AddressHate Research Scholar at NYU’s Center for the Study of Antisemitism | Lead, Decoding Antisemitism — PI, Decoding Hate | Research Advisor, AddressHate | Editor-in-Chief, Digital Hate Review

Special thanks to Emily Carmeli, President of AddressHate, for the original idea behind this case study.

Executive Summary

Between July 8 and 14, 2026, Braden Peters (“Clavicular,” b. 2005), one of the most prominent figures of the “looksmaxxing” subculture, visited Israel — streaming from Tel Aviv, visiting the Western Wall, and appearing alongside an IDF soldier who became the second target of the backlash. Six months earlier, Peters had made international headlines as part of the Miami nightclub group singing along to Ye’s “Heil Hitler.” His visit divided pro-Israel and Jewish public discourse between a pro-engagement camp — which hoped his enormous reach could carry positive impressions of Israel into a Gen-Z male milieu no conventional actor reaches — and critics who held that engagement conferred legitimacy at unacceptable cost. Yet much of the argument on both sides rested on assumptions about how Peters’ audience and surrounding online milieu would receive the visit. To our knowledge, no empirical analysis of the publicly observable reception attached to Peters’ own account existed. This study addresses that gap by examining the public reply sections attached to Peters’ own X account.

Research question: How did Clavicular’s own online community publicly interpret and respond to his visit to Israel? This is not a study of whether Peters himself changed; it is a study of how his community reacted when one of its central figures publicly affiliated with Israel.

We — Decoding Hate, the research hub at NYU's Center for the Study of Antisemitism, in collaboration with the think tank AddressHate — collected all replies recoverable from six X threads posted during the visit (1,215 unique authors) and annotated them as 1,379 analytical units — text, images, and memes, with longer replies split into separately coded sub-units — using the project’s peer-reviewed Decoding Antisemitism Lexicon (cited in full under Research design). Every borderline case was resolved through documented, corpus-specific rules.

The findings reveal a clear overall pattern despite substantial variation across the six threads. More than two units in five (42.6 %) carry antisemitic content — and the engagement distribution was starker still: antisemitic replies received more than nine out of every ten likes recorded on coded replies (see Overall statistics). Antisemitism peaked at 59.6 % in the "STATUSMAXING" thread (X-4) and 58.2 % in the arrival thread (X-2).

And the antisemitic share does not exhaust the hostility: a further 5.9 % of units attack Peters or the trip with no group-based content at all — insults, hypocrisy charges, plain rejection — rising to 17 % once non-antisemitic servility and other group-based hostility are counted. These categories are mutually exclusive by construction and together account for roughly six in ten units (Table 1). What varies is not whether the visible reply environment rejected the visit but which vocabulary the rejection employed. Replies showing any genuine rethinking of Israel — surprise, curiosity, a revised view — are rare; where users defended anyone, they overwhelmingly defended Peters, not Israel, Israelis, or Jews. Within hours, the tone was set by the milieu’s two most famous voices. Their replies — declaring the trip career-ending, cursing Israel’s existence, and recasting Peters as a manipulated asset — drew approximately 54,000 likes combined and open the Introduction.

Four findings structure the analysis, and one conceptual claim organizes them: affiliation with Jews or Israel can itself become the trigger for the antisemitic reclassification of the affiliating non-Jew.

• First, the dominant hostile storyline is not abstract “Jewish control” rhetoric but its personal application: Peters recoded as bought, captured, and honey-trapped, and finally declared Jewish himself (”Stfu jew,” “Kikevicular,” memes editing a kippah onto his head) — antisemitism deployed as a disciplinary tool against one of the community’s own, what this study terms in-group discipline.

• Second, the same step-by-step pattern — disappointment, then “he’s being controlled,” then “the Jews control him,” then “he is one of them” — was documented in the reactions to Jordan Peterson’s Netanyahu interview (Becker 2024) and reappears here in a younger, more visual influencer milieu; the study calls it the dethroning arc.

• Third, part of the outrage responds to events that never occurred: a fabricated photo of Peters with Minister Itamar Ben Gvir — watermarked by a self-labeled parody account — circulated as real and anchored a demonstrable cluster of hostile replies. Part of the factual ground these replies reacted to was synthetic.

• Fourth, the antisemitism and the misogyny are not separate problems but one fused system: in the July 13 thread, 31 of 47 misogyny-coded replies are simultaneously antisemitic, and their primary target is the soldier — the only participant in the entire episode to face institutional sanction.

The answer to the research question, within the visible reply sections on Peters’ own account (one component of the visit’s reception, though the most public one), is stark: not new thinking about Israel, but conspiracy narratives, public punishment for the affiliation itself — what this study terms affiliation punishment — and, at the edges, open neo-Nazi imagery. The pattern is consistent with ideological self-stabilization performed in public: potentially disconfirming evidence was not ignored but repeatedly reinterpreted, visibly, through established narratives of capture, betrayal, and conspiracy.

Table of Contents

• Introduction

• Dataset

• Research design

• Empirical findings

  • Overall statistics

  • The registers in detail

• Discussion

• Why this matters

• Conclusion

• References

Introduction

The visit quickly became a national and international controversy. Israeli media reported Peters’ walkout from a Channel 13 interview when asked to apologize for the “Heil Hitler” episode (Jerusalem Post), the disciplinary action taken against the soldier who had appeared with him (i24NEWS, Ynet), and sharp disagreement among pro-Israel voices over whether engagement represented strategic outreach or irresponsible legitimization (JTA / Jerusalem Post); international coverage framed the episode variously as influencer diplomacy, reputational laundering, and a failed propaganda exercise, and by week’s end even a United States congresswoman had been asked to comment (Jerusalem Post). Within the debate, three positions crystallized.

Strategic engagement: Peters’ reach — approximately 2.2 billion third-party clip views in March–April 2026, per a marketing-affiliated and unaudited figure — extends into a manosphere-adjacent milieu whose encounters with Israel are otherwise mediated through conflict imagery, conspiracy narratives, and adversarial influencer culture, and whose discourse, as this project’s research documents, recurrently combines antisemitism, misogyny, racism, and homophobia with distrust of democratic values and a tendency toward authoritarian worldviews; as one Jerusalem Post opinion piece put it, “the alternative isn’t a better messenger. For most of this audience, it’s no messenger at all” (Gerstley, Jerusalem Post).

The legitimization critique: hosting Peters normalized a figure embedded in antisemitic and misogynist environments — a culture, in another Jerusalem Post op-ed’s phrase, that “mistakes influence for integrity” (Abramowitz-Silverman, Jerusalem Post).

The attention-arbitrage reading: the visit was neither diplomacy nor redemption but monetized controversy — Peters said openly that he came because Israel was trending and no major influencer had streamed from it. (The debate had a further, sharply gendered dimension, taken up below in the misogyny analysis, where it belongs.)

Across that debate, participants assumed something about how Peters’ own audience would respond. Supporters assumed exposure might soften attitudes — that nightlife, food, and friendships from Tel Aviv could reach a milieu whose image of Israel is otherwise entirely secondhand. Critics assumed the visit would normalize extremism or provoke further antisemitism. These claims pointed in opposite directions, and all of them were speculative: no study examined how Peters’ community actually received the visit. This paper is that study. Its question: how did Clavicular’s own online community publicly interpret and respond to his visit to Israel? Did users express curiosity or revise prior assumptions? Did they respond to ordinary scenes from Tel Aviv as potentially disconfirming experiences? Or did established narratives absorb the visit instead — recasting Peters as bought, controlled, sexually manipulated, or secretly a Jew?

On July 10, 2026, Peters announced his first Tel Aviv livestream: “Live on Kick! With special guest…” The reply section answered before the stream began. Dan Bilzerian, one of the manosphere’s most-followed figures, wrote: “Good way to destroy your reputation” — at approximately 26,000 likes the single most-engaged reply in the corpus. The comment contains no antisemitic keyword, no slur, no conspiracy claim; in the context of the thread, it presents contact with Israel as reputationally disqualifying. Sneako — Peters’ erstwhile associate from the Miami “Heil Hitler” episode — posted “Fuck Israel free Palestine” into the same thread (≈21,000 likes). Three days later, under Peters’ photo beside an Israeli soldier, Sneako supplied the conspiratorial completion: “You need to know that you were her assignment” (≈7,000 likes). Between these three utterances — rejection, delegitimization, capture — runs the arc this study will document across more than a thousand replies. The argument, stated at the outset: in this corpus, one major function of antisemitic repertoires — alongside, not in place of, the direct hostility toward Jews the corpus contains in abundance — is the disciplining of a perceived in-group member’s affiliation with Jews and Israel: antisemitic repertoires operating as a social technology for regulating the community’s own boundaries.

A word on scope. Reply data cannot measure silent viewers, private attitude change, or the full reception of the livestream; and three terms this study keeps distinct are worth fixing here: Peters’ audience includes millions who never comment; his milieu is the broader subcultural formation he belongs to; the reply environment is the observable discourse attached to his posts. This study analyzes the reply environment — exhaustively, within the replies recoverable at the time of collection — and infers from it only what a reply environment can show: how the visit was publicly interpreted, rewarded, rejected, and reframed in the space attached to Peters' own account.

Because several of this study’s terms are related but not interchangeable, their hierarchy is stated here once. Affiliation punishment is the overarching social process: the public sanctioning of an individual for establishing, or appearing to establish, social proximity to a stigmatized group. The process is general; this study analyzes its antisemitic form — affiliation with Jews or Israel, hereafter simply affiliation. In-group discipline is that process’s social function — what the punishment accomplishes for the community that performs it. The dethroning arc is the recurring sequence through which the process unfolds in this and comparable corpora. Reclassification is the mechanism of identity reassignment operating within that sequence, and judaisation is its strict antisemitic form — the rewriting of the non-Jewish target as Jewish. When later sections use one of these terms, they refer to that level of the hierarchy and no other.

Dataset

The corpus consists of all replies recoverable from six trigger posts on Peters’ X account (@Clavicular0) at the time of collection (July 15, 2026): replies from 1,215 unique authors, posted July 9–15. The collected material was annotated exhaustively. The six threads, chronologically: a July 9 post from the hours before the trip became the dominant topic of his reply sections (138 units), the July 10 arrival announcement (359 units), Peters’ apolitical defense of later that day (358 units), the July 11 “STATUSMAXING” post (141 units), the July 13 critique of Israeli media coverage (121 units), and the July 13 post beside the IDF soldier (262 units): 1,379 annotated units, of which 16 % carry attached media, inspected wherever recoverable. The corpus flow, stated once: all recoverable replies from the six threads were collected on July 15 and segmented into 1,379 adjudicated analytical units — longer replies containing separable quoted material were divided into separately coded sub-units — which form the quantitative results reported in the tables. Units are individual X replies, with attached media coded as part of the same unit under the same annotation protocol, quoted material coded against the replying author’s stance, and author handles logged as a separate signal layer rather than as unit content; one unit counts once toward prevalence regardless of how many codes it carries. Media items were inspected wherever recoverable; units whose media could not be retrieved are excluded from visual analysis and marked as such. All comments are quoted verbatim; author handles of private individuals are not republished in public outputs, and the soldier is referenced only through the already-public record. A legal note: Peters is the defendant in a pending civil lawsuit alleging sexual assault of a minor, which his counsel denies; no criminal charge exists, escalated versions of these allegations circulate in the reaction discourse, and this study treats all such circulating versions as rumor artifacts — objects of analysis, never statements of fact.

A note on platform selection. X hosted the most directly attributable and methodologically recoverable public response to Peters’ own trip-related posts. The trip’s video content ran on Kick, but livestream chat is ephemeral — it scrolls in real time, carries neither stable threading nor per-message engagement metrics, and no complete and verifiable archive was available to the research team. Peters’ Instagram presence contributed only a handful of tangentially related posts during the trip window, generating no comparable reply discourse; and YouTube carried the episode mainly through third-party clip channels, whose comment sections react to a re-uploader’s framing rather than to the influencer’s own speech acts (the project’s YouTube-based studies — the Antisemitism in Real Time series, the Kirk analysis, Catchy Hatred — cover that reception mode; this study deliberately extends the research line to X). The choice of X is therefore methodological, not residual: Peters’ account hosted the trip’s discursive events themselves — the announcements, the apolitical defense, the soldier photo — with public reply threads attached directly to his posts, stable URLs and timestamps enabling temporal analysis, and public like counts enabling the engagement-weighted measures reported below. X provided the only directly attached, stable, and retrospectively collectable reply corpus identified for this study.

Research design

Annotation was conducted by the Decoding Hate research team following the linguistic and multimodal close-reading protocol of the AiRT series, guided by the DA Lexicon (Becker et al. 2024). In contrast to previous Decoding Antisemitism and Decoding Hate studies, which typically annotate a defined sample of each comment section (on the order of 100–200 units per thread), this study annotates the entirety of each collected reply section — every reply in each of the six threads, exhaustively rather than by sampling — which removes selection effects within threads and permits genuine thread-level comparison. Annotation proceeds at the unit level with three layers per unit: track, i.e. the hate ideology addressed (antisemitism; misogyny; other group-based hostility; supportive/counter-speech), concept(s), i.e. the stereotypes and other imagery drawn from the antisemitic or other repertoire, and register (explicit / coded / co-textual, with a separate visual register for media-borne concepts). Two corpus-specific coding decisions require statement. First, bare servility allegations — “sellout,” “how much do they pay you” — without ethnoreligious or conspiratorial markers are tracked as their own register (the individual-scale analogue of the subordination frame in Becker 2024 and 2026b) and are not counted as antisemitism; they upgrade only where the mechanism recruits Jewish-money, goy-servitude, covert-control, or soul-devil concepts. Second, Israel-directed slogans, toponym erasure, and legal-institutional atrocity framings are coded in this corpus as delegitimization of the Jewish state: addressed to an influencer’s lifestyle stream rather than to news coverage of a war, such utterances can be read, in this interactional context, as performing pariah-marking rather than policy advocacy. In this interactional setting the delegitimization family is a constitutive part of the corpus’s antisemitism, not an adjacent category, and it is counted as such; because the boundary is drawn differently from AiRT Part IV (§ Forms of Rejection), where the same forms appear in news-commentary contexts, the decision is documented here explicitly. A further category, adjudicated for this corpus: units performing rejection over the visit — unfollowing, “stay there,” career-over declarations — are coded as affirmation of (Israel-related) antisemitism only where the context independently supplies the Israel-as-pariah norm these units enact — the unit replies to or endorses an antisemitic proposition, carries a group marker or recognized formula, or its author or attached media establishes the referent in the local exchange; generic rejection that cannot be contextually resolved remains affiliation punishment outside the antisemitism count. For Table 1’s summary rows, the definitions are: rejection without antisemitism covers units coded as attacking Peters or rejecting the trip — insult, mockery, hypocrisy charges, dismissal, staged-stream claims, political critique — including those carrying servility or other group-based hostility, but no antisemitic or misogynistic content and neither supportive nor counter-speech coding; its narrower core, units with no group-based coding of any kind, comprises 81 units (5.9 %). Both rejection rows count only units affirmatively tracked as hostile or critical during annotation: replies whose hostility was incidental to an otherwise untracked unit — passing mockery, hypocrisy asides, generic insult without a coded stance — remain outside the count, so the rejection figures, and the cumulative hostile-in-any-register row built on them, are conservative floors rather than exhaustive tallies of negative sentiment. Hostile in any register is the sum of the antisemitic and rejection rows, which are exclusive by construction. Supportive covers units endorsing Peters — his take, his conduct, or the visit; units welcoming the visit or speaking positively of Israel itself form the Israel-positive subset discussed under Overall statistics (the apolitical-defense thread’s large supportive bloc endorses Peters’ right to silence, not Israel). Counter-speech covers units contesting the hostile discourse itself — naming or pushing back on the antisemitism — including critical Jewish and Israeli voices rebuking Peters’ prior conduct; pro-visit endorsements are not counter-speech and are counted as supportive. Affirmation of an antisemitic norm is part of the phenomenon, and the corpus counts it accordingly. Attitude revision was coded only where a unit met at least one defined indicator: explicit surprise, explicit revision of a stated prior view, acknowledgment of a previous misconception, a humanizing observation about Israelis or Jews as people, or admiration replacing prior hostility; absence of these indicators, not impressionistic reading, grounds the finding that such units are rare. We adjudicated every boundary class against a documented, corpus-specific rulebook to ensure internal consistency. A language model (Claude Fable 5, Anthropic) was used to structure and organize the discourse material and the research team’s working notes; it did not determine inclusion or make final coding decisions — all analytical categories, boundary rules, and adjudications follow the project’s expert annotation framework and were reviewed by the research team.

Empirical findings

Overall statistics

Across the 1,379 units, 42.6 % carry antisemitic content, with sharp variation by trigger: 58.2 % in the arrival thread (X-2), 59.6 % in the "STATUSMAXING" thread (X-4), 37.8 % in the thread beside the soldier (X-6), 30.4 % in the pre-trip July 9 thread (X-1), 33.8 % in the apolitical-defense thread (X-3), and 27.3 % in the media-critique thread (X-5). The last two carry the corpus’s two sizeable supportive blocs — 24 % and 22 % — supporting, in each case, Peters’ stance rather than Israel. A further tenth of the corpus carries the servility register (including units co-tagged with antisemitism); homophobic, transphobic, ableist, racist, and anti-Muslim content without a Jewish/Israel referent runs throughout; counter-speech totals about 1.5 %. The range extends to open neo-Nazi iconography and far-right meme culture in a mainstream reply section: the arrival thread carries a photograph of what appears to be George Lincoln Rockwell, founder of the American Nazi Party, captioned “No one cares about your jew bullshit” — and around it circulate Pepe the Frog variants, the “happy merchant” caricature, Hitler-valorizing macros, and Nazi meme templates, the standard visual repertoire of far-right online spaces surfacing in the reply section of a lifestyle influencer (the visual layer is analyzed in §10). Two prevalence figures must be read together, because they measure different things. Counted by unit, with every analytical unit weighted equally, 42.6 % carry antisemitic content. Counted by attention, the picture inverts: of the roughly 113,700 likes recorded on coded replies across the six threads, 94 % sit on replies containing antisemitism-coded content; under the corpus-specific rules documented above — including contextual affirmation of an antisemitic pariah norm — all ten of the corpus’s most-liked replies are antisemitism-coded, and the top fifty carry over 90 % of their combined like-mass on antisemitic units. Antisemitism is not the majority of voices in these threads, but it is the overwhelming majority of attention: the 42.6 % measures prevalence across units; the 94 % measures its share of engagement recorded on coded replies. Although like counts do not reproduce X’s ranking algorithm, their concentration shows that antisemitic replies occupied the most publicly rewarded layer of the discussion — and the concentration holds at every level: nearly every reply above 400 likes across the hostile threads is antisemitism-coded, affirmation-coded, or servility-coded. (Like counts as recorded at collection.) This visibility pattern is partly a function of who posted — several of the highest counts belong to major accounts whose reach precedes any single reply — so we treat it as evidence about which replies were most publicly rewarded and therefore likely to be especially visible, not as a measure of ideological distribution among ordinary users.

Figure 1. The visit as a sequence of triggers: antisemitic share of replies by thread, in posting order (bars show share of 100 %).

The rate varies sharply with the trigger — the threads also differ in timing, participants, and reach, so the variation is descriptive rather than causal: it more than doubles between Peters' apolitical defense and his status performance one day later, and even the pre-trip thread — whose trigger has nothing to do with Israel — reaches 30.4 % once the Israel frame saturates his reply sections. The corpus's lowest rate, 27.3 %, appears in the media-critique thread, which also contains one of the study's two sizeable supportive blocs.

Figure 2. What was said vs. what was rewarded (six threads; bars show share of 100 %).

A further register must be read alongside these: 81 units (5.9 %) attack Peters or reject the trip without any group-based content at all — hypocrisy charges, insults, staged-stream claims, plain political critique — rising to 234 units (17.0 %) when units carrying servility or non-antisemitic group hostility — but no antisemitic or misogynistic content — are included (Table 1). This register matters twice over. First, it shows that hostile criticism of Peters without antisemitism was a fully available, well-populated option in this discourse — the 42.6 % who reached for antisemitic material were making a marked choice, not following the only script on offer. Second, it identifies the reservoir: the non-antisemitic rejection register shares its grievance structure with the antisemitic one — sellout, hypocrite, fallen idol — and that structure can at any point be filled with antisemitic concepts, as the pre-trip thread's intra-day escalation, documented below, demonstrates. The plain attacks are not the opposite of the antisemitic ones; they are their recruitment surface.

The other side of the ledger is small but must be reported with the same care. The adjudicated corpus contains 136 supportive units (9.9 %) and 21 units of explicit counter-speech (1.5 %) — users directly contesting antisemitic replies. Supportive units concentrate principally in the apolitical-defense thread (85 of 136), with a second bloc in the media-critique thread (26), and its content is revealing: it consists overwhelmingly of in-group loyalty to Peters — defense of him against his critics — rather than endorsement of the visit, positive statements about Israel, or any revision of prior views. The bloc’s top unit is exemplary: “This is genuinely the best take I have seen from a 20 year old” (≈574 likes) endorses his right to silence, not the country he was standing in. Only units that speak positively of the visit or of Israel itself — “Have a great time! Enjoy Israel just like you would enjoy any other awesome vacation destination” (≈436 likes) — actually push against the pariah norm the antisemitic registers enforce, and these are a small minority within this already small bloc. Counter-speech, at roughly 1.5 % of the corpus, was outnumbered by antisemitic units by roughly 28 to one.

The July 9 thread sharpens the picture from its edge. Its trigger concerns alcohol advice and a statement about a Miami club — not the trip — and its replies from the morning of July 9 run 15 % antisemitic, in the thread’s native register of drug-hypocrisy mockery. From that afternoon onward, as news of the Israel trip saturated Peters’ environment, the same thread’s rate climbs to 37 %: reclassification, collective-marker formulas, slogans, and an Epstein–Mossad blackmail attribution flood a conversation about nightclub liability. The Israel frame did more than dominate the Israel threads: it became salient even within a thread whose original trigger concerned something else entirely.

Table 1. Corpus overview and headline rates. Each cell shows n (within-thread %); corpus column shows n · share of 1,379. Rejection, core = hostile/critical units with no group-based coding of any kind; Rejection, broad adds units carrying only servility and/or other-group hostility tags; both are exclusive of the Antisemitic row, and the Hostile row sums the Antisemitic and broad-rejection rows. Threads chronological: X-1 pre-trip post (Jul 9) · X-2 arrival announcement (Jul 10) · X-3 apolitical defense (Jul 10) · X-4 status performance — trigger word “STATUSMAXING” (Jul 11) · X-5 media critique (Jul 13) · X-6 photo with the soldier (Jul 13).

Most-engaged coded reply per thread: X-1 “Jew cock sucker” ≈ 12 · X-2 “Good way to destroy your reputation” ≈ 26K · X-3 child-torture attribution ≈ 4.2K · X-4 “JEWMAXXING” ≈ 7.9K (8:1 over the 1K trigger) · X-5 welcome-to-the-club reclassification ≈ 98 · X-6 “you were her assignment” ≈ 7K.

Multi-track units are counted in every applicable register row; percentages are of thread n. Row definitions: see Research design.

Table 2. Antisemitic trope families by thread — what dominated where. AS-coded units; multi-label counts; thread columns chronological (thread key as in Table 1). ◀ = the family’s highest-frequency thread. Families aggregate DA taxonomy concepts — power/capture: power, conspiracy theories (AS-C2-2, AS-C2-3); demonization: evil & moral corruption, religious antisemitism (AS-C1-2, AS-C1-5); atrocity: genocide, terrorist state, Nazi analogy, blood libel & child murder (AS-C5-5, AS-C5-4, AS-C5-1, AS-C1-3); hostility/violence: insults, threats, death wishes, deportation/exclusion commands (AS-C6-1, AS-C6-2, AS-C6-4, AS-C6-8); affirmation of antisemitism (AS-C4-3); reclassification/foreignness (AS-C1-1); delegitimization: denial of Israel’s right to exist (AS-C5-7); physiognomy/dehumanization (AS-C1-4). A unit counts once per family regardless of how many of its codes it carries.

In one sentence: the arrival announcement concentrated nearly every register, the self-defense drew the atrocity charges, and the woman drew the misogyny-fused conspiracy.

The registers in detail

The subchapters below present the qualitative material in two blocks. Sections 1–3 document the disciplinary registers aimed at Peters himself — capture, servility, reclassification and expulsion — in the order of the dethroning arc they compose. Sections 4–6 document the image of Israel that licenses that discipline — atrocity, demonization, delegitimization — escalating from alleged acts to attributed essence to denied existence. Sections 7–10 then treat the fusion registers (misogyny, homophobia) and the carrier layers (lexicon, visuals) that run through both blocks. Within this structure the ordering broadly tracks corpus frequency (Table 2): the power register is the largest family, and the servility register is reported directly after the power register it scaffolds.

1. Power and conspiracy: the capture of a gentile

The dominant register personalizes Jewish-power conspiracy as the capture of the affiliating gentile. In the arrival thread the mechanism is economic — purchase, shekels, orchestration:

“Pick a Goy and have him push the body dysmorphia agenda. But we’ll call it looksmaxxing.” (≈2,800 likes — a protocols-format fabricated quote attached to a photo of a Haredi man with a walkie-talkie)

— flanked by an Epstein-blackmail warning and, at the fringe, historical-conspiracy stacks (USS Liberty, JFK). The repertoire’s historical depth surfaces in a single aside, a medieval canard revived intact in a Gen-Z reply section:

“Careful with your coins, they might clip them.”

The capture claim also takes its most bodily form — the influencer as a literally operated object:

“A sea of kikes and you’re some billionaire jew puppet, your fame is fading buddy. More like puppet maxxing, get the hand out of your ass, loser.”

The puppet figure renders the control thesis physical: not influenced but operated, a body with a hand inside it, in an environment described through the slur as an immersive mass (”a sea of kikes”). The unit stacks four registers in two sentences — slur, puppet-servitude, the “-maxxing” morphology, and the fame-loss verdict — a compact demonstration of how the capture, carrier, and total-loss layers travel together.

Three days later, once a woman entered the frame, the mechanism turned sexual-conspiratorial: the honeypot/handler/”assignment” complex dominates the July 13 thread (31 units), from full articulation down to bare two-word formulas — a frame lexicalizing in real time:

“You need to know that you were her assignment.” (Sneako, ≈7,000 likes)

“she’s telling someone else everything you say to make sure you’re not a spy”

“Honeypot maxing”

The handler figure is not left abstract: the corpus names one. The rabbi publicly reported as the visit’s facilitator is recast, in the reply sections, as an intelligence officer running an asset:

“call @FarroYossi for permission. You’re handlers”

Around this core, the register runs the full scale of explicitness — from two-word capture celebration to fully engineered-world paranoia:

“They got em boyz”

“he is being programmed by the folks around him”

“a hollow facade engineered to manipulate children…” (≈46 likes — the industry itself as an engineered deception, agent unspecified at saturation)

And the register has its emblem. The arrival thread carries the classic control-iconography of the tradition, unmodified: a serpent marked with the Star of David strangling the American eagle.

Jewish power figured as a snake choking the United States — the visual grammar of a century of antisemitic caricature, reposted into a Gen-Z reply section.

The narrative also has a long-game variant, the fully articulated infiltration timeline: Peters as a plant rather than a purchase, his role to defuse the awakening at its critical moment:

“The s guys a jew asset. He will trickle bullshit out there trying to slowly manipulate until right when the goyim are coming together to fight against tiny hats then say some shit like jews are our friends”

And the capture narrative has a terminus. In its completed form, the asset, once used, is liquidated — the corpus repeatedly forecasts Peters’ death at the hands of the power that allegedly bought him, and it has already verbified the scenario through the killing of Charlie Kirk, whose digital afterlife this project analyzed in an earlier study:

“Get ready to be Charlie Kirked.”

“After your botched nose job, if you suddenly die we genuinely won’t know if it was Mossad or if you actually killed yourself.” (≈18 likes)

“She’ll be running your podcast within 2 yrs while we argue over how much you loved Israel before your exploding face puff powder death”

The verbified Kirk conspiracy is not this corpus’s invention but the newest entry in a long lineage of assassination attributions: the JFK killing laid at Israel’s door (present here in its own right — “JFK died for this”), the Epstein “didn’t kill himself” formula, whose logic the Mossad-or-suicide unit above reproduces exactly, and, most recently, the preemptive “Remember, Dan is not suicidal” declarations that circulated around Dan Bilzerian, analyzed in an earlier study from this project. The corpus even closes that loop: Bilzerian himself — the figure around whom the not-suicidal formula was coined — appears in these reply sections as the single most-liked voice (§3).

Taken together, these formulas construct a sequence running from purchase through capture and use to liquidation — the shape of the corpus’s largest trope family (250 units, Table 2). The conspiracy grants its target no exit, a point the expulsion analysis below (§3) returns to.

One temporal detail deserves emphasis. The photo with the soldier — the ostensible “evidence” for the honeypot claim — was posted July 13. But the corpus contains the same frame four days earlier: a July 9 reply already casts Peters as controlled by a Jewish girlfriend (the unit is presented in §7), at a point when no woman had yet appeared in his trip coverage at all. The honeypot narrative was not inferred from the photo; it existed before its evidence did, and the photo was slotted into it. A possible second early instance within the July 13 thread itself is under timestamp verification.

2. The servility register — the antechamber

Directly adjacent to the conspiracy register, and deliberately reported beside it, sits the bounded servility discourse:

“Sellout.” (≈105 likes)

“What a money whore”

“You really selling yourself HARD, dont ya”

“Dumbest decision you’ve made so far.”

These units name no Jewish mechanism and are not coded as antisemitism — but they supply the narrative scaffolding on which the coded units build, as in the Peterson and Trump-era corpora (Becker 2024, Group B1; 2026b). The correlation is the point: in both threads, the servility register and the capture register tell the same story at two levels of explicitness, and individual commenters move between them freely. Where the servility register names its mechanism — shekels, “good little goy,” a soul sold (the soul-transaction necessarily supplying a devil as counterparty) — it crosses into the coded corpus. The crossing point has its own codeword:

“The final boss of humiliation rituals. Fucking pathetic” (≈81 likes — “humiliation ritual,” the far-right formula for degrading acts demanded of public figures by a hidden power: the sellout charge upgraded into orchestration)

3. Reclassification: the gentile expelled

A boundary note before the material: judaisation — in the strict sense defined in the Introduction — operates here through slurs, names, attributed religious markers, physiognomic stereotypes, or manipulated imagery, a pattern first documented at scale in the reactions to Jordan Peterson’s Netanyahu interview (Becker 2024). It does not cover every capture claim or purchase allegation: capture denies the target’s autonomy; judaisation rewrites his attributed identity; expulsion removes him from the in-group. The three are stages of one broader process — affiliation punishment, as defined in the Introduction — whose social function this study calls in-group discipline.

The corpus documents a recurrent, multi-stage apparatus for expelling the affiliating gentile from gentile status. Onomastics — the same suffix-Jewification pattern as “Juden Peterstein” in the Peterson corpus and “Trumpstein” in the Trump corpora:

“Kikevicular” (≈490 likes)

“Claviculberg”

Direct address:

“Stfu jew” (≈730 likes)

“we know you are a kike, stop capping”

“Aren’t you Jewish you psyop” (posted July 9 at 07:26, hours before the arrival announcement — the crypto-Jew question live on day one, and already fused with the asset frame)

“Sell out. You used to call out the shabbos goyim. Now you are one.” (the canonical servitude concept — the shabbos goy, a gentile in Jewish service — deployed as a before/after verdict: the entire dethroning in eleven words)

The apparatus then turns visual, the reclassification performed on Peters’ own appearance:

“now we know the reason he got a nose job, it was so he could get closer to kiss the wall” (≈67 likes)

“Did you slam yer nose into the wall before you kissed it?” (≈260 likes — posted from a handle carrying the “110” expulsion codeword)

The subculture that built its identity on measuring faces here converts its own core practice into the oldest physiognomic stereotype in the antisemitic repertoire. The logic completes itself in a single image: a meme of a celebrating woman with the US flag captioned “110” — the far-right expulsion-count codeword (Jews expelled from “109 countries,” America the 110th) — aimed at the reclassified Peters with the caption “Stay there”: first recoded as a Jew, then included in the imagined expulsion. The same codeword operates one layer deeper in the platform’s architecture: the author of one of the corpus’s most-liked physiognomic taunts posts from a handle carrying “110” in its name — the expulsion fantasy worn as identity.

Rejection as affirmation — the total-loss family. The expulsion has an enforcement arm, and it contains no antisemitic vocabulary at all:

“Unfollowed.”

“Stay there.”

“It’s shalomver” (a nonce blend of shalom and the doomer formula “it’s over”)

“Good way to destroy your reputation” (Bilzerian, ≈26,000 likes — the corpus’s most-engaged utterance)

“Have you ever seen mafia movies? Resignation from the mob is usually gruesome.” (membership as mob, exit as death — the discourse articulating its own enforcement logic)

Each enacts the same proposition: contact with Israel means losing everything, because Israel is a pariah whose touch contaminates — and, per the mafia unit, that the alleged membership can be neither kept nor safely resigned. One unit narrates the loss as a completed biography, with both endpoints named:

“Went from funny nick fuentes streams to a goy jew lover kissing the wall. Insane fall off” (≈41 likes)

Where most total-loss declarations assert ruin without specifying what was lost, this one states the community’s value scale outright: the prelapsarian state is the Fuentes-sphere affiliation (”funny nick fuentes streams”), and the fall is the affiliation with Jews — “goy jew lover kissing the wall” stacking the servitude label, the reclassifying address, and the submission ritual into a single fallen identity. The unit is the dethroning arc told as before-and-after in one breath, by the community, about itself.

The expulsion even has a stated readmission price. Where the servility register taunts Peters for kissing the Wall, the rejection register names the ritual that would undo it — desecration:

“The first mfer to stream going to the wall and spitting or pissing on it, is gonna ascend to mythical levels of legendary… TRUE ASCENSION SON” (an incitement echoed in the arrival thread: “pull a KING move and piss on that wall”)

The site of his submission is designated as the site of his possible redemption; the community that expelled him specifies the profanation through which he could return. Coded as affirmations of Israel-related antisemitism, these units ratify the shun-norm for every onlooker — which is how a norm becomes ambient without ever being argued — and their survival across surfaces (a slang token, a Hebrew-marked coinage, a marketing observation) shows the frame circulating beneath the threshold of every keyword-based detection system.

4. The atrocity register: genocide, child-killing, and Nazi analogy

The corpus’s third-largest register (108 units) attributes atrocity: genocide, child-killing, and Nazi-analogical framings directed at Israel and, through his presence there, at Peters himself:

“Fuck each and every zionst. Child killing freaks” (≈8,800 likes — the arrival thread’s second-most-engaged reply)

“You hang out with people responsible for genocide”

“Did they let you shoot a kid or rape a prisoner, or is that only for jews?” (Jake Shields, ≈4,800 likes)

The register also arms itself, converting the attribution into a dare:

“More like GENOCIDEMAXXING. Why don’t you walk up to an IDF soldier and ask him to finish your botched nose job with his rifle butt?” (≈260 likes)

The taunt presupposes that Peters’ actual experience — present, safe, filmed — is false, and that contact with the “real” Israel would maim him: the disconfirming evidence is not disputed but converted into a challenge whose terms already assume the atrocity frame. It is the self-stabilization mechanism of the Discussion performed in a single reply, with the violence wish routed through the genocide attribution and the whole packaged in the subculture’s own morphology.

The complicity transfer is the register’s signature move — and the Shields unit shows the register’s essentializing core: atrocity figured not as a state’s crime but as a Jewish privilege, the same slide from individual accusation to collective essence that this project documented in an earlier analysis of abuse allegations recast as “organized state policy”. The family is internally heterogeneous, and the coding preserves the differences: Israel-directed atrocity accusation, Nazi analogy and inversion, Holocaust denial, Holocaust derision, and collective transfer to Jews or to the influencer are distinct forms, antisemitic on different grounds, and are reported here as one frequency family only for the corpus overview. The derision form has its own coinage in this corpus:

“From clubacaust to Tel Aviv within 6 months” (23 likes)

“Clubacaust” — the Miami “Heil Hitler” club episode blended with the Holocaust — compresses the entire affiliation arc into six words while trivializing the Shoah in a single morpheme: the community narrates Peters’ trajectory from Nazi-adjacent transgression to Israel affiliation in Holocaust-derisive vocabulary, treating both endpoints as content for the same joke. The soldier-photo thread escalates the coinage from derision to demand — a call for “clubocaust 2” — completing the move from mocking the episode to requesting its repetition. The register peaks in the thread where Peters declared himself apolitical: 42 of that thread’s antisemitic units carry it, led by a second Jake Shields reply at ≈4,200 likes — “You are helping a country that tortures and kills children. Every International body says Israel is committing war crimes” — which places two Shields units at a combined ≈9,000 likes atop the register, elite amplification in the most literal sense — the claim to be outside politics was answered with the heaviest accusatory material in the corpus, because within this discourse there is no apolitical presence in Israel: location is complicity. At the register’s edge, atrocity attribution shades into Shoah derision and inversion — Holocaust-denial and oven imagery documented in the visual layer below — completing a spectrum that runs from pseudo-juridical accusation to open celebration of the crime it invokes.

5. Demonization: Israel as evil

Distinct from political rejection, a family of 112 units attributes evil as essence — the oldest theological layer of the antisemitic repertoire, in Gen-Z orthography:

“Israhell” (recurring across threads as a fixed spelling)

Israel as “demon country”

The theology is most fully articulated in the corpus’s soul-sale units, where affiliation is figured as a Faustian bargain and Israel as the devil buying:

“Not worth your soul, buddy.” (≈216 likes — one of the register’s most-engaged units)

“They’ll entice you… All for your soul. It’s the devil’s game.”

This is where demonization meets the purchase register (§1): the soul-sale formula requires a devil as counterparty, so every “sold your soul for paper” unit presupposes that the buyer is evil in its essence. The frame also attaches to the people, not only the transaction — Israelis figured as a “murderous cult,” Israel as a “demonic psychopath” — and, at its sharpest, folds back onto Peters himself, the evil-attribution completing the reclassification of §3:

“demonic Jew demon” (Peters addressed directly — demonization and reclassification collapsed into one phrase)

The essence-attribution also arrives in the subculture’s own morphology:

“Lol your statusmaxing in the most hated country in the world why don’t you go rapemaxxing now that your there” (≈31 likes)

Two moves happen in one sentence. “The most hated country in the world” states the pariah norm as settled global consensus — the universal-hatred formula. And “go rapemaxxing now that your there” treats mass rape not as an allegation about specific acts but as a standing national activity, available to any visitor the way nightlife is: atrocity attributed as essence, rendered in the “-maxxing” template that folds it into the milieu’s everyday idiom. The unit is the corpus’s terminus of a trajectory this project analyzed in the mainstream press cycle it feeds on — the recasting of individual abuse allegations as “organized state policy” (Becker 2026c): once atrocity has been promoted from act to essence, it can be offered, casually and in a lifestyle vocabulary, as something one does in Israel.

In plain terms: where delegitimization says Israel should not exist, demonization says Israel is evil in its being — and it is the second claim that licenses everything the atrocity register (§4) asserts.

6. Delegitimization

Israel-directed slogans and erasures are coded in this corpus as delegitimization of the Jewish state (see research design):

“Fuck Israel free Palestine” (Sneako, ≈21,000 likes)

“This is literally Palestine, dude” (≈1,700 likes)

“It’s called PALESTINE!”

— alongside stolen-land framings and “from the river to the sea.” Addressed to a lifestyle stream, they function here as pariah-marking rather than policy argument, and they cluster almost entirely in the arrival thread: the slogan register fires at the event; the conspiratorial-misogynist register fires at the woman, three days later.

7. Fusion with misogyny

The public meta-debate around the visit was itself sharply gendered, and that context belongs here rather than in the introduction. Women who warned about a figure facing sexual-assault allegations reported being dismissed as failing to see “a greater strategy that we women don’t see because we’re emotional about sexual assault” (Silverman, Jerusalem Post); a second Jerusalem Post essay read the episode through a longer communal pattern in which those who warn about a predator inside the tent — most often women — are recast as the problem (Tayer, Jerusalem Post). The punitive energy followed the same line: creators who collaborated with Peters received death threats; a prominent pro-engagement advocate received character-directed attacks; men advancing the same arguments drew substantive rebuttal without equivalent character destruction. And the episode produced exactly one institutional sanction: the IDF disciplined the soldier who had appeared in the livestream — a suspended jail sentence, professional disqualification, and transfer — the only consequence imposed on any participant in the entire affair, an asymmetry Israeli commentary itself began to name, asking why she was the only participant ostracized while the male influencers who hosted and amplified Peters faced none.

Inside the corpus, the structure is dyadic: Peters is punished as the allegedly captured man; the soldier is punished as the alleged instrument of capture. Misogyny does more than accompany the antisemitic narrative here; it supplies its causal mechanism: the Israeli woman is represented as the sexual instrument through which Jewish or Israeli power captures the male target. In the July 13 thread, misogyny is accordingly not a parallel phenomenon: 31 of its 47 misogyny-coded units are simultaneously antisemitism-coded, and roughly 29 target the soldier or Israeli/Jewish women specifically — as honeypots, “hags,” state property “ran through” by the military, or transphobically denied womanhood:

“she is a trap” (a deliberate double meaning, fusing the transphobic slang with the honeypot claim)

“YOU GET BENT OVER BY YOUR JEWISH WHÓŘE GF DAILY.” (July 9 thread, 09:41 — heavy evasion diacritics; posted four days before the soldier photo existed — the temporal priority analyzed in §1)

Nor does the fusion confine itself to the soldier. When a journalist’s interview with Peters entered the corpus as a quote-tweet trigger, the same apparatus turned on her:

“Bari Weiss bff / concubine is rocking w you” (≈120 likes — the journalist recast as sexual property within a Jewish-network frame)

“have they let you take Hebrew classes yet or are they still treating you like a half-breēd reject?” (≈29 likes — racialized exclusion and misogyny in one address, evasion diacritic included)

And the register can invert its target: rather than degrading the women as Jewish instruments, one unit antisemitically reclassifies them as false whites — deploying the far-right “fellow whites” meme, which alleges that Jews deceptively claim whiteness to subvert from within:

“1. ‘Looksmaxing’ influencer thinks ‘All these [jewish] women are good-looking.’ You can’t make this stuff up… 2. ‘Well, hello there, fellow White women.’ Follow @[account] as we max our way out this mess” (≈100 likes)

Peters’ compliment to Israeli women is recoded as a category error — they are not white women but Jews posing as such — and the correction doubles as recruitment: the reply closes with a follow appeal, the thread functioning as an acquisition surface for white-nationalist content.

Standalone sexism is the minority pattern; the misogyny is instrumental to the antisemitic frame, degrading the woman in order to deny the relationship’s reality and complete the capture narrative. The arrival thread, by contrast, contains almost no misogyny (7 units) — the register rises sharply when a woman enters the frame. That the only institutional sanction of the entire episode fell on this same woman gives the fusion its bitter real-world coda.

8. Fusion with homophobia, and the pedophilia attribution

Two further hostility registers fuse with the antisemitic frame, and the corpus’s imagery and address forms make both unmissable. The first is homophobia as the idiom of capture: the affiliating gentile is not only bought and controlled but sexually subordinated, and the subordination is homophobically coded. The slur and the reclassification arrive fused in a single breath:

“FAGGOT JEW.”

“Ok jew, yer a faggot.”

“Fucking fag. Bending over for the Jews” (the sexual-subjugation formula with the collective marker — coded as affirmation of antisemitism)

“jew cocksucker” / “Jew cock sucker” (recurring across threads as a fixed formula)

And the register runs from bare slur to erudite allusion:

“See if they’ll hook you up with a free Magnus Hirschfeld surgery.” (≈57 likes — a Weimar-sexology dog-whistle: Hirschfeld, the Jewish founder of Berlin’s Institut für Sexualwissenschaft, invoked as the origin figure of the Jews-as-engineers-of-sexual-deviance canard; decoding it requires a historical literacy the register’s bare slurs do not)

The same logic drives the corpus’s second fabricated image — Peters kissing a kippah-wearing man, from the same self-labeled parody account as the Ben Gvir fake (see chapter 10) — and the replies that treat it as evidence (”was this your handlers making you kiss a dude?”): homophobic degradation supplied the visual proof of capture, just as misogyny supplied its causal mechanism. At the state level, the register generalizes into degeneracy attribution — Israel figured as a nation of sexual deviance — a small register in this corpus but among its most theologically loaded.

The second register is the pedophilia attribution — the oldest child-focused libel in the antisemitic repertoire. It is coded only where the child element is explicit in the unit, as in direct attribution:

“Of course all the little children are at a tiny hat club, they love them young. Disgusting.” (the kippah codeword fused with the child-predation charge)

“Be careful that they don’t record you secretly screwing underaged girls and boys. They are known to blackmail influencers and politicians (Epstein…).”

— and as morphology: the circumcision-ritual mockery coinages documented below (”Metzitzahmaxxing,” “babyforeskinmaxxing,” ≈570 likes). A boundary note governs the adjacent Epstein complex: Epstein references default to the blackmail/kompromat family, not to the pedophilia attribution, unless the minor element is stated. “No wonder you visiting Israel to run a blackmail operation like Jeffery Epstein as Mossad” therefore codes as conspiracy (§1), not as child-focused libel; the same rule holds for the trafficking insinuation aimed at the Miami club and for the AI-generated Israeli flag bearing Epstein’s face documented in the visual layer, whose recoverable content is the blackmail-state emblem. Only where a unit fuses the two — as in the “underaged girls and boys” warning above — is the child-focused concept counted. The pattern completes the disciplinary logic: the community that welcomed Peters is figured as predatory toward children, so that his presence among them becomes complicity in the ultimate crime — the same complicity transfer the atrocity register performs, executed through the libel with the longest history.

9. Lexical innovation

The reaction discourse is conducted in a subcultural vernacular fusing with antisemitic concepts in real time. The “-maxxing” template alone yields:

JEWMAXXING (≈7,900 likes) · shekelmaxxing (as a username) · goymaxxing · goyslopmaxxing · smallhatmaxxing · Metzitzahmaxxing · babyforeskinmaxxing (≈570 likes) · Genocidemaxxing

The template’s carrying capacity is demonstrated in a single unit — one author running the full classical inventory through the subculture’s morphology:

“Pedophile maxxing. / Genocide maxxing. / Satan maxxing. / Antichrist maxxing. / Goy maxxing. / Cannibal child eater maxxing. / Epstein Fortnite account maxxing.” (≈84 likes)

Blood libel (”cannibal child eater”), demonization, atrocity attribution, goy servitude, and the Epstein complex — the classical repertoire in one suffix. Beyond the template, the coinage work extends to slur formation (”kikussy”) and to the discourse’s evidence culture, self-satirized in a four-word deceit formula that ventriloquizes the Jewish voice as inherently untrustworthy:

“Source trust me goy”

The lexical layer, finally, extends into the author field itself. Handles and display names in this corpus carry the repertoire as identity rather than utterance: the goy- and shekel-morphology adopted as usernames, ethnic slurs embedded in account names, a handle declaring misogyny as a self-description, the numeric expulsion codeword “110” worn as a name, an account calling itself “tired of Jews,” and display names adopting the name of a US antisemitic hate group (Goyim Defense League) — the last already encountered attached to two of the visual exhibits above. Per protocol these accounts are not named here, but the pattern matters analytically: a comment is composed for a moment, a handle is chosen once and worn across every interaction the account will ever have. Where the reply text shows the repertoire activated by an event, the nominal layer shows it installed as self-presentation before the event existed — and it marks a moderation blind spot, since handle-level signals are invisible to systems that evaluate comments one at a time. The wall-pilgrimage taunt lexicalized as “kiss the wall (goy),” “wall kisser,” and the single-word coinage “Kissthawallmaxing” — the servility taunt compressed into the subculture’s own morphology; the far-right canon supplied “Every Single Time,” “Shut it down,” “goy cattle,” “tiny hat,” and the numeric “110”; counter-discourse answered in the same morphology (”slutmaxxing,” in a critical commentator’s coinage — evidence of the template’s cross-camp productivity). None of these items is legible to standard lexicon-based detection; all were fully legible to the in-group. Moderation-evasion orthography (”K!ke,” “jevv,” a Cyrillic о in “Gоу,” a leet-spelled “l< i l< €,” an evasion diacritic in “half-breēd”) documents users engineering around exactly the keyword filters whose insufficiency this vocabulary demonstrates.

10. The visual layer and the synthetic substrate

The image layer did not illustrate the discourse; it escalated it. Every major register that appears text-only in some units appears image-only in others: Holocaust denial entered the corpus exclusively as an image (a Flork meme mocking the “totally not made up, historically accurate number — 6 MILLION”); Holocaust derision fused with the greed stereotype in a single exhibit — “Ovenmaxxing oh vey” over a photograph of an open oven with a trail of dollar bills leading into it, the subculture’s morphology carrying a Shoah joke and the money stereotype simultaneously; Hitler valorization as an “ABSOLUTE CINEMA” macro; the canonical “happy merchant” caricature appears holding a chained, collared gentile on a leash (”That’s right goy get em”); a “Judas goat” explainer meme casts Peters as the trained animal that leads the sheep to the slaughterhouse “again and again”; theological antisemitism as a Revelation 2:9 “synagogue of Satan” scripture card; movement signaling as a “FREEDOM FROM ISRAEL & ZOG” t-shirt selfie and the apparent Rockwell photograph; and the ambient far-right meme repertoire — Pepe the Frog variants, Nazi meme templates, and the “happy merchant” leash image quoted above: goy servitude rendered literal — running through the threads as the discourse’s visual vernacular (the serpent-and-eagle control image is presented in §1). The purchase rumor acquired its own artifact — a fabricated State of Israel check:

Most consequentially, part of the discourse responds to synthetic events. The early comment sections are flooded with deepfakes pairing Peters with National Security Minister Itamar Ben Gvir in front of the Wailing Wall — Peters giving a thumbs-up beside the minister — watermarked by a self-labeled parody account, the same source as a fabricated image of Peters kissing a kippah-wearing man. The image recurs at least six times in the arrival thread and continues circulating in subsequent threads, anchoring a measurable cluster of hostile replies, including a 915-like condemnation and units that treat the “meeting” as established fact:

“Were you really chillin with Ben-Gvir?”

“Is that picture of you with Ben Gvir in front of the wall real?” (July 9 thread — the fabrication circulating as an open factual question)

No press coverage reports any Ben Gvir meeting; the detailed trip reconstructions record only a coincidental encounter with a Netanyahu adviser. Further judaisation imagery in the corpus is explicitly labeled “Made with Grok Imagine,” and commenters routed the discourse through AI systems in both directions. In one direction, racial classification itself is outsourced to the platform’s built-in model:

“Hey @grok is Clavicular jewish?”

Seven words, and the question “who is a Jew” — the reclassification apparatus of §3 — is handed to an AI system for adjudication. In the other direction, generative tools manufacture the “evidence” for both hostile narratives at once: the capture story (the wall-kiss image that drove “Free Clav”) and the far-right-alliance story (the Ben Gvir image). One parody account supplied the factual substrate for both. On the evidence available, the defensible formulation is this: a measurable cluster of replies treated an apparently fabricated image as evidence of a meeting for which no corroboration exists in any reporting or trip reconstruction.

Discussion

The controversy surrounding Peters’ visit produced sharply opposing normative judgments but almost no empirical evidence. The present findings allow that debate to be revisited from the perspective of the receiving audience.

What the community’s visible response contained. The corpus cannot measure silent viewers’ attitudes, livestream reception, or circulation on other platforms, and this limitation is stated plainly: reply sections over-represent engaged and hostile voices. The findings describe one influencer, one event, and one platform; how far they generalize remains an empirical question, for which the Peterson comparison provides only preliminary evidence. But the guiding question — what happened inside his own community — has a direct answer at its most public surface, the reply environment on Peters’ own account, the space every onlooking fan and creator could read. It contained: a 26,000-like verdict that Israel contact destroys careers; a 21,000-like delegitimization slogan; an 8,800-like child-murder attribution; a capture narrative so dominant it lexicalized into single words; and the systematic recoding of the visiting gentile as a Jew. Within this environment, presence in Israel is itself coded as a conspiratorial act — the “apolitical” framing collapsed under pressure from both political directions. Whatever happened in channels this study cannot observe, any strategy premised on influencer diplomacy toward this milieu must reckon with the finding that its most visible discourse space possesses a disciplinary apparatus already fitted to this case, and that within that space pre-existing disciplinary frames frequently displaced engagement with the substance of the visitor’s content.

Antisemitism as discipline, and the dethroning arc. A definition first. In-group discipline refers to the public punishment of a perceived group member for violating a norm of affiliation, loyalty, or identity. What distinguishes it from ordinary criticism is its structure: the target is construed as corrupted, his autonomy denied, his identity rewritten, his membership revoked — and the punishment is performed publicly, for others to see. What reads as insult functions as exemplary punishment. The theoretical core of the case is this: rather than expressing hostility toward Jews alone, the discourse in the reply environment studied here polices the boundaries of the in-group by transforming affiliation with Jews into evidence of corruption. The accusation performs a disciplinary rather than merely descriptive function — it does not only assert something about Jews; it does something to the affiliating member, and through him to every onlooker weighing the cost of proximity. The general form of the claim: antisemitic repertoires here operate as social technologies for regulating in-group boundaries — the tropes are old, but alongside their familiar out-group hostility they perform an additional function in this corpus: boundary maintenance.

These concepts join established literatures rather than replacing them. Boundary maintenance and symbolic exclusion are established objects of sociological research (Lamont & Molnár 2002); stigma and spoiled identity are central concerns of symbolic interactionism (Goffman 1963); and in-group policing has been examined in conflict research (Fearon & Laitin 1996). Affiliation punishment, in this framing, is a specific empirical manifestation of norm enforcement. What it specifies in the present analysis is that the sanction’s content is drawn from an out-group hate repertoire and that its terminal move reclassifies the sanctioned member into the stigmatized out-group itself. Generic norm enforcement disciplines a member as a deviant member; the mechanism documented here disciplines him by rewriting what he is — the punishment for proximity to Jews is to be made a Jew. That specification is what makes the process antisemitic in form and not merely in occasion. The register inventory identifies the mechanisms through which affiliation is converted into corruption: each register supplies part of the sanction’s vocabulary, and their density indicates how extensively the discourse performed this disciplinary function. The corpus’s central mechanism so understood — antisemitic concepts deployed against the affiliating gentile — closely reproduces the four-group sequence documented for the Peterson–Netanyahu reactions (Becker 2024): (A) disappointment and desacralization; (B1) victim of a vague external power — the servility register; (B2) victim of Jewish power — honeypot, handler, shekels, orchestration; (C) “true colors” — “There’s was always something about you we hated… Now we know”; (D) judaisation and exclusion — Kikevicular/Claviculberg as the exact suffix-pattern of “Juden Peterstein.” The recurrence of a closely related pattern across target type (academic intellectual → Gen-Z influencer), platform (YouTube → X), and four years suggests — pending a formal comparative analysis with a shared instrument — that the dethroning arc may be a recurrent interactional pattern of this milieu rather than a case-specific reaction — a hypothesis for systematic comparative testing. The Clavicular corpus adds a fifth stage the Peterson corpus lacked: AI-generated visual judaisation. The arc is an analytical sequence, ordered by conceptual escalation — its stages co-occur within threads, and the corpus demonstrates the recurrence of the configuration, not that individual authors traverse the stages in time. The arc, displayed:

Figure 3. The dethroning arc.

Affiliation
↓
Disappointment
↓
Servility attribution
↓
Capture (Jewish power)
↓
Judaisation / reclassification
↓
Expulsion
↓
Public disciplinary signal to the surrounding community

Synthetic media can inject fabricated “evidence” at any stage.

The arc’s function is therefore deterrent as well as expressive: it publicly demonstrates the costs of affiliation, addressing every onlooking creator as much as Peters himself.

The synthetic substrate. In plain language first: some users in this corpus were reacting not to events that happened but to fabricated images that supplied apparent evidence for narratives already circulating. The result is what this study calls a synthetic factual substrate — false visual material functioning as the evidentiary basis for real hostility. Hostile movements have always fabricated evidence — photographs, documents, quotations. What generative systems change is the economics: they lower the cost and time required to manufacture apparent visual evidence and insert it into an unfolding controversy, at platform speed, customized to the narrative that needs it. Conspiracy discourse has always fabricated explanations: hidden hands behind visible events. Here it fabricated the evidence as well — not an interpretation of a meeting, but the meeting itself. This corpus documents that operation in real time rather than in prospect. That one self-labeled parody account supplied the “evidence” for two mutually exclusive hostile framings — Peters as captured victim and Peters as far-right ally — and that the discourse absorbed both without friction, suggests the epistemic structure documented in the conspiracy-mutation literature operating at meme speed: the frames selected their evidence, and when reality declined to supply it, the tools did. The subject’s own reported attempt to have the content deleted (documented only in the fabricator’s self-published screenshot) completes the recursion: the sole testimony about the fake’s fakeness comes from the faker. The loop, displayed:

Figure 4. The synthetic evidence loop.

Real event (the visit)
↓
Fabrication (parody-account images, watermarked)
↓
Circulation stripped of framing (reposted as authentic)
↓
Uptake as evidence (hostile replies cite the image as fact)
↓
Contested only within the loop (the subject’s reported deletion demand survives solely in the fabricator’s self-published screenshot)

Each pass hardens the fabrication’s status as shared factual ground.

Methodological implications. Three follow. First, text-only analysis materially undercounts and mischaracterizes this discourse: the visual pass raised the arrival thread’s coded rate by several points, flipped several units from harmless to explicit, and contained the corpus’s only Holocaust-denial and neo-Nazi-iconography units. Second, keyword-based detection is structurally blind to the corpus’s most consequential material — the total-loss family, the “-maxxing” coinages, the numeric codewords, the evasion orthography, and Bilzerian’s keyword-free verdict. Third, headline percentages are boundary-dependent: this study’s coding decisions — delegitimization and norm-affirmation counted as constitutive, servility reported adjacent to but outside the coded corpus — are therefore documented explicitly in the research design, and the servility boundary demonstrates that the instrument retains a genuine stopping point.

Why this matters

For influencers and creators: affiliation with a stigmatized out-group can trigger coordinated identity punishment rather than ordinary disagreement — the response documented here was not criticism of content but revision of the person.

For policymakers and platforms: moderation systems are built to detect hostility directed at protected groups; this corpus demonstrates hostility directed at a non-member through antisemitic repertoires — a category current governance frameworks do not name — and the disciplinary sequence runs through coded vocabulary, usernames, images, and fabricated evidence, with the nominal layer (hate worn as a handle) invisible to message-level moderation entirely.

For advocacy organizations: reach is not persuasion; before strategic engagement with any milieu, the receiving community’s norms — here, a ready-made apparatus for punishing this specific affiliation — need assessment, because the messenger’s audience may discipline the messenger faster than the message can land.

For educators: this case supports teaching how conspiracy narratives regulate in-group loyalty — antisemitism appearing not only as hostility directed at Jews but as a repertoire used to punish non-Jews for perceived proximity to Jews or Israel — several of this corpus’s most-liked units contain no Jewish referent at all and are legible only through that lens.

For researchers: the meaning here is distributed across text, imagery, handles, interaction, and prior discourse; instruments that read messages one at a time, in one modality, will systematically undercount the phenomenon.

Conclusion

The central disagreement surrounding the visit was never simply whether Peters should have been invited. It was whether such engagement would generate visible reconsideration within his public online environment. To our knowledge, this study provides the first empirical evidence bearing on that question. His own reply sections supply the one part of the answer that is directly observable — and it is stark. The visible milieu received the visit as a betrayal, processed it through a ready-made disciplinary apparatus, enforced it through casual enactments of a pariah norm, and pronounced its verdict through its two most prominent voices, whose three replies drew approximately 54,000 likes combined. In the dominant and most highly rewarded layer of the reaction, the gentile who publicly affiliated with Israel was not engaged on the substance of the visit; he was recoded and expelled.

The visit placed this community in front of an opportunity to reinterpret a familiar object through direct exposure. Within the observable reply environment, the findings consistently indicate one outcome: we found little evidence of public attitude revision — and the stronger formulation is warranted: the reply environment did not ignore the visit but repeatedly reinterpreted it through pre-existing narratives of betrayal, capture, and exclusion. The case is consistent with a broader capacity of established ideological frameworks to metabolize what contradicts them without altering themselves. Not new thinking, but the old thinking, reinforced — and armed with new tools.

References

Becker, M. J. (2024). Jordan Peterson and conservative antisemitism online: The dethroning of an intellectual icon following his interview with Netanyahu. In M. J. Becker, L. Ascone, K. Placzynta, & C. Vincent (Eds.), Antisemitism in online communication: Transdisciplinary approaches to hate speech in the twenty-first century (pp. 47–74). Open Book Publishers. https://doi.org/10.11647/OBP.0406.02

Becker, M. J. (2025). Charlie Kirk’s murder and the digital afterlife of political violence: A YouTube discourse analysis. Decoding Antisemitism Substack, September 2025. https://decodingantisemitism.substack.com/p/charlie-kirks-murder-and-the-digital

Becker, M. J. (2026a). Antisemitism in real time, Part III: “Israel made America do it” — YouTube reactions to the US–Israel campaign against Iran. Decoding Hate, NYU Center for the Study of Antisemitism. https://decodingantisemitism.substack.com/p/antisemitism-in-real-time-part-iii

Becker, M. J. (2026b). Antisemitism in real time, Part IV: “A fake feud” — YouTube reactions to the Trump–Iran deal and the Trump–Netanyahu rift. Decoding Hate, NYU Center for the Study of Antisemitism, June 29, 2026. https://decodingantisemitism.substack.com/p/antisemitism-in-real-time-part-iv

Becker, M. J. (2026c). From abuse to “organized state policy.” Decoding Antisemitism Substack, May 2026. https://decodingantisemitism.substack.com/p/from-abuse-to-organized-state-policy [responds to Kristof’s NYT column of May 11, 2026]

Becker, M. J. (n.d.). At the bottom of the iceberg. Decoding Antisemitism Substack. https://decodingantisemitism.substack.com/p/at-the-bottom-of-the-iceberg

Becker, M. J., Deutsch, I., & Verser, J. (2026). Catchy hatred: YouTube reactions to Kanye West’s “Heil Hitler” — Sympathetic amplification: Host framing and audience response across ten YouTube threads. Decoding Antisemitism Substack. https://decodingantisemitism.substack.com/p/catchy-hatred-youtube-reactions-to

Becker, M. J., Troschke, H., Bolton, M., & Chapelan, A. (Eds.). (2024). Decoding antisemitism: A guide to identifying antisemitism online. Palgrave Macmillan. https://link.springer.com/book/9783031492372

Fearon, J. D., & Laitin, D. D. (1996). Explaining interethnic cooperation. American Political Science Review, 90(4), 715–735.

Goffman, E. (1963). Stigma: Notes on the management of spoiled identity. Prentice-Hall.

Lamont, M., & Molnár, V. (2002). The study of boundaries in the social sciences. Annual Review of Sociology, 28, 167–195.

This case study was conducted under Decoding Hate, the research hub at NYU’s Center for the Study of Antisemitism, and draws on the Decoding Antisemitism Lexicon. It connects to the DH study “Catchy Hatred” — on the YouTube reception of Kanye West’s “Heil Hitler,” the song performed at the Miami club episode in which Peters participated — and to the Antisemitism in Real Time series.*

Matthias J. Becker — AddressHate Research Scholar at NYU’s Center for the Study of Antisemitism | Lead, Decoding Antisemitism — PI, Decoding Hate | Research Advisor, AddressHate | Editor-in-Chief, Digital Hate Review

Isabelle (Izzy) Deutsch

Jameson Verser — Decoding Hate Researcher, NYU’s Center for the Study of Antisemitism

Executive Summary

In May 2025, Kanye West released “Heil Hitler,” a song whose title, lyrics, and music video explicitly draw on Nazi symbolism, archival Hitler audio, and antisemitic conspiracy narratives. Major streaming platforms removed the song within days, but it continued to circulate widely on YouTube through news coverage and reaction videos. This secondary circulation makes the case particularly well suited for examining how explicitly antisemitic content is framed, propagated, and recontextualized within platform discourse. The study also explores how the song’s openly antisemitic rhetoric is taken up across different online environments, and which forms of endorsement, rejection, elaboration, and normalization emerge in audience responses.

This study analyzes 1,000 comments from ten YouTube videos posted across the first ten days after release. All threads examined engage the same song; what varies is the communicative context created by different hosts — adversarial hosts who condemn the song (e.g., Piers Morgan), right-leaning commentators, progressive hosts, reaction-creators, and independent commentators. The corpus also includes an instance of sympathetic amplification (Joe Rogan), in which the host repeats or normalizes central interpretive framings of the song on camera. One pre-release comparator YouTube thread is retained as immediate context. All comments were coded using a structured antisemitism lexicon and a three-annotator consensus process.

Four findings stand out.

First, antisemitic communication appears in substantial volume across every host condition examined. Across the corpus, 26.0% of all coded comments contain one or more antisemitic elements; per-thread rates range from 12% to 46%. Antisemitic content does not concentrate only in unambiguously extremist spaces but also surfaces in comparatively mainstream YouTube formats.

Second, host condition shapes the communicative norms of a thread. The phenomenon of sympathetic amplification makes this especially visible. We define sympathetic amplification as an on-camera move in which a host treats the song’s overt Nazi references and conspiracy-ideological message as plausible, treats its removal from platforms as confirmation of that message, endorses substantive claims of the song, or expressly praises its message. Under this definition, sympathetic amplification does not simply increase antisemitic content; in this corpus, it changes the character of the discussion itself. Host condition shapes what we call audience permission structures — the communicative norms of a thread: which statements are treated as self-evident, which draw pushback, and which forms of antisemitic articulation appear as legitimate. Under adversarial hosts, commenters attack the host and defend the song. Under sympathetic hosts, they extend the song’s framings, confirm antisemitic tropes, and normalize their open articulation. At the same time, the Joe Rogan thread produces the corpus’s densest substantively engaged counter-speech, indicating that sympathetic amplification mobilizes both endorsement and heightened public pushback.

Third, and most striking: even a pop song whose Nazi reference could hardly be more explicit does not, in the YouTube discussions examined here, meet with widespread rejection. Across every host condition, the corpus surfaces substantial endorsement, further development of antisemitic narratives, and — at a density we have not observed in prior studies in this research line — open affirmation of the Nazi reference itself. The extraordinary explicitness of the song does not produce correspondingly strong social sanction; instead, it opens a form of antisemitic follow-on communication that extends beyond the reproduction of classic antisemitic conspiracy narratives. Substantive counter-speech, by contrast, accounts for only 3.2% of coded entries corpus-wide (32 of 1,000), with one thread containing none at all.

Fourth, the results suggest that the societal opportunity structures for antisemitic communication have shifted in recent years. What now proves resonant is no longer only coded or indirect antisemitic material. Even openly Nazi-referential communication can, in broader digital publics, meet with endorsement, applause, and discursive further development. That these dynamics appear not only in unambiguously extremist spaces but also in comparatively mainstream YouTube formats is among the study’s most notable findings. The present case therefore documents not only the reception of one song. It also points to a broader shift in the conditions under which antisemitic communication circulates in digital space, acquires public traction, and seeps into online discourse more widely.

Introduction

Public communication about identifiable social groups does not remain on the stage where it is first articulated. It circulates through digital platforms, is amplified by news organizations, commented on by influencers, and further developed thousands of times in online discussion spaces. In this process, statements are not simply repeated. They are affirmed, contested, intensified, reinterpreted, and endowed with new meanings. Particularly within digital publics, the societal significance of public communication therefore arises not only from the original speaker but also from the secondary discourse that the speaker’s statements generate.

This dynamic becomes especially consequential when public communication operates through demonization. We use this term to describe representations that attribute to an identifiable group an essential malevolence, hidden coordinated power, or destructive collective purpose. Such narratives do more than shape perceptions of the targeted group. They alter the communicative conditions under which public discourse unfolds. Political disagreement gives way to enemy identification; complex social developments are reduced to conspiratorial explanations; and democratic conflict increasingly comes to be framed as a struggle against a supposedly existential adversary. These effects are structural: they obtain regardless of the speaker’s intentions or declared political commitments.

The consequences extend well beyond the targeted group itself. Demonizing communication can contribute to intimidation and withdrawal from public discourse, normalize hostility toward minority communities, and gradually shift the boundaries of what becomes publicly acceptable. Particularly significant is the trivialization of historical atrocity through popular culture. When Nazi symbolism, Holocaust references, or Hitler imagery become components of mass entertainment, the question is no longer merely how widely such content circulates, but how digital publics respond to it. What is ultimately eroded is not only recognition of the historical crime itself, but also the mutual recognition upon which democratic public life and social cohesion depend.

The present study examines precisely this question. It does not attempt to infer Kanye West’s intentions; discourse analysis cannot establish what a speaker ultimately believes. Instead, it investigates the communicative consequences of the song and asks how its openly antisemitic and Nazi-referential message is received, interpreted, contested, elaborated, and normalized across different digital publics.

The analytical significance of this case lies in the communicative object itself. Prior studies in this research program have examined audience responses to coded, indirect, or ambiguously-signalled forms of antisemitic discourse embedded in political speeches, media commentary, interviews, and public debate. Here, by contrast, the communicative object is a song built around explicit Nazi references. “Heil Hitler” combines antisemitic conspiracy narratives with Nazi symbolism in a form whose explicitness can scarcely be exceeded. The central question is therefore no longer whether audiences recognize antisemitism when it is communicated indirectly, but how digital publics respond when antisemitic and Nazi-referential communication requires no interpretation at all.

The case is analytically unusual for another reason. Although major streaming platforms removed the song within days of its release, it continued to circulate widely through YouTube news coverage, commentary programs, podcasts, and reaction videos. As a result, the same communicative object repeatedly entered different interpretive environments, where hosts condemned, defended, normalized, or contextualized it in markedly different ways. That trajectory extended well beyond the initial release window. By April 2026, the artist now known as Ye was on a stadium comeback tour. The album “Bully” debuted at No. 2 on the Billboard 200 in late March, and two SoFi Stadium shows sold out in early April. Days later, the United Kingdom denied him entry and the Wireless Festival he was set to headline was cancelled. A CNN feature on 13 April 2026 described the moment as a “dichotomy of a star both banned and celebrated” (Respers France 2026). The precondition for the tour was a January 2026 full-page Wall Street Journal advertisement headlined “To Those I’ve Hurt,” in which West apologized for prior antisemitic remarks and attributed them to an undiagnosed brain injury and untreated bipolar type-1 disorder — “I am not a Nazi or an antisemite,” he wrote, “I love Jewish people.”

Alongside that contrition, the audience-facing circulation of the earlier content has continued. Days before the WSJ apology, video circulated of far-right figures — Nick Fuentes, Sneako, Andrew Tate, Myron Gaines — singing along to “Heil Hitler” at the Miami Beach nightclub Vendôme, drawing condemnation from the venue, the mayor, and the Auschwitz Jewish Center Foundation. This article documents one stage of that trajectory: the YouTube comment discourse the song generated when it was released a year earlier. The May 2025 release itself sits within a three-year trajectory of antisemitic content from Kanye West, from the “death con 3 on JEWISH PEOPLE” tweet of October 2022 through the May 2025 song release, as traced by the Blue Square Alliance Against Hate’s Command Center.

The variation was substantial. Some commentators treated the song as indefensible antisemitic provocation; others framed it as taboo-breaking or as evidence of censorship. In one of the JRE clips analyzed for this study, Rogan describes the song as “kind of catchy” and articulates a framework in which platform removal functions as confirmation of the conspiracy the song articulates: “If you ban it, then people want to hear it more… and then it kind of supports what he says, which is that there’s this concerted effort, if you talk about Jewish people, that they’re going to remove you from everything” (Gilson 2025a). Piers Morgan condemns the song on-camera as “indefensible” in two of the threads analyzed here. Outside the present corpus, Russell Brand defends the song as artistic provocation and calls West “uncancellable” (Gilson 2025b). This variation creates an unusually favorable setting for discourse analysis: the object of discussion remains largely constant while the surrounding context changes.

To examine these dynamics, the study analyzes 1,000 comments posted beneath ten YouTube videos discussing the song during the first ten days following its release. The central question is what role host framing plays once the communicative object itself is already maximally explicit. Rather than asking only whether antisemitic responses occur, the analysis examines how different communicative environments influence which forms of audience participation become plausible, legitimate, or socially expected. Under what conditions do users defend the song, attack its critics, elaborate its conspiracy narratives, reject them, or openly affirm its explicit Nazi references? Addressing these questions allows the study to move beyond measuring the prevalence of antisemitic discourse toward identifying the communicative mechanisms through which explicitly antisemitic messages acquire legitimacy, become normalized, and are further developed within contemporary digital publics.

Contribution

The study makes three contributions:

1. To our knowledge, it provides the first cross-comparison of audience responses to a song built around explicit Nazi references across multiple YouTube host conditions.

2. It introduces audience permission structures as a discourse-analytic construct operationalized from comment data.

3. It identifies explicit Affirmation of Antisemitism as a distinct response register — surfacing at densities not documented in prior speaker-driven studies of coded antisemitic discourse in this research line.

The Track and the Artist

The Track

“Heil Hitler” is a single song, not an album. It was released on 8 May 2025 via West’s X (formerly Twitter) account as the intended third single from his then-forthcoming studio album CUCK, a project that later leaked and was scrapped. The song is written by Dave Blunts, produced by Quadwoofer and Sheffmade, and features additional vocals from West’s group The Hooligans. Unlike the speaker-driven cases examined previously in this research line, the song’s Nazi referent is explicit rather than coded. The title is the Nazi salute in its original German; the song incorporates archival Hitler audio from a 1935 speech; the lyric articulates a standard set of antisemitic conspiracy claims (Jewish institutional control, banking conspiracy, persecution-of-the-truth-teller framing) within a verse-and-chorus pop structure designed for circulation; and the accompanying music video depicts a warehouse of Black men in militaristic formation, draped in animal skins and mouthing the chorus alongside The Hooligans. The Nazi reference operates at the surface of the song’s title, audio, lyrics, and visuals, rather than requiring interpretive reconstruction.

The song was removed from major streaming platforms (YouTube, Spotify, Apple Music, SoundCloud) and is banned in Germany under §86a StGB, but remained accessible on X and Meta-owned platforms during the May 2025 reception window this corpus documents. West released two variants that month: an instrumental version on 14 May 2025 (“The Heil Symphony,” later renamed “Hit Symphony”) and a partially censored gospel-themed rewrite on 31 May 2025 (“Hallelujah”) that replaced the chorus’s Nazi salute with the title word. After meeting Orthodox rabbi Yoshiyahu Yosef Pinto in November 2025 and apologizing, West’s company YZY began issuing copyright strikes against reuploads of the original song. The January 2026 Vendôme nightclub incident discussed in the Introduction is one documented instance of the song’s circulation continuing well beyond the May 2025 window this corpus samples.

The song’s methodological significance is its boundedness: a single object, under five minutes long, encountered as the same lyrical-musical-visual referent across every host context in the corpus. All ten videos are organized around this same referential object, though they engage it differently: some play extended clips, some quote excerpts, some summarize, and some show the music video, with on-screen attention ranging from seconds to minutes. The song is therefore stable across the corpus in a sense not available to speaker-driven studies that rely on distributed, multi-segment discourse; the host-framing variable captures the differences in how the videos present, quote, and contextualize it.

The Artist

Kanye West is one of the most-followed cultural producers of the past two decades. The 2025 release is not his first antisemitic episode; the 2022 cycle — including the “death con 3 on JEWISH PEOPLE” tweet, the Alex Jones interview, and the “I like Hitler” remarks — had already established an audience familiar with and engaged by his antisemitic content. Those 2022 statements were widely treated by platforms, advertisers, and commentators as a public scandal: brand partnerships dissolved, accounts were suspended, and mainstream commentary largely converged on condemnation. The 2025 release drew similarly intensive institutional criticism — streaming platforms removed the song within days, public condemnation was widespread, and the WSJ apology in January 2026 acknowledges the trajectory of fallout. What differs between the two cycles is less the volume of friction than its form of circulation. In 2022, condemnation occupied the discursive foreground; in 2025, the explicit Nazi reference circulated alongside condemnation through reaction-and-commentary content on YouTube, where extended host engagement and audience comments engaged the song directly across multiple host conditions. The three-year trajectory is part of what the case documents.

Dataset

The corpus comprises 1,000 user-generated comments from ten YouTube videos around the “Heil Hitler” release, spanning roughly ten days (9–17 May 2025), plus one near-predecessor video (KW-01, 6–7 May 2025) retained as immediate context. The first 100 chronologically ordered comments per video form the per-thread sample; top-level comments and reply comments are coded and counted the same way, with reply chains preserved for sequence and relational analysis. Spam and duplicates were excluded.

The corpus listing below is ordered chronologically by date posted. Each entry is tagged with its host-context type, since the analysis turns on host framing variation. In plain terms: we classify each video by how the host frames the song on camera — condemning it, defending it, reacting to it, or treating it as proof of a conspiracy. The host conditions are operationalized as follows:

• Adversarial-host. The host explicitly condemns the song and rejects its claims. The song is treated as content to be opposed, not engaged on its merits.

• Progressive-host. Left-leaning political commentary that condemns the song but situates it within a broader political-cultural critique.

• Right-host. Conservative political commentary; positioning toward the song ranges within the right-political spectrum from condemnation to defense.

• Independent / hybrid. Centrist or politically heterogeneous commentary; evaluation not anchored to a primary partisan position.

• Reaction-creator. Single-creator reaction content; primary mode is real-time response rather than political-editorial framing.

• Sympathetic-amplifier. The host repeats, legitimizes, or normalizes one or more of the song’s central framings during on-camera commentary. The category is operationalized as a decision rule: a host is coded as sympathetic-amplifier when at least one of the following criteria is met on camera. (i) The host treats the song’s conspiracy claim as plausible or worth taking seriously, rather than dismissing it. (ii) The host treats the song’s removal, demonetization, or other moderation outcome as confirmation of the conspiracy the song articulates. (iii) The host endorses a substantive factual premise of the song (e.g., that a particular industry or institution is controlled by Jewish actors). (iv) The host explicitly praises the song’s message or its rhetorical effectiveness in a way not separated from the song’s antisemitic content. Aesthetic praise alone (e.g., “the beat is good”) does not satisfy the criterion unless accompanied by engagement with the song’s substantive claims; condemnation of the song’s content while expressing curiosity or amusement does not satisfy the criterion. The criteria are intended to distinguish hosts whose on-camera articulation supplies discursive material the audience can pick up (Rogan’s articulation of (i) and (ii) above) from hosts whose framing remains evaluatively external to the song even when politically sympathetic to West more generally (e.g., Knowles, who is coded as right-host rather than sympathetic-amplifier because his on-camera articulation does not meet any of (i)–(iv)).

• Sympathetic-ambiguous. Host positioning is favorable to the song but qualified — favorable evaluation with conditionality that reads as rhetorical hedge rather than substantive constraint. The category is retained for analytical scope: it has no instances in the present ten-video corpus, but is preserved in the typology because the distinction between sympathetic-amplifier and sympathetic-ambiguous remains analytically operative for future cases and for comparison with prior research-line studies in which such hosts appear.

The Adversarial-host and Progressive-host categories above admit substantive sub-types that emerged inductively from the data rather than being specified a priori, and are tracked in the results below as data-driven refinements of the base typology. Adversarial-secular (KW-01, KW-04) and Adversarial-religious (KW-05) distinguish two grounds of adversarial framing observed in the corpus: secular-liberal condemnation (Morgan) and Christian-conservative condemnation (CBN News), each of which surfaces a partly different audience-permission-structure. Meta-progressive (KW-10) was induced from KW-10 specifically: Seder critiques Rogan’s reception of the song rather than commenting on the song directly, producing a distinctive argument about antisemitism rather than articulation of it register. The Meta-progressive subtype is therefore a category generated by the empirical material rather than imposed on it, and it does not currently extend beyond the single thread that motivated it. Whether it generalizes to other meta-discursive progressive cases is a question for future corpora. These sub-types are reported where they bear on the results; the base typology above remains the analytical frame.

Taken together, the typology contains six base host families (Adversarial-host, Right-host, Progressive-host, Reaction-creator, Sympathetic-amplifier, Independent/hybrid), with two of those families sub-divided into observed variants from the corpus (Adversarial → secular vs religious; Progressive → standard vs meta-progressive). This produces eight observable host conditions instantiated in the corpus: Adversarial-secular, Adversarial-religious, Right-host, Progressive-host, Reaction-creator, Sympathetic-amplifier, Independent/hybrid, and Meta-progressive. Sympathetic-ambiguous is preserved in the base typology for future cases but has no instances in the present ten-video corpus. The cross-thread results table below uses the eight-condition vocabulary.

KW-01 (Piers Morgan walk-out) is included as a near-predecessor to the song’s release: the interview was recorded on 6 May 2025, two days before the 8 May release, and addresses Kanye West’s antisemitic record in general rather than the “Heil Hitler” song specifically. It is retained in the corpus as immediate context for the song’s reception environment and as one half of the Morgan host-framing pair (with KW-04, the Piers Morgan Panel discussion of the song itself). Its inclusion is flagged here so that readers can weight its analytical contribution accordingly.

Chronological Corpus Listing

Research Design

Method

Analysis followed the principles of qualitative content analysis, with native-speaker coders applying linguistic and multimodal close reading supplemented by sequence and relational analysis where comments form part of reply chains. Coding combined deductive categories from the Decoding Antisemitism Lexicon (Becker et al. 2024) — 40 antisemitic concepts organized across categories including Otherness, Moral/Supernatural Defamation, Dehumanization/Blood Libel, Deceit/Blame, Power/Conspiracy/Economic Domination, Victimhood Inversion, Holocaust & Memory Politics, Israel-Related antisemitism, and Aggressive Speech Acts — with inductive coding of formations specific to the Kanye West case. The Lexicon accounts for dog whistles, sarcasm, irony, visual and emoji-based signaling, and inversion structures.

In plain terms: every comment was read and coded against a structured antisemitism framework covering conspiracy claims, direct Nazi references, Holocaust distortion, Israel-related abuse, and direct threats — with three annotators coding independently and resolving disagreements through discussion until they reached consensus.

Two case-specific coding decisions bear noting. First, comments endorsing the song or the artist on the basis of the song's antisemitic content (e.g., “hail ye,” “I stand with Ye!”) are coded as Affirmation of Antisemitism (C4-3) with a secondary Conspiracy code (C2-3) where the endorsement is framed in terms of “they're hiding the truth” reasoning. Second, comments directed at the hosts — most prominently Piers Morgan as “Mossad asset,” “puppet for Israel,” “koshervative” — are coded under Conspiracy (C2-3) when they invoke puppet-family logic (asset, puppet, agent, pawn, stooge, payroll, controlled-by, on-behalf-of), per the team's standing puppet-rule. The rule fires only on the commenter's own on-behalf claims: quoted or mocked puppet-constructions do not trigger C2-3. Power (C2-2) is reserved for bare-influence claims that do not invoke string-puller logic (e.g., “Zionist megaphone”).

The study applies the same annotation protocol and counting unit as prior pieces in this research line.

Coding Protocol

The unit of analysis is the YouTube comment, whether top-level or a reply. The first 100 chronologically ordered comments per video form the per-thread sample; top-level comments and reply comments are coded and counted the same way. Comments were excluded if they contained no semantic content (empty, single emoji, single character), if their language fell outside the coder team’s working competence and machine translation did not yield a stable reading, if they addressed a topic unrelated to the song in a multi-topic video, or if they were clear duplicates.

All comments were independently coded by three annotators (Becker, Deutsch, Verser), who resolved disagreements in adjudication sessions until they reached consensus. The coding scheme is the Decoding Antisemitism Lexicon (Becker et al. 2024) — 40 concepts plus sub-registers and multimodal signals — supplemented by a small set of corpus-specific standing rules, such as treating puppet-family metaphors as Conspiracy rather than Power and coding aesthetic praise of “Heil Hitler” as Affirmation absent explicit distancing. These rules and their full rationale are documented in the Standing Coding Rules appendix. The per-thread rates and concept-frequency counts in this essay reflect the post-adjudication consensus.

Analytical Categories

The six-category schema below adapts the Decoding Antisemitism Lexicon and the Temple Israel framework to the distinctive features of this case — specifically, a song built around explicit Nazi references and an Affirmation register that prior speaker-driven cases did not surface at comparable density. Categories are ordered by their empirical density in this corpus.

1. Affirmation of Antisemitism. This category captures direct endorsement of the song’s Nazi reference and Hitler-rehabilitation content, including explicit slogans, salute language, and “Hail Ye”-style praise. In this corpus we also group certain Otherness, Deceit, and Historical-Precedent moves (for example the “109 countries” trope or theological delegitimization) under Affirmation when they function as scaffolding for explicit endorsement rather than as standalone themes.

2. Taboo and Self-Victimization. This category covers “you can’t criticize them” narratives, persecution-of-the-truth-teller framing, and cancellation-as-evidence inversions, along with Relativization moves that defend antisemitic content as free speech, mere aesthetics, or misunderstood provocation.

3. Power and Conspiracy Narratives. This category combines Power claims (that Jewish actors control media, politics, finance, or public opinion) with Conspiracy claims about hidden coordinated action behind specific events, including host-as-puppet recoding, historical-event revisionism, and “AshkeNazi”-style Hitler rehabilitation. Power and Conspiracy are separate Lexicon codes, but in this corpus they sit on the same terrain — “they run things” and “they orchestrated this event” — and are therefore presented together.

4. Threats and Calls for Retaliation. This category includes eliminationist content, Holocaust-completion language, threats against specific targets, and Blame-for-Antisemitism moves that attribute responsibility for antisemitism to its victims.

5. Israel-Related Antisemitism. This category covers Nazi and fascism analogies applied to Israel, Blood-Libel projections onto Israeli military forces, instrumentalization of antisemitism and the Holocaust as deflection from the song’s content, and Gaza/genocide framings used to reorder discursive priority.

6. Holocaust Distortion and Denial. This category captures numerical revision (for example “6 million” vs “271k” substitutions), false-equivalence framings, and denial of the Holocaust’s historical scale. It is conceptually distinct from Affirmation: Affirmation endorses the Nazi reference itself, whereas Holocaust Distortion contests the historical record of Nazi violence. The two can dual-code (for example in Hitler-rehabilitation comments) but function as separate discursive moves.

A Schadenfreude / Sarcasm register operates across the corpus but in this case functions principally as affective cover for the Affirmation register (e.g., “I did Nazi that coming”); where this dual-coding is operative, the comment is coded under Affirmation of Antisemitism and contributes to the antisemitism rate through that route (per the humor-as-affirmation rule; see Standing Coding Rules appendix).

These categories structure the empirical chapter that follows; detailed sub-registers (e.g., Affirmation A1–A4) and illustrative comments are discussed there rather than in this section.

Three-Phase Circulation Model

The analysis maps onto the three-phase circulation model developed in prior pieces in this research line (Becker, Beacken, and Sabra 2026): the song, held constant across the corpus; host framing, varied across the ten videos from adversarial (Morgan, CBN News) to sympathetic-amplifier (Rogan); and the audience response in the comment sections. Compared to prior speaker-driven studies that examined coded antisemitic discourse, the Kanye West design varies host framing while holding the discursive object constant in a form built around explicit Nazi references.

Key Terms

Five terms carry the analytical weight in this article. Each is defined here and used consistently below.

• Discursive object. The content the audience is responding to. In this corpus, the discursive object is the song “Heil Hitler” — held constant across the ten threads (except KW-01, where the song had not yet been released and West’s general antisemitic record served as the discursive object instead).

• Host framing. How the host on camera positions the discursive object — condemning it, defending it, playing it, quoting it, contextualizing it, or treating it as evidence of a broader claim. Host framing is a variable, coded per video.

• Host condition. The typological category into which a host framing falls — adversarial-secular, right-host, progressive-host, reaction-creator, sympathetic-amplifier, hybrid, meta-progressive, or the Adversarial-religious sub-type induced from KW-05. Host framing is what a specific host does; host condition is the class to which their framing belongs.

• Sympathetic amplification. A specific type of host framing in which the host, on camera, treats the discursive object’s claim as plausible, treats institutional response to it as confirmation of that claim, endorses a substantive premise of the object, or praises its content. The KW-07 (Joe Rogan) thread is the corpus’s single sympathetic-amplifier instance.

• Permission structure. The set of discursive moves that commenters in a given thread treat as ordinary, uncontroversial, and not requiring defense against pushback. Permission structures are inferred from what circulates without friction, what draws counter-articulation, and what is marked as transgressive within the comments themselves. They are an analytical abstraction, not an independently observable variable.

Two adjacent terms recur in the article and are used in more specific senses. Response register refers to a discursive mode audiences activate — Affirmation, Taboo, Conspiracy, Relativization, and so on — rather than to individual codings. Discursive scaffolding refers specifically to how one register (e.g., Taboo) creates the terrain on which another register (e.g., Conspiracy elaboration) operates.

Empirical Findings

Results

Quantitative Overview

The corpus comprises 1,000 audience comments coded across ten YouTube threads (KW-01 through KW-10) covering eight distinct outlets across the first ten days after the song’s 8 May 2025 release. All ten threads were independently coded by three annotators (Becker, Deutsch, Verser) and adjudicated to team consensus; per-thread antisemitism rates, counter-speech counts, and dominant conceptual registers are summarized below.

Figure 1. Antisemitism rate across the ten threads. Bars are colored by host-condition typology; the dashed line marks the corpus mean (26%).

Three patterns visible at the cross-thread level shape the qualitative findings that follow.

First, the host-condition gradient that a simple version of the audience-permission-structures argument would predict — adversarial low, sympathetic high — does not appear cleanly in the rates: the right-host thread KW-02 tops the corpus at 46%, above the sympathetic-amplifier thread KW-07 at 32%, progressive threads sit at 24% (KW-03) and 31% (KW-09), the adversarial-secular pair KW-01 (12%) and KW-04 (26%) span the corpus’s floor-to-middle range, and the hybrid and meta-progressive threads (KW-08, KW-10) sit at 17%. Host condition is associated with which registers commenters activate, but it does not map linearly onto total antisemitism density. In this corpus, host condition mainly affects the mode of audience response — what kinds of moves surface — rather than the overall volume of antisemitic material. Notably, the corpus minimum (KW-01, 12%) is the one thread in which the song is not yet the discursive object — a point developed under Finding 1.

Second, the song-release variable produces the clearest within-corpus signal. The KW-01 (pre-release, 12%) versus KW-04 (post-release, 26%) comparison — same host, same adversarial framing, same channel environment, a 14-point shift — shows the strongest within-host association between introduction of the song and a change in audience discourse. This is not a controlled isolation: the discursive object shifts (West’s antisemitism in general vs the “Heil Hitler” song), timing differs, and the broader public discourse moves across the seven days between threads. The 14-point shift is nonetheless the largest within-host change in the corpus and coincides, under constant adversarial framing, with Affirmation content that the pre-release thread does not surface.

Third, counter-speech is unevenly distributed across host conditions. KW-07 and KW-09 each contain 6 of the corpus’s 32 counter-speech codings — but the two sixes differ qualitatively. KW-07’s counter-speech engages the antisemitism register substantively (naming moderation logic, rebutting conspiracy elaboration), giving the sympathetic-amplifier condition a 32% antisemitism rate alongside the corpus’s densest engaged counter-speech. KW-09’s counter-speech consists predominantly of explicit pro-ban stances toward the song, coded as counter-speech under the standing rule that an explicit pro-ban position on openly Nazi-referential content constitutes rejection of it (see Standing Coding Rules appendix). By contrast, the reaction-creator thread (KW-06) contains no counter-speech at all — the corpus’s only such thread — and the adversarial-religious thread (KW-05) contains a single instance. Where counter-speech does appear, it tends to coincide either with a substantive antisemitism register the audience can engage directly (as in KW-07’s conspiracy elaborations) or with explicit antisemitism-defending clusters that invite rebuttal (for example, KW-02’s Nazi-endorsement comments and the humor-as-affirmation and Blame-for-Antisemitism clusters in KW-08, KW-03, and KW-09).

Concept Distributions

The cross-thread concept-frequency matrix shows which of the Lexicon’s concept codes appear in which threads. The compressed view below covers the twelve codes with the highest corpus-wide frequencies; the full matrix, including rare-incidence codes, appears in the supplementary appendix at the end of this piece for readers who want the complete distribution. Each code is named in plain language alongside its Lexicon identifier so the matrix reads independently of the technical apparatus.

Cross-thread concept-frequency matrix (top 12 codes by corpus total)

The matrix surfaces three patterns: (1) Affirmation (C4-3) is the corpus’s modal concept code with 141 codings, and operates at substantial density across every thread in the corpus — with the highest concentrations in KW-02 (27) and KW-06 (20), and the lowest in KW-01 (3, the pre-release thread where the song is not yet the discursive object); (2) Relativization (C3-4, 47 codings — the corpus’s clear second register, reflecting the widened Relativization standard and the anti-ban coding rule documented in the appendix), Taboo (C3-8), Conspiracy (C2-3), and Power (C2-2) form a secondary cluster that operates at moderate density across host conditions, with Religious Antisemitism (C1-5) concentrating in KW-05; (3) the Israel-related family (C5-1) concentrates in KW-09 and is thin elsewhere — the Israel-related register is strongly host-condition-dependent in this corpus.

*KW-04 records 0 Power (C2-2) and 2 Conspiracy (C2-3) codes against 6 Insults (C6-1) — the highest Insults count in the corpus — plus the corpus’s only 2 Threats (C6-2) codings — consistent with a shift in host-attack registers between KW-01 and KW-04. In KW-01 (pre-release), hostility toward Piers Morgan runs through puppet-family constructions that code as Conspiracy or Power; in KW-04 (post-release), structured ideological host-attack thins and hostility operates instead through a slur-and-threat register (documented in the KW-04 per-outlet summary and in §4 below), while the Affirmation slot (14 C4-3) carries the structured ideological engagement. The KW-01 → KW-04 shift is thus not simply additive (host-attack plus Affirmation) but a redistribution of audience hostility across registers once the song becomes the primary discursive object.

Audience Permission Structures: A Working Definition

The three findings below use audience permission structures in a specific, discourse-analytic sense: the set of discursive moves that commenters in a given thread treat as ordinary and uncontroversial — moves that are not actively defended, hedged against pushback, or marked as transgressive in the comments themselves. This is a description of the discursive environment, not of individual cognition: permission structures are an analytical abstraction inferred from recurring interactional patterns, based on which moves surface, which draw counter-articulation, and which circulate without friction. When we speak of host conditions “being associated with” or “shaping” a permission structure, the claim is about the comment environment audiences encounter, not about their internal sense of license; references to moves “being available” or “surfacing” likewise refer to what the thread actually contains.

One epistemological clarification is worth stating explicitly. Audience permission structures is an analytical abstraction inferred from recurring interactional regularities in the comment corpus — patterns of what circulates without friction, what draws counter-articulation, what is marked as transgressive — rather than an independently observable variable. The risk of circularity — the construct being inferred from comments and then used to explain those same comments — is mitigated by the cross-thread comparison structure: a permission-structure claim predicts which moves should and should not circulate under a given host condition, and that prediction can be checked against the corpus. The Morgan within-host comparison (KW-01 vs KW-04) is one instance of that structure.

The inference rule is worth stating in one place. A discursive move is coded as available under a given permission structure when it satisfies at least three of the following five markers: (i) it circulates at meaningful frequency in the thread (not a single isolated comment); (ii) it appears in interactional uptake — subsequent commenters treat it as ordinary and build on it rather than mark it as transgressive; (iii) it draws proportionally little or no counter-articulation from other commenters; (iv) it is not prefaced with justification, hedging, or in-comment defense against anticipated pushback; and (v) it operates alongside compatible moves in the same thread as an organized register rather than as an isolated occurrence. Moves that fail multiple markers — surfacing but drawing counter-speech, or appearing only under hedge structures — are coded as present but not available. This distinction matters because permission structures are not read from frequency alone: a register with three occurrences that circulates without friction is more diagnostic of what the thread makes available than a register with ten occurrences all met by counter-articulation.

Two patterns organize the three findings below. The first concerns volume and effect: sympathetic amplification does not simply increase antisemitic volume but alters the discursive object commenters engage and the operations they perform on it. The second concerns the register the song’s explicit Nazi reference makes available: Affirmation of Antisemitism appears here at densities the research line’s prior speaker-driven studies did not document — a within-case pattern in which the song’s explicit Nazi content, more than sympathetic amplification alone, appears to account for the surface presence of this register. The corpus’s cleanest within-host test of these patterns is the two-thread Morgan comparison (KW-01 pre-release vs. KW-04 post-release), developed under Finding 1 below, where the same host under the same adversarial framing engages the audience under two different discursive objects.

The Analytical Role of Each Thread

Before turning to the findings, it is worth stating what each of the ten threads contributes to the analysis. The corpus is not ten instances of the same observation. It is a structured comparison in which each thread performs distinct analytical work, with the ten together supplying the variation the three findings rest on.

The Morgan pair (KW-01, KW-04) is the corpus’s controlled within-host comparison. Same host, same adversarial framing, same channel environment, seven days apart, with only the discursive object varying (Kanye West’s antisemitism in general in KW-01; the “Heil Hitler” song specifically in KW-04). The pair isolates the effect of the song’s discursive presence from the effect of host condition, and supplies the strongest within-corpus association between the song’s introduction and a change in audience discourse (Finding 1).

The right-host, progressive-host, and adversarial-religious threads (KW-02, KW-03, KW-05, KW-09) vary host condition across the corpus’s central ideological range while holding the song constant as discursive object. These threads test whether host condition shapes the mode of audience response — which registers surface as available — independently of overall antisemitism density. KW-02 (Michael Knowles, right-host) produces the corpus’s highest density (46%) but through Affirmation-dominant uptake; KW-09 (Young Turks, progressive-host) produces the corpus’s third-highest density (31%) with distinctive Israel-comparison and Blame-for-Antisemitism registers absent elsewhere, alongside a Relativization cluster that is the corpus’s densest. KW-05 (CBN News, adversarial-religious) surfaces the Religious Antisemitism register at concentrations no other thread produces.

The sympathetic-amplifier thread (KW-07, Joe Rogan) is the corpus’s single instance of the host condition in which the host, on camera, articulates a framework in which the song’s platform removal confirms the conspiracy the song articulates. It supplies the primary case for Finding 2 (conspiracy elaboration under sympathetic-amplifier framing) and contains the corpus’s densest engaged counter-speech (6 I0c, tied in count with KW-09 but distinct in kind — see the third pattern above) alongside its second-highest antisemitism density — the empirical basis for the refinement that sympathetic amplification appears to make space for audience engagement with the antisemitism register in both directions.

The reaction-creator thread (KW-06) and the hybrid thread (KW-08, Brad Polumbo) test what happens under host conditions that do not fit the adversarial / sympathetic-amplifier axis directly. KW-06 supplies a reference case in which host reaction operates without extended commentary; KW-08 supplies the corpus’s clearest case of the humor-as-affirmation register, where entertainment framing operates as indirect endorsement via trivialization.

The meta-progressive thread (KW-10, Sam Seder) occupies a distinctive position: the host critiques Rogan’s reception of the song rather than the song directly, placing the audience response at one remove from the discursive object. The thread supplies a co-dominant Affirmation/Relativization profile unique in the corpus (8 C4-3 tying 8 C3-4) and its clearest illustration of how meta-discursive framing produces argument about antisemitism rather than uniform articulation of it.

Together the ten threads perform three analytical operations. The Morgan pair isolates the discursive-object variable within-host. The four ideological-variation threads plus the sympathetic-amplifier thread isolate the host-condition variable across the corpus’s central range. The reaction-creator, hybrid, and meta-progressive threads supply reference cases against which the central-range findings can be calibrated. Each thread contributes distinct analytical material; the variation across the ten is the corpus’s design, not its noise.

Finding 1 — Host recoding under adversarial framing

The Piers Morgan threads (KW-01, KW-04) furnish the corpus’s primary case of audience response under sustained adversarial host framing. Morgan condemns the song on camera, identifies its content as antisemitic, and challenges West (and West’s sympathetic interlocutor Sneako) directly. The audience response is consistent with a structural recoding of Morgan himself as Mossad-controlled, with the conspiracy claim articulated as defense of the song against the host. Comments treat Morgan as “Mossad asset,” “puppet for Israel,” “koshervative,” and “Jew worshipper” — formulations directed at the host rather than at the song. Where comments do engage the song’s content, they typically do so to argue that Morgan’s hostility to it confirms his subordination to the system the song names.

The structural significance: adversarial host framing in this corpus does not function as suppression of audience uptake. It redirects audience attention from the song to the host, while the conspiracy framework the song articulates is preserved intact and reactivated as defense against the host’s adversarial position. The pattern is consistent with a framing-dependence mechanism noted in prior work in this research line: adversarial framing functions as suppression only when the audience grants the framer legitimacy in the first place. Where that legitimacy is denied — as it is in this corpus for Morgan — adversarial framing provides additional evidence absorbed into the conspiracy framework rather than functioning as challenge to it.

The two Morgan threads in the corpus permit a within-host comparison that bears on the audience-permission-structures argument. KW-01 (6–7 May 2025) is the pre-release Sneako walkout interview; the song had not yet been released, and the discussion concerns Kanye West’s general antisemitism rather than the song specifically. KW-04 (13 May 2025) is the post-release panel discussion in which the song itself is the subject of the segment. The host is the same, the host’s adversarial framing is the same, the political positioning is the same, and the channel environment is the same (both videos are Piers Morgan Uncensored uploads). The audience composition across the two threads cannot be verified from comment data alone — channel-overlap inference is plausible but not directly demonstrated. The discourse profiles diverge meaningfully. KW-01 is dominated by host-attack content, with commenters treating Morgan as proxy for the system the broader Kanye discourse names; Affirmation of the song’s Nazi reference is thin (3 C4-3 instances) because the song is not the discursive object. KW-04 retains the host-attack register but adds substantial Affirmation content — comments engaging the song directly (“Its a banger though,” “the greatest statement of freedom of speech ever,” “The noticing continues. Thank you, Kanye!”) coexist with the Morgan-as-Zionist-asset construction (14 C4-3 instances). The observable discourse pattern under the same adversarial host condition shifts from host-attack-dominant to host-attack-plus-Affirmation when the song enters the discourse environment. The song’s discursive presence is associated with what commenters articulate under the host’s framing, even when that framing remains constant. This is consistent with the broader argument that audience permission structures depend on the interaction between host condition and discursive object, not on host condition alone.

Figure 2. Effect of the song’s release — KW-01 vs KW-04. Same host (Morgan), same adversarial framing, before vs after the song’s release. Counts here are at the coding level. Affirmation codings increase from 3 (KW-01) to 14 (KW-04) — an eleven-coding rise once the song is the discursive object. Host-attack content persists across both threads but redistributes across registers: KW-01’s structured puppet-family constructions (5 Conspiracy/Power codings) give way in KW-04 to a slur-and-threat register (6 Insults + 2 Threats codings against 2 Conspiracy and 0 Power). Other categories of antisemitic coding and counter-speech remain comparable. The eleven-coding Affirmation shift is the strongest within-host association between the song’s introduction into the discussion and a change in audience discourse the corpus contains; it is not a controlled isolation of the song variable alone, since topic, week, guests, media coverage, discursive object, and comment population all shift between the two threads (see discussion in the second pattern, above, and in Limitations).

Finding 2 — Conspiracy elaboration under sympathetic-amplifier framing

The Joe Rogan thread (KW-07) furnishes the corpus’s primary case of audience response under sympathetic-amplifier framing. Rogan describes the song as “kind of catchy,” articulates the “you can’t criticize them” framework on camera, and treats the platform removal of the song as confirmation of the conspiracy the song articulates. Where the Morgan threads produced host-attack uptake, the Rogan thread produces a register in which the host drops out of audience commentary almost entirely. Audience comments elaborate the conspiracy framework directly, building out the media-control, banking-control, and “they run everything” registers Rogan introduces. The Taboo / “you can’t criticize them” formulation operates in this thread as discursive scaffolding for the audience’s conspiracy elaboration — consistent with the pattern of host articulation and audience elaboration moving in the same direction, without the redirection-to-host-attack visible in the adversarial threads. The pattern is consistent with prior observations of the JRE audience under different speaker conditions: a parallel discourse-analytic study of the Rogan / Murray / Smith episode of April 2025 documented that comment sections under Rogan’s framing converge on dismissal of credentialed counter-positioning and elaborate the conspiracy register rather than challenging it, with antisemitic content present in roughly a third of the analyzed sample (Becker and Sabra 2025).

A refinement worth flagging from the empirical pattern in KW-07. The scaffolding the host’s framing provides does not produce uniform audience uptake of the kind documented under right-host or progressive-host conditions, where commenters converge on a single discursive register (Affirmation in those cases). Instead, KW-07 commenters produce both substantial conspiracy-elaboration of the host’s framing and substantial direct counter-articulation challenging it.

Under Comment 19 alone — the audience comment treating Kanye as proving who can’t be criticized — the reply chain contains roughly four substantive antisemitism articulations of the media-control / Israel-criticism / Jewish-ownership-of-media register alongside multiple direct counter-articulations operating in two modes. Some engage the antisemitic content itself: one identifying the antisemitism as echoed genocide, one quoting the Heil Hitler refrain to challenge the framing that it constitutes criticism, one naming Kanye as actively Nazi-aligned, one rejecting the conflation of Hitler-praise with Israel-criticism, one quoting the antisemitic chorus to dismantle the Taboo framing. Others engage the antisemitism-protective argument structures Rogan advances: countering the Taboo claim via empirical counter-example, countering the hate-speech-as-free-speech framework that protects the song’s circulation, and countering the Taboo claim’s logical form via reductio.

Counter-speech in KW-07 ties with KW-09 for the corpus’s highest count (6 each), and constitutes the corpus’s densest substantively engaged counter-speech — KW-09’s count is carried largely by explicit pro-ban stances coded I0c under the standing rule, whereas KW-07’s engages the conspiracy elaboration directly. This coexists with the corpus’s second-highest conspiracy-elaboration density. The sympathetic-amplifier host condition appears to make space for audience engagement with the antisemitism register in both directions — elaboration and challenge — rather than producing the Affirmation-monoculture KW-03 exhibits or the host-attack redirection KW-01 and KW-04 exhibit. The scaffolding claim holds in the sense that the host’s articulation of “you can’t criticize them” produces an audience discourse that engages the substantive antisemitism register directly. Whether that engagement produces elaboration or counter-articulation appears to depend on the individual audience member’s prior position rather than on the host’s framing alone.

The Rogan thread supports the more precise formulation of the central finding: sympathetic amplification appears to reshape what the audience does with the song (elaborate it, extend it, challenge it, integrate it with adjacent conspiracy material) rather than how much of it the audience produces. The host-condition pattern visible in Figure 1 is consistent with the volume-versus-mode distinction: KW-07 (32%) and KW-02 (46%) both surface high antisemitism densities under structurally different host conditions — right-host with no audible host conspiracy framing versus sympathetic-amplifier with such framing on-camera. The two threads differ substantially in what the audience does with that rate — KW-02 concentrates in Affirmation (C4-3 ×27), KW-07 in Affirmation (C4-3 ×18) combined with Taboo (C3-8 ×8) and Conspiracy (C2-3 ×3). Comparable-high antisemitism density across differing concept distributions is the simplest visible form of the article’s central claim.

Finding 3 — Affirmation of Antisemitism as a corpus-distinctive register

Across the corpus — with the highest density in KW-02 (right-host, 27 instances of C4-3) and KW-06 (reaction-creator, 20 instances), followed by KW-03 and KW-07 (18 each) — commenters produce a register of direct endorsement of the song’s explicit Nazi reference. Formulations such as “hail ye,” “Yedolf turning up the gas,” and “I did Nazi that coming” do not operate through the plausible-deniability scaffolding that characterized comparable content in event-conditioned corpora (e.g., the Temple Israel “elbows to the temple” register) or the coded scaffolding that characterized comparable content in prior speaker-driven studies in this research line. The concept-frequency heatmap (the Supplementary Appendix’s Figure) makes the corpus’s modal pattern visible at a single glance: Affirmation (C4-3) is the only code that operates at moderate-to-high density across every thread in the corpus. The cross-thread comparison (Figure 1) underscores the pattern’s host-condition reach — Affirmation surfaces across right, progressive, adversarial-religious, reaction, sympathetic-amplifier, hybrid, and meta-progressive conditions alike, with the lowest density in the pre-release KW-01 thread (where the song is not yet the discursive object).

The pattern is consistent with the possibility that the song’s explicit Nazi reference reduces the discursive labor of coding endorsement: where the song already names the Nazi reference, commenters do not need to construct plausible-deniability scaffolding to endorse it, and the Affirmation register operates openly. The category architecture of digital antisemitism, on this reading, is not exhausted by the conspiracy-cluster registers that have dominated prior speaker-driven studies; a song built around explicit Nazi references can be associated with an additional register — direct Affirmation of its content — that coded discourse did not surface at comparable density in the cases we have on record. The relationship to host framing is worth flagging: Affirmation appears across the corpus, with the densest instances in KW-02 (right-host) and KW-06 (reaction-creator), and the lowest post-release density in KW-10 (meta-progressive, 8). The corpus contains only one sympathetic-amplifier thread, so claims about that condition’s relationship to Affirmation density rest on a single observation; the pattern that Affirmation is widely distributed across host conditions and concentrated in some of them is more robustly supported.

Summary of Comment Patterns by Outlet

The cross-thread table above gives the headline counts. The per-outlet summaries below describe the distinctive empirical signature of each thread — what register dominated, what sub-register patterns surfaced, where the thread diverges from the corpus norm. Comment IDs identify a comment’s position within a thread: single numbers refer to top-level position (e.g., Comment 4), hyphenated IDs to position within a reply chain (e.g., Comment 17-02). All comments — top-level and reply — are counted equally in the per-thread antisemitism rate.

• KW-01 — Piers Morgan Uncensored (Morgan + Sneako, walk-out): antisemitism rate 12% (100 entries). Predates the song release (6–7 May 2025); addresses Kanye West’s general antisemitic record rather than “Heil Hitler” specifically. Dominant register: host-attack recoding Morgan as Mossad/AIPAC/Zionist proxy — codes primarily as Conspiracy (C2-3) under the puppet-family rule (asset, puppet, stooge, payroll formulations), with Power (C2-2) reserved for bare-influence claims like “Zionist megaphone” and unnamed-controller constructions (“the Powers that be… can make or break you overnight”). Affirmation of Antisemitism appears thinly (3 instances). Netanyahu-comparison jabs at the host without puppet lexicon or a Jewish collective referent (“Piers is netanyahoo light”) code I0 per the final calibration. The corpus’s only thread where host-attack dominates over song-engagement — and, at 12%, the corpus floor.

• KW-02 — Michael Knowles (The Daily Wire): antisemitism rate 46% (100 entries). Earliest verified primary reaction in the corpus (9 May 2025, day after song release). Dominant register: Affirmation of Antisemitism (27 instances of C4-3), including the “NHH” / “N**s Heil Hitler” salute formula and the “I now love Kanye” positive-alignment sub-register. The corpus’s highest antisemitism density.

• KW-03 — Breaking Points (Krystal Ball & Emily Jashinsky): antisemitism rate 24% (100 entries). Affirmation-monolithic profile — of the 25 concept codings in the thread, 18 are C4-3 (72%), almost all aesthetic praise of the song without distancing (A4 sub-register); the remainder concentrate in Relativization (5 C3-4). Host visibility in comments near-absent. Notable as the corpus’s first progressive-host thread; antisemitism density is comparable to right-host and adversarial-religious conditions, undermining the predicted host-condition gradient. Four counter-speech codings.

• KW-04 — Piers Morgan Uncensored Panel: antisemitism rate 26% (100 entries). Same host, framing, and channel environment as KW-01, but post-song-release. The KW-01 → KW-04 shift involves a register redistribution: structured-ideology host-attack thins (puppet-family Conspiracy/Power codings drop), while substantial Affirmation surfaces (14 C4-3 codings — “Its a banger though,” “the greatest statement of freedom of speech ever,” “The noticing continues. Thank you, Kanye!”). Host-attack routes through the slur-and-threat register in the corpus’s densest threat sub-thread, producing the corpus’s highest C6-1 (Insults) count, its only C6-2 (Threats) codings — including the weapon-emoji threat and the death-wish location solicitation — and its first explicit incitement cluster. Also notable: the corpus’s most sophisticated Holocaust-denial construction (Comment 4, the 6 million / 271k numerical substitution).

• KW-05 — CBN News (Raj Nair + Joshua Washington): antisemitism rate 30% (100 entries). Faith-context outlet (CBN = Christian Broadcasting Network); the guest, Joshua Washington, directs the Institute for Black Solidarity with Israel. Distinctive register: religiously-grounded antisemitism through Christian theology, including the Christian-Zionist inversion cluster, supersessionism (both the “synagogue of Satan” variant and the chosen-people-inversion / scriptural-imposture variant), conversionist antijudaism, deicide-inversion, and Christian-historic-antisemitism-as-heritage. The corpus’s densest Religious Antisemitism concentration (6 C1-5 codings) and the most conceptually diverse thread (13 distinct concept codes). One counter-speech coding (Comment 25-4, “Jesus was Jewish though,” rebutting the antijudaic “noticer” frame of its parent).

• KW-06 — Knox Hill: antisemitism rate 25% (100 entries). Reaction-creator host providing critical-aesthetic engagement with the song. Host visibility in comments near-absent — the discourse stays on the song rather than recoding the host. Dominant registers: Affirmation (20 C4-3 codings, primarily aesthetic praise of the song, the corpus’s second-highest Affirmation count) and a substantial Relativization cluster (7 C3-4 codings) elaborating minimization and free-speech defenses around the platform’s failure to remove the song. Zero counter-speech — the corpus’s only thread with none.

• KW-07 — JRE Clips (Joe Rogan): antisemitism rate 32% (100 entries). Second-highest density in the corpus, after KW-02. Distinctive registers: a platform-removal-as-confirmation conspiracy cluster (the “jewtoob” framing — Comments 17-02, 17-25, 19-027); an explicit Affirmation register (18 C4-3 codings, including “NHH”-formula and numerological-decoding constructions); a substantial Taboo cluster (8 C3-8 codings); an Instrumentalization register (Comment 17-32, “oi vey!” as mock-performance of Jewish complaint, I1a + C3-5); and 6 I0c codings — the corpus’s densest substantively engaged counter-speech — operating in two modes: direct engagement with antisemitic content, and engagement with antisemitism-protective argument structures. The sympathetic-amplifier condition appears to make space for engagement in both directions rather than producing uniform Affirmation.

• KW-08 — Brad Polumbo (Brad vs Everyone): antisemitism rate 17% (100 entries; tied with KW-10 as the lowest post-release density in the corpus — the overall corpus floor is the pre-release KW-01 at 12%). Conservative-critical / hybrid condition: Polumbo defends free speech while explicitly naming Kanye’s content as antisemitic. Distinctive registers: the corpus’s largest humor-as-affirmation cluster — anchored by Comment 9 (”I’m Jewish and Kanye’s crash out is just hilarious... beat at the end honestly goes hard if you ignore the blatant antisemitism”) with its Confirmation chain (9-01, 9-02, 9-14) — where entertainment framing operates as indirect Affirmation via trivialization; a thin aesthetic-praise cluster (3 pure C4-3 codings: 1-4, 6-2, 13); a Power-Media + Blame-for-AS cluster (Comment 16 “jew music moguls,” Comment 29-8 “Jewish record executives profit off the plight of impoverished Black Americans” — the latter triple-coded C4-3 + C1-9 + C2-2); and Instrumentalization via a villain-comparison chain (Comments 21-05 / 21-06, juice-box 🧃 letter-substitution). Notable single cases: Comment 18-4 “271k at most” (the cross-thread Holocaust-distortion register, also KW-04 Comment 4) and Comment 18-5 (Neturei Karta weaponization register). Substantial counter-speech (5 I0c codings): Comment 18 offers the corpus’s most substantive counter-speech articulation, and the Comment 9 chain contains the corpus’s rare interactional case where humor-as-affirmation and counter-speech (9-06, 9-08, 9-10) occur in direct dialogue.

• KW-09 — The Young Turks (Cenk Uygur & Jordan Uhl): antisemitism rate 31% (100 entries). Third-densest in the corpus, after KW-02 and KW-07 — and within one point of KW-07. Progressive-host condition, but the audience profile diverges from KW-03’s monolithic A4-aesthetic pattern. Distinctive registers: the corpus’s densest Israel-comparison cluster (6 C5-1 Nazi Analogy codings + 3 C3-5 Instrumentalization); the corpus’s densest Relativization cluster (12 C3-4 codings, carried substantially by anti-ban stances coded under the standing anti-ban rule) running in parallel with direct Affirmation (10 C4-3); the Jewish-music-industry-profits trope (Comments 3-2, 5-2, same user); the corpus’s first clean manufactured-hate-crimes register (Comment 31); the first visual-identity-marker euphemism (Comment 3-5, “the little hats”). The most conceptually diverse progressive-host thread in the corpus. Six counter-speech codings — predominantly explicit pro-ban stances coded I0c under the standing rule.

• KW-10 — The Majority Report (Sam Seder): antisemitism rate 17% (100 entries; tied with KW-08 as the lowest post-release density). Meta-progressive condition — the host critiques Joe Rogan’s defense of the song (KW-07) rather than the song itself, positioning the audience response at one remove. Distinctive registers: a dense Relativization cluster (8 C3-4 codings, tying Affirmation and giving KW-10 a co-dominant Affirmation/Relativization profile unique in the corpus — Kanye-as-edgy-trolling, mental-illness, Israel-comparison, free-speech defense, sports-analogy minimization, Jewish-Nazi-collaborator tropes); the corpus’s most multi-concept antisemitic construction (Comment 7-030, supersessionism + Greed + Power + Evil, with the supersessionist frame carrying Religious Antisemitism); the first active Nazi-revisionist propaganda recommendation (Comment 7-052, “Europa the Final Battle”); the first esoteric antisemitism-conspiracy reference (Comment 7-062, anti-Frankist); the most explicit Blood-Libel-projected-onto-Israel construction (Comment 7-005); and substantial political-positioning argument in the 85-reply mega-thread under Comment 7. Direct Affirmation at 8 C4-3 codings — consistent with meta-discursive framing producing argument about antisemitism rather than uniform articulation of it. Two counter-speech codings, including the mega-anchor Comment 7 itself (“According to right wingers, even if you make a song glorifying Hitler, that still somehow doesn’t make you a Nazi” — I0c, 834 likes, 85 replies), whose counter-speech status frames the entire dominant reply chain.

1. Affirmation of Antisemitism

This is the response register that distinguishes the Kanye West corpus most sharply from the research line’s prior speaker-driven studies. The Affirmation register is treated here as a hierarchy of four sub-registers rather than a single category. The distinction matters because the sub-registers carry different evidential weight: a comment endorsing Hitler directly is analytically distinct from a comment making a pun about Nazism, even though both circulate the Nazi referent. Coders distinguish among:

A1. Direct ideological endorsement. Explicit endorsement of the Nazi reference itself, with the Nazi content as the endorsed object: “Hail Hitler,” “Hitler was right,” “based,” and similar. Coded as Affirmation only when the Nazi referent is the substantive endorsed content, not a generic edginess marker.

A2. Positive alignment with the song’s framing. Endorsement of Kanye West’s framing of his grievance as containing legitimate substance, including the conspiracy claim the song articulates: “He’s calling out your bosses,” “109 countries btw,” “they don’t understand the things I say.” The alignment is with the song’s discursive content rather than with Nazi ideology per se; the two converge in this case because the song’s framing is itself the Nazi referent.

A3. Humorous normalization. Antisemitic content delivered through wordplay, puns, or irony whose humorous frame is built on the antisemitic referent rather than against it: “I did Nazi that coming,” “Yedolf turning up the gas,” “Karaoke night will never be the same.” Coded as Affirmation when the joke’s structure requires the antisemitic content to function — the antisemitic referent is the punch line’s load-bearing element, not its target.

A4. Aesthetic endorsement linked to Nazi content. Praise of the song’s musical or production qualities (“that beat is truly undeniable,” “the song slaps,” “Ye cooked”) delivered in commentary contexts where the song’s title, lyric, and visual content are the explicit Nazi reference. A4 is the most interpretively demanding sub-register, and the coding rule reflects a corpus-specific judgment. For songs in which the antisemitic content lives in subtext rather than in the title and chorus, the strict default would be to hold ambiguous aesthetic praise for second-coder review. For “Heil Hitler” the strict default is loosened, because the song’s title and chorus are the Nazi referent: positive aesthetic engagement with the song is not reliably separable from positive engagement with the Nazi referent it names without an explicit distancing move in this artifact-specific context. The coding rule is therefore: aesthetic praise of the song was treated as A4 unless an explicit distancing move was present in the comment itself — formulations like “I wish the lyrics were changed”, “shame about the message”, “despite the content”, or an irony marker that distances the speaker from the Nazi referent specifically. Distancing must appear within the comment; subsequent self-disavowals in reply chains do not retroactively change a coded A4 instance. Comments that praise the music while making the lyrics the punchline of irony are coded A3 (humorous normalization) rather than A4. The distinction between A4 and A2 (Positive alignment) is whether the comment articulates substantive alignment with the song’s framing: “This song goes so hard” is A4 (aesthetic only); “He’s calling out your bosses, song slaps” is A2 (aesthetic plus framing alignment).

The section’s analytical claim — that a song built around explicit Nazi references activates registers that coded discourse did not activate at comparable density in the cases we have on record — anchors the corpus’s contribution to the coded/explicit distinction. The hierarchy above is also the article’s primary methodological refinement: prior research-line studies treated Affirmation as a single category, but the Kanye West corpus surfaces enough sub-register variation to warrant the four-level decomposition.

The clearest A1 case in the corpus is KW-07 Comment 17-30:

“NHH = 1488 because N is the 14th letter” (KW-07 Comment 17-30)

The construction articulates Direct Ideological Endorsement through numerological decoding: the commenter spells out, for other readers, how the “NHH” abbreviation circulating across multiple threads (N****s Heil Hitler) maps onto the white-supremacist numerological signature 1488 (the Fourteen Words plus “HH” — Heil Hitler — as the eighth letter doubled). The comment does not merely deploy the dog whistle; it teaches it, functioning as in-thread decryption service for the white-supremacist vocabulary the song’s audience is circulating. The comment codes as I1b + C4-3 (Affirmation, A1 sub-register) — contextual rather than co-textual, since the 1488 signifier requires world knowledge to interpret.

Christian-historic-antisemitism-as-heritage surfaces in KW-05 Comment 25-1:

“To be a true Christian is to be ‘antisemitic’. Man up and be like your glorious ancestors.” (KW-05 Comment 25-1)

The construction — posted by an account whose handle carries the 1488 numerological signature — frames antisemitism as constitutive of authentic Christian identity and invokes “glorious ancestors” as the tradition to be inherited. Coded as I1 + C4-3 (Affirmation) + C1-5 (Religious Antisemitism). This is the corpus’s clearest case of historical-precedent-as-justification: historic Christian antisemitism is presented not as a cautionary example but as an ancestral standard the commenter urges others to live up to.

The corpus also surfaces adjacent historical-precedent and relativization registers that interact with Affirmation but are conceptually wider. One such register is expulsion-as-evidence: claims that historical expulsions of Jews from “109 countries” prove the validity of contemporary antisemitism. In prior coded-discourse studies these arguments primarily scaffold conspiracy narratives; in this corpus they more often push toward direct Affirmation of historical anti-Jewish violence, consistent with the explicit-supply pattern noted across the article.

Supersessionist Christian antisemitism concentrates in KW-05 and KW-10 and operates as a theological variant of this historical-precedent register. The most compressed case is KW-05 Comment 5-1 — “Amen — Revelation 2. 9” — which invokes the “synagogue of Satan” verse as assent to a parent comment’s warning that God will “get [Kanye’s] attention,” and codes as I1b + C1-5 (Religious Antisemitism). The blood-curse variant appears in KW-02 Comment 4-21, which cites the Pilate blood-on-their-hands passage and “a certain synagogue of a certain red feller,” coding as I1b + C1-5 + C2-3 (the covert-agent construction carrying the Conspiracy code). Conversionist antijudaism appears in KW-05 Comment 24 — “needa bring the jews to jesus” — coded I1 + C1-5. A structurally distinct KW-05 case is Comment 15-1, which weaponizes the chosen-people frame against contemporary Jews: scripture is invoked to affirm biblical Israel’s special status, then reversed to strip current Jews of that status. The move is imposture — contemporary Jews cast as counterfeits — and codes with the hidden-identity logic structurally parallel to the puppet-family constructions. A more elaborated version appears in KW-10 Comment 7-030, which layers supersessionism, Greed, Power, and Evil across a single block of text (coded I1b + C2-1 + C2-2 + C1-2 + C1-5, with the supersessionist framing carrying the Religious Antisemitism coding). Across these comments, supersessionism functions as the master frame: historical, economic, and identity-based grievances are presented as evidence for a Christian-theological claim about Jews, rather than as standalone complaints.

An adjacent register operates through name-fusion. KW-01 Comment 5 fuses “Satan” and “Netanyahu” into:

“Satanyahu” (KW-01 Comment 5)

attaching the Devil-attribution register directly to the sitting Israeli prime minister. The construction codes as I1b + C1-2 (Evil / Moral Corruption, via the Satan-identification — contextual rather than co-textual, since the pun requires recognizing the Netanyahu reference) + C1-9 (Blame for Antisemitism, via the surrounding argumentative structure) + C5-5 (Genocide, via the wider argumentative context in which the name-fusion is deployed). The construction is analytically distinctive because it compresses a triple attribution — demonization, blame-shifting, and genocide-framing — into a single lexical device. The name-fusion register echoes historically documented antisemitic constructions (e.g., “Zio-nazi,” “Rothschild-vermin”) but adapts them to a contemporary head-of-state target, illustrating how digital vocabulary generates compact vehicles for legacy antisemitic frames.

The corpus also surfaces a Nazi-vocabulary redeployment register in KW-06 Comment 11-11, where the term “untermensch” — the National Socialist designation for dehumanized populations — is turned on the commenter’s interlocutor: “How do you like your state forced vaxx undermench?” The redeployment sits inside a construction that simultaneously minimizes Nazism itself (“Adolf was a sad loser on meth that lost. Its the least dangerous ideology around today, the one that’s actually dangerous is mainstream democrat globalist communism”). The comment codes as I1b + C4-3 (Affirmation, via the minimization-of-Nazism move) + C3-4 (Relativization & Denial of Antisemitism). The register is analytically notable for its double operation: Nazi persecution vocabulary is kept in active circulation as an insult resource, while Nazi ideology is simultaneously reframed as the least dangerous option on the contemporary political menu. A contrasting deployment of the same term appears in KW-04 Comment 4-5, where “untermensch” is used in counter-speech — reminding a Kanye defender that “Hitler would view you also as untermensch” (coded I0c) — demonstrating that Nazi-historical vocabulary circulates in this corpus in both affirming and counter-speech modalities.

A cross-thread Christianized-Heil register surfaces in KW-02 and KW-05, in which the Nazi salute is fused with Christian affirmation formulas:

“Christ is Lord! HH” (KW-02 Comment 1-6) “Hail Christ Jesus 💪.. MAGA 🇺🇸🏁” (KW-05 Comment 20)

The first attaches the “HH” (Heil Hitler) abbreviation directly to a Christological credo; the second fuses the salute formula with a Christian invocation and a contemporary political slogan. Both comments code as I1a + C4-3 (Affirmation) with the Christianized framing carrying an additional Religious Antisemitism (C1-5) coding in the KW-02 case where the “HH” + Christological pairing operates as a substantive theological claim rather than a bare adjacency. The pair is analytically distinctive because it illustrates how the salute formula circulates in the corpus not as a standalone Nazi reference but as a modular element that combines with adjacent identity-marking vocabulary — Christian, political, aesthetic — depending on the discursive environment. Its appearance in a right-host (KW-02) and an adversarial-religious (KW-05) thread suggests that the Christianized-Heil register operates as a cross-condition identifier of a specific fused ideological position rather than as a host-condition-dependent register.

The corpus surfaces a distinctive humor-as-affirmation register whose paradigmatic case is KW-08 Comment 9:

“I’m Jewish and Kanye’s crash out is just hilarious (Edit: I’ve never heard the whole song the beat at the end honestly goes hard if you ignore the blatant antisemitism).” (KW-08 Comment 9)

The comment codes as I1a + C4-3 (Affirmation) + C3-4 (Relativization) — dual-coded because the entertainment-framing and the aside-from-content bracketing operate simultaneously. The commenter’s Jewish self-identification does not neutralize the trivialization move; if anything, it authorizes an “even Jews find it funny” trivialization for other commenters in the reply chain. Confirmation comments in the reply chain endorse the “don’t take it seriously” framing:

“lol love this. People really shouldn’t take everything to heart so deeply...” (KW-08 Comment 9-01) “He is truly comedy material to laugh at!” (KW-08 Comment 9-02) “there is no such thing as bad publicity” (KW-08 Comment 9-14)

Comment 9-03 illustrates the register’s operation even in comments that register concern about the song’s spread:

“It’d be a whole lot funnier if I didn’t look at the comment sections... teenagers being introduced and AGREEING with it” (KW-08 Comment 9-03)

The “funnier” framing performs the trivialization that the concern nominally opposes. The register is analytically important because it operates independently of aesthetic-praise or ideological-endorsement modes: humor as such becomes the vehicle by which antisemitic content is normalized in the commenting environment.

The Comment 9 chain also surfaces the corpus’s clearest interactional case of humor-as-affirmation and counter-speech in direct dialogue. The same reply chain contains three counter-speech comments — 9-06, 9-08, 9-10 — that push back against Comment 9-05’s “he wants you to get mad” trivialization framing, naming the harm concretely (a cited 400% increase in hate crimes; “actual people at the end of his shock trick”). No other thread in the corpus shows this pattern of concentrated humor-as-affirmation and concentrated counter-speech coexisting within a single reply chain. The finding is worth flagging as an empirical instance of the audience-permission-structures argument: the same discursive space, under the same host framing, can host both the trivialization register and the counter-speech register in dialogue.

The most analytically dense Blame-for-Antisemitism case in the corpus is KW-08 Comment 29-8:

“He’s already said in an interview he’s doing it because he feels Jewish record executives profit off the plight of impoverished Black Americans. Also because he feels they’re fucking with him... few Jewish individuals in the industry that openly threatened him with black mail... So I don’t really blame him for being mad.” (KW-08 Comment 29-8)

The comment codes as I1a + C4-3 (Affirmation) + C1-9 (Blame for Antisemitism) + C2-2 (Power) — triple-coded because it simultaneously (a) endorses Kanye’s antisemitic content by justifying it as reactive, (b) attributes responsibility for that content to Jewish media control, and (c) invokes the Power register through the “record executives” framing. The case is important because it articulates the “they made him do it” logic as a coherent argument rather than as a slogan — the commenter constructs a full justification narrative that ends with an explicit exculpation (”I don’t really blame him for being mad”). The construction is a compressed instance of the broader Blame-for-Antisemitism register that surfaces across multiple threads in the corpus.

2. Taboo and Self-Victimization

Taboo and self-victimization is the register that operates as the discursive scaffolding through which the sympathetic-amplifier effect functions. Four sub-registers organize the corpus’s Taboo content: the “you can’t criticize them” framework as Rogan articulates it on the podcast and the audience picks it up; the cancellation-as-evidence inversion (platform removal reframed as the cause rather than the effect of the song’s circulation); the persecution-of-the-truth-teller framing applied to Kanye West himself; and the personal-experience-of-moderation sub-register, in which commenters cite their own account suspensions as confirmation of the conspiracy framework. Commenters frequently interpret platform-moderation responses as confirmation of that framework when the responses arrive without an accompanying account of why the content meets the moderation threshold — an observation the Conclusion develops further.

The Taboo register in this corpus is structurally inseparable from the Conspiracy register documented under §3 — the corpus’s clearest Taboo cases pull double duty as Conspiracy cases:

“U really think jewtoob would let him type that?” (KW-07 Comment 17-02) “the song has been removed from every mainstream platform... they often get accused of controlling media. Logic dummy.” (KW-07 Comment 19-027)

The first articulates Taboo through the “letting” construction (someone is permitted or not permitted to speak) and Conspiracy through the “jewtoob” identification of who is doing the permitting (I1b + C2-3 + C3-8). The second articulates Taboo through the platform-removal anchor and Conspiracy through the elaboration that follows. The dual-coding is the analytical point: in this corpus, the Taboo register predominantly serves as the entry-point for Conspiracy elaboration rather than as a stand-alone discursive move.

The cancellation-as-evidence inversion appears in compressed form as a recurring formula across multiple threads:

“the more they ban it the more people listen” “this only makes it bigger”

and the recurring inversion in which platform-moderation actions are reframed as marketing rather than moderation. The construction codes as I1b + C3-8 (Taboo) where the implicit Jewish-control identification is the discursive context, and as I1a + C3-8 where the construction names “they” directly. The structural feature worth noting: counter-speech directed at this register rarely redirects the discussion away from the Taboo framing in the corpus. Among the corpus’s counter-speech (I0c) codings, the ones that engage the cancellation-as-evidence formula directly (KW-08 Comment 18, KW-07 Comments 14, 19-005) are themselves treated by subsequent responders as further evidence of the Taboo claim — the counter-speech move is recoded by the comment ecosystem as part of the silencing it tries to challenge.

The highest-engagement Taboo anchor in the corpus is KW-07 Comment 19 — “He’s proving a point. He’s showing who you can’t criticize. That’s who’s in charge.” — which received approximately 5,024 likes and generated a multi-comment counter-speech (I0c) reply cluster. The comment codes as I1a + C3-8 (Taboo of Criticism) and articulates the who-you-can’t-criticize framing in compact form. The interactional structure around it is analytically distinctive: rather than the more common pattern in which counter-speech is met with reaffirmation of the Taboo claim, the reply chain under Comment 19 sustains consecutive counter-speech comments that name the moderation logic directly and reject the removal-as-evidence inference. The cluster is analytically important for two reasons. First, it demonstrates that concentrated counter-speech engagement is empirically possible against a high-engagement Taboo anchor under sympathetic-amplifier conditions — the register is not immune to sustained pushback. Second, it illustrates the specific discursive circumstance under which pushback holds: where the counter-speech names the moderation rationale rather than only denying the removal-as-evidence claim, the reply chain sustains rather than collapses. The counter-speech that succeeds is register-naming rather than assertion-denying — consistent with the pattern documented earlier in the “noticing”-pair analysis.

3. Power and Conspiracy Narratives

The initial expectation from prior pieces in this research line was that Power and Conspiracy together would form the dominant antisemitic register in the corpus. The matrix surfaces a different pattern: Affirmation (C4-3, 141 codings) is modal; Power (C2-2, 20) and Conspiracy (C2-3, 19) operate as a secondary cluster.

Power and Conspiracy nonetheless remain the register through which the Morgan–Rogan contrast operates, and are presented first here because the audience-facing logic that connects them — “they run things” and “they orchestrated this specific event” — sits on the same conceptual terrain.

This section covers two operationally distinct but conceptually adjacent codes in the Decoding Antisemitism Lexicon: Power and Subordination claims (C2-2 — background influence and control as stated fact) and Conspiracy Theories (C2-3 — hidden coordinated action behind specific events). The Lexicon treats them as separate codes; this article presents them together because the audience-facing logic links them.

Sub-registers under Power observed in the corpus: dog-whistle and orthographic substitution; the “Zionist megaphone” / media-influence register (most densely in the Joe Rogan thread, where the song’s removal from streaming platforms is treated as evidence of Jewish media influence — note that control claims here, where they invoke string-pulling logic, sit under Conspiracy below); generalized claims of Jewish influence over institutions without specific hidden-coordinator agency.

Sub-registers under Conspiracy observed in the corpus: host-attack formulations that frame the host as a puppet/asset/agent of a hidden controlling body (the puppet metaphor, with its inherent string-puller logic, codes as Conspiracy — most densely in the Piers Morgan threads, where Morgan is recoded as “Mossad asset,” “AIPAC puppet,” and “Zionist agent”); the USA-as-puppet-state framing (same puppet logic at the state level) and the AIPAC-capture / Jewish-state-capture register (capture implies covert seizure of institutions by hidden coordinated agency); historical-event conspiracy framings (JFK, Lavon Affair, USS Liberty, dancing Israelis); and deep-state and institutional-conspiracy elaborations. The “AshkeNazi Zionist pawn” Hitler-rehabilitation revisionism is treated separately under §1 (Affirmation), since its central move is Hitler-rehabilitation rather than conspiracy proper.

The relationship between the two registers in this corpus: the Power register supplies the meta-frame (Jews/Israel/Zionists as influential and controlling) under which specific Conspiracy claims about individual events or actors are organized — per the hierarchical multi-target organization principle established in prior work in this research line.

The clearest sympathetic-amplifier elaboration in the corpus appears in KW-07 (Rogan), where commenters articulate the Conspiracy register through the song’s removal from streaming platforms. The anchor, Comment 19 (“He’s proving a point. He’s showing who you can’t criticize. That’s who’s in charge.”), frames the removal as demonstration; the elaboration arrives in the reply chain:

“use logic dumbass. The song has been removed from every mainstream platform. Who did Hitler hate the most? They often get accused of controlling media. Logic dummy.” (KW-07 Comment 19-027)

The construction is a chain — platform removal → identification of who removed it → confirmation that they control media — in which each step is presented as inferential follow-through rather than ideological articulation. The anchor codes as I1a + C3-8 (Taboo of Criticism); the reply’s elaboration carries the Conspiracy content. The Comment 19 chain represents the cleanest case of the Taboo-to-Conspiracy bridge the article’s broader argument identifies: platform moderation becomes confirmation of the conspiracy frame precisely because the moderation arrives without an accompanying account of why the content meets the threshold.

Adjacent in KW-07 is the “jewtoob” construction:

“U really think jewtoob would let him type that?” (KW-07 Comment 17-02) “Jew toob? 😂” (KW-07 Comment 17-25)

The construction names YouTube through a Jewish-portmanteau that articulates platform ownership as a fact-claim about Jewish influence. Comment 17-02 codes as I1b + C2-3 (Conspiracy Theories) + C3-8 (Taboo) — the “would let him” construction invokes the hidden-permitter logic; Comment 17-25, appearing later in the same thread, codes as I1a + C2-2 (Power) — the bare portmanteau restates the ownership claim without the string-puller machinery. The same word does different work depending on where it sits in the elaboration sequence: the pairing surfaces how a single coined term carries either the full Conspiracy construction or its residual Power claim depending on the surrounding syntax.

The corpus’s most efficient global-Conspiracy formulation is KW-05 Comment 25-2:

“He’s just acutely aware of international jewery...” (KW-05 Comment 25-2)

The construction codes as I1 + C4-3 (Affirmation, endorsing Kanye’s “awareness”) + C2-3 (Conspiracy, “international jewery” invokes the trans-national-Jewish-coordination architecture). One twelve-word comment performs both endorsement of the song’s framing and full-scale global-conspiracy elaboration.

A note on register visibility. The “noticing” construction — coded euphemism for identifying (alleged) Jewish influence, widely circulating in alt-right and reactionary discourse — appears in the corpus in both endorsement and counter-speech modalities. Its endorsement form surfaces in KW-04:

“The noticing continues. Thank you, Kanye!” (KW-04 Comment 15)

where the register is affirmed positively as identification with Kanye’s discourse. Its counter-speech form surfaces in KW-03 Comment 33-10:

“More like ‘I can notice patterns and will ensure there is a world left for my grandchildren’” (KW-03 Comment 33-10)

where the commenter mockingly ventriloquizes both the “noticing” frame and the Great Replacement / demographic-threat frame (”a world left for my grandchildren”) as a substitute-quotation offered in response to a parent comment characterizing the song as alt-right pipeline content. The KW-03 case codes as I0c (counter-speech); the KW-04 case codes as I1a + C4-3 (Affirmation) + C2-3 (Conspiracy Theories) — co-textual, since the thread supplies the “noticing” anchor, and dual-coded because the formulation simultaneously thanks Kanye for the identification and invokes the coded conspiracy frame. The pair is analytically distinctive because it demonstrates that the same coded formation circulates across host conditions in different discursive positions — as sincere affirmation in the adversarial-secular thread (KW-04, where Morgan condemns the song and the audience defends it) and as ventriloquized target in the progressive-host thread (KW-03, where the host condemns the song and the audience-side counter-speech mocks the coded scaffolding). Some commenters, in other words, recognize the pipeline’s coded markers well enough to articulate them from the outside as objects of critique — a form of register-naming counter-speech distinct from direct content-challenge. That the same construction shifts sign under different host conditions, without changing its lexical form, is consistent with the broader argument that audience permission structures affect not just what content commenters produce but which stance they take toward circulating antisemitic vocabulary they can be assumed to recognize. A related instance worth noting for the cross-thread reach of the register: KW-05 Comment 25-3 (“in fact, you could say jesus himself was a bit of a ‘noticer’ too 🤔”) extends the coded term into the theological frame, coded I1b + C4-3 + C1-5 (Religious Antisemitism).

The corpus also surfaces a distinctive manufactured-hate-crimes construction in KW-09 Comment 31:

“More manufactured Jewish hate crimes to pass more Antisemitic laws.” (KW-09 Comment 31)

The structure inverts the standard moderation narrative — antisemitic incidents are reframed as Jewish self-staging to legislate against critics — and codes as I1a + C2-3 (Conspiracy Theories) + C3-5 (Instrumentalization of Antisemitism). This register matters because it pre-empts one likely counter-speech move — the move that the existence of antisemitism could be used to challenge the conspiracy framing — by claiming the incidents themselves are coordinated.

4. Threats and Calls for Retaliation

Eliminationist content directed at Jewish targets, individual and collective — including Holocaust-completion language, target-individual threats with direct combat-game-register address, and the corpus-distinctive dynamic flagged in the team’s adjudication in which commenters call each other “Jew” as a pejorative when one disagrees with another’s pro-West stance, performing Jewish identity as the discursive marker of opposition to the song and completing the inversion through which the comment ecosystem frames defense of the song as defense of truth.

The corpus’s densest threat cluster is in KW-04, in a sub-thread where the reply-target’s channel handle itself signals a kill-the-target framing. The cluster contains an escalation pattern across six replies:

“Jew” — deployed as slur to a Jewish-coded interlocutor (KW-04 Comment 9-5) “Jew” / “jewy” — slur escalation (KW-04 Comments 10-02 / 10-04) “OK rabbi” — substitution-slur (KW-04 Comment 10-06) “me too Jew. Me too. 🔫“ — slur + gun emoji (KW-04 Comment 10-08) “Jew send location” — meme-formula threat (KW-04 Comment 10-09)

The cluster shows the in-thread acceleration from slur-as-insult to slur-paired-with-violence-emoji to explicit hunt-for-target threat. The progression matters because each step is locally legible — the slur is the slur, the gun emoji is the gun emoji, the “send location” phrase is the “send location” phrase — but the cumulative thread carries the escalation that no single comment articulates. The cluster codes through C6-1 (Insults) for the slur-deployment and C6-2 (Threats) for the violence content: Comment 10-08 codes as I1a + C6-1 + C6-2 + MMD (the first-person adoption of “can’t wait to meet one of you,” addressed to a Jewish-coded target and fused with the weapon emoji, constitutes an implied threat rather than abstract violence endorsement), and Comment 10-09 as I1a + C6-1 + C6-2 (Threats, death wish — the location solicitation downstream of the weapon-emoji threat). KW-04’s documented absence from the Power (C2-2) and Conspiracy (C2-3) registers suggests the thread’s host-attack continuation operates through the slur-and-threat register rather than the structured-ideology registers KW-01 supplied.

The Jew-as-pejorative-marker dynamic surfaces also in KW-07 (Rogan), where commenters use “Jew” to mark other commenters’ disagreement with the pro-West / pro-Rogan position — performing Jewish identity as the discursive position of opposition to the song. The construction is structurally distinct from the KW-04 cluster because it does not require its target to be Jewish-coded; it simply marks the target’s discursive opposition through Jewish-coding. Coded as I1a + C6-1, plus the inversion-completion the register performs: defense of the song is reframed as defense of truth against a Jewish-coded opposition that the comment ecosystem has just generated through the labeling act itself.

5. Israel-Related Antisemitism

The register operating principally as deflection — moving the discursive object from the song’s explicit Nazi content to Israel/Palestine framings. Under the progressive-host condition in particular, Israel-related framings become one of the principal discursive routes through which the explicit Nazi referent is displaced: the permission structure makes Israel-criticism content available as a frame the audience can pick up, and the song’s Nazi content recedes from the discursive object. The corpus surfaces Gaza-genocide deflection formulations alongside moves in which commenters acknowledge the antisemitism category but reorder its priority relative to other framings.

The corpus’s most consequential Israel-related construction is KW-10 Comment 7-005, which projects Blood Libel — the medieval Christian charge that Jews ritually murder children for religious purposes — onto Israeli military forces. Coded as I1b + C1-3 (Blood Libel) + C3-5 (Instrumentalization of Antisemitism & the Holocaust). The interest is the route of transmission: Blood Libel is a Christian-historical antisemitic register that, in this corpus, surfaces only in Israel-comparison contexts, suggesting that the Israel/Palestine framing supplies a contemporary scaffold for circulating historic antisemitic content that would otherwise be too explicitly identifiable as antisemitism. The Blood Libel content is delivered through Israel-as-perpetrator narrative rather than through Jews-as-perpetrator narrative, which permits its circulation in spaces where the explicit Christian-historical framing would be foreclosed.

KW-09 (The Young Turks) carries the corpus’s densest Israel-comparison cluster. The relevant codings include I0 (legitimate criticism of Israeli policy), I1b + C5-1 (Nazi-analogy applied to Israel as collective antisemitism), and I1a + C3-5 (Instrumentalization of Antisemitism through Israel-criticism deflection). The internal distinction matters because the three modes do different work: legitimate Israeli-policy criticism is not antisemitism and does not enter the antisemitism count; Nazi-analogy applied to Israel slides into antisemitism through the Jewish-collective-as-Nazi claim; and the Instrumentalization-through-Israel-deflection move — typically:

“why do we have to talk about Kanye when Israel is committing genocide” (KW-09, recurring formulation)

— functions to reorder discursive priority rather than to articulate antisemitism directly. KW-09’s distinctive concentration of these registers (the thread carries the corpus’s only multi-instance C5-1 cluster) reflects the progressive-host condition’s audience-permission-structure for Israel-criticism content, which absorbs both legitimate criticism and antisemitic material under shared discursive surface.

6. Holocaust Distortion and Denial

Where Affirmation of Antisemitism endorses the Nazi reference itself, Holocaust Distortion and Denial operates on the historical record of Nazi violence: contesting death tolls, offering false equivalences, or redistributing responsibility for the Holocaust. The two registers can dual-code in the same comment — Hitler-rehabilitation constructions in particular typically carry both — but they are analytically distinct discursive moves. The Holocaust Distortion register operates at substantially lower corpus density than Affirmation (C3-3 surfaces in only a few threads), but the constructions it produces are often the corpus’s most sophisticated in their argumentative machinery. Two sub-registers organize the corpus’s Holocaust Distortion content:

Hitler rehabilitation. Favorable historical comparisons framing Hitler as a competent national leader rather than a genocidaire; “AshkeNazi Zionist pawn” revisionism in which Hitler is reframed as a Jewish agent and therefore not a legitimate object of antisemitism critique.

Holocaust distortion through false-equivalence. Structural arguments that admiration of Hitler is no different in kind from religious worship or from praise of other historical figures, deployed to claim parity between the song’s content and routine speech.

Holocaust distortion through numerical revision appears as a distinct sub-register, identifiable through a recurring numerical marker. A brief gloss for readers outside Holocaust-distortion literature: 6 million is the widely-accepted scholarly and memorial figure for Jews murdered in the Holocaust, established through forensic, archival, and demographic research compiled in the postwar period and now standard in Holocaust historiography. Within the Holocaust-denial and white-supremacist online ecosystem, 6 million functions as the targeted figure — the number “deniers” claim to disprove or revise downward. The counter-figure 271,000 (often abbreviated 271k) circulates in revisionist literature as a substitute count derived from selective readings of Wehrmacht statistical documents (most prominently the Korherr Report) interpreted against the historical consensus. When 271k appears in proximity to 6 million in online comment registers, the construction operates as a coded denial speech act: the writer is articulating Holocaust-revisionist content through a numerical-substitution dog whistle that codes as innocuous to readers outside the ecosystem and as full denial to readers inside it. The construction is a recurrent pattern in online Holocaust-denial discourse — a coded form that reads as arithmetic to outside readers while functioning as denial inside the ecosystem.

The most sophisticated Hitler-rehabilitation case is KW-05 Comment 11-3:

“You mean the Austrian Painter that enacted Animal Rights Laws, anti-child labor laws and worked almost his whole life for the prosperity of the German people and succeeded in reaching almost full employment at a time America was in the grips of the Great Depression?” (KW-05 Comment 11-3)

an accomplishment-listing register that recodes Hitler as a competent national leader rather than a genocidaire, delivered through the “Austrian Painter” euphemism that itself circulates as a coded Hitler reference. Coded as I1a + C4-3 (Affirmation). The construction is the prototype of historical-rehabilitation through selective-accomplishment listing — the Holocaust is absent from the list precisely because it is what the construction is rehabilitating around. Where rehabilitation-by-accomplishment codes as Affirmation, the corpus’s numerical-revision constructions (KW-04 Comment 4; KW-08 Comment 18-4, both I1b + C3-3) carry the Holocaust Distortion code proper: the two registers work the same rehabilitative terrain from different directions.

The corpus’s only multi-trope Affirmation construction with full Holocaust-distortion machinery is KW-05 Comment 9-07, a 120-word comment articulating four antisemitism registers simultaneously:

“the exaggerated tales of suffrage” (KW-05 Comment 9-07, excerpt)

The comment articulates Holocaust distortion (the excerpt above), Lie/Deceit about Holocaust narratives, Power (Zionist influence on US policy), and Instrumentalization. It requires contextual reading throughout (I1b ideation) and codes with the full four-concept set (C3-3 + C1-6 + C2-2 + C3-5). The interest is structural — at the comment’s length and density, the construction approaches the form of essayistic antisemitic commentary rather than reaction-text, signaling that some YouTube commenters use the comment space as an extended-prose vehicle for antisemitic discourse rather than reactive engagement.

A cross-thread 271k Holocaust-distortion register circulates in the corpus as a compressed numerical-substitution device. The construction operates by displacing the widely cited “6 million” figure for Jewish victims of the Holocaust with the number “271k” — a formulation drawn from Holocaust-denial literature that argues (falsely) for radically lower death tolls. The construction surfaces in KW-04 Comment 4 (the panel-thread post-release) and in KW-08 Comment 18-4 (Brad Polumbo’s hybrid-critical thread), where it is deployed inline in response to Comment 18’s quotation of the antisemitic slogan “6 million was not enough.” Both comments code as I1b + C3-3 (Holocaust Distortion & Denial). The register is analytically distinctive because it operates through a single number rather than through explicit denial rhetoric: the numerical substitution carries the denial while the surrounding text remains ostensibly conversational, making the construction difficult to moderate at the platform level. Its cross-thread circulation — appearing in adversarial-secular and hybrid-critical conditions alike — suggests that Holocaust-denial numerical dog whistles are becoming a portable vocabulary that migrates independently of host framing.

The corpus’s most direct Nazi-propaganda recommendation is KW-10 Comment 7-052:

“Go watch Europa the final battle.” (KW-10 Comment 7-052, from a longer comment that opens “The word antisemitic means nothing… you are brainwashed into what you currently think.”)

The reference is to a 2017 Holocaust-revisionist documentary that circulates in white-supremacist online discourse as a primary Holocaust-denial vehicle. The comment codes as I1b + C3-4 (Relativization & Denial of Antisemitism, via “the word antisemitic means nothing”) + C4-3 (Affirmation) + C2-3 (Conspiracy Theories, via the brainwashing frame). Notable because it is the first instance in the corpus of active Nazi-revisionist propaganda recommendation — the comment goes beyond circulating antisemitic content and directs other commenters to additional Holocaust-denial material as if to a reading list. The recommendation register surfaces in KW-10 (Sam Seder, meta-progressive condition), suggesting that even under conditions strongly hostile to antisemitism, the comment-space functions as a discovery channel for white-supremacist material.

Comparative Observations

The corpus intersects with prior work in this research line on two structural points, both worth flagging briefly.

First, on cross-study observation: the prior Bilzerian study (Becker 2026c) documented audience uptake of a coded speaker-driven mode in which conspiracy elaboration was the dominant audience response. The Kanye West corpus documents an explicit speaker-driven mode in which direct Affirmation of the Nazi reference itself appears at densities that coded discourse in the prior study did not surface. This is a within-research-line contrast, not a controlled comparison: the two studies differ in speaker category, artifact form, and supply mode, and the design does not let us isolate which difference accounts for the shift. The Limitations section develops this caveat; the within-corpus host-framing comparison developed in Findings 1 and 2 does not depend on the cross-study reading.

Second, on the conversion of institutional moderation into conspiracy input: prior event-conditioned studies (Becker 2026a, Becker 2026b) documented the same mechanism observed here. At Temple Israel, the FBI’s antisemitism designation was reframed as cover story; at WHCD, the Secret Service response was reframed as evidence of staging; in the Kanye West corpus, streaming platforms’ removal of the song is reframed as confirmation of the conspiracy the song articulates. The mechanism appears across cases; the input differs.

Discussion and Implications

For journalists, policymakers, and platform teams. The full six-point discussion follows below. In brief:

• Explicit Nazi-referential content is not confined to extremist spaces. The corpus surfaces substantial endorsement of a song built around explicit Nazi references across mainstream-adjacent YouTube formats, including under hosts who condemn the song on camera. Content-moderation frameworks calibrated to coded or indirect antisemitism may not fully capture what explicit material generates in audience response.

• How a host frames the content shapes what the audience does with it. Adversarial hosts see audience hostility redirect at the host rather than the content; sympathetic hosts see the content’s framings extended and elaborated. Volume of antisemitic content is a poor stand-alone metric — what the audience does with that content varies substantially by host condition.

• Platform removal, on its own, can be recoded as evidence for the framework it aims to disrupt. When a moderation action arrives without a public account of why the content meets the moderation threshold, conspiratorial audiences can absorb the removal as further confirmation of the very narrative it was meant to interrupt. The framing around a moderation decision — how it is communicated, whether its rationale is publicly legible — affects the discursive environment as much as the decision itself.

• Counter-speech capacity is unevenly distributed. Substantive counter-speech accounts for only 3.2% of coded comments corpus-wide, with one thread containing none at all. The conditions producing the most antisemitic content are not necessarily the conditions producing the least pushback. Interventions calibrated to a specific host condition are likely to be more empirically informed than platform-wide generic strategies.

• The civic-stakes frame extends beyond the specific target group. When the historically specific vocabulary of mass violence — the salute, the archival audio, the numerical vehicles of Holocaust denial — is routed through the affective architecture of pop music and circulates as joke material and meme material, what thins over time is the shared public capacity to recognize such atrocity as an event that grounds present-day obligations. This effect is not confined to antisemitism; it operates wherever the historical persecution of an identifiable community is metabolized into entertainment.

The findings above concern audience response patterns in one corpus. They are not, on their own, evidence for a general theory of platform-mediated antisemitism, nor do they license causal claims about specific host-audience relationships beyond this case. The relations documented in this article are within-case associations under a documented set of host framings — informative for how audience response to explicitly Nazi-referential material varies with host framing on a single platform, more tentative when extrapolated to other artifacts, platforms, or discursive environments. The discussion below is offered in that register: implications the findings appear to have for platform moderation, public discourse, and further research, with the case-bounded character of each implication marked where relevant.

The first implication concerns the discursive labor moderation performs. Streaming platforms removed “Heil Hitler” within days of release. Removal did not end the song’s circulation: the song continued to circulate on YouTube through reaction and commentary videos, on Meta-owned platforms directly, and in audience-facing spaces where the song’s Nazi content and the removal decision itself became discursive objects the audience could take up. The Rogan formulation — that platform removal “kind of supports what he says, which is that there’s this concerted effort” — is not incidental to how the audience response unfolds. Under the sympathetic-amplifier condition, the removal is reframed as confirmation of the conspiracy the song articulates: the moderation act is metabolized as evidence for the framework it was intended to disrupt. This pattern is consistent with the general observation, documented across the research line’s prior case studies, that institutional counter-actions absent an accompanying public explanation of moderation rationale can be discursively converted into inputs for the frameworks they aim to counter. For platform-level moderation policy, this suggests that the discursive frame around a moderation decision — how the decision is communicated, whether the moderation rationale is publicly legible — is not merely a communication matter downstream of the decision but is part of what determines the decision’s effect on the discursive environment.

The second implication concerns the relationship between host framing and audience response. The corpus does not support the simplest version of the sympathetic-amplifier thesis, in which sympathetic host framing produces higher antisemitism density in the comment environment. The right-host thread (KW-02) exceeds the sympathetic-amplifier thread (KW-07) in antisemitism density; the corpus floor is the pre-release adversarial thread (KW-01, 12%), with the hybrid and meta-progressive conditions producing the lowest post-release densities (17%). What the corpus does support is that host framing appears associated with the mode of audience response — which registers surface as available discourse — rather than with volume alone. Under adversarial framing, hostility toward the host becomes an available register. Under sympathetic-amplifier framing, direct conspiracy elaboration becomes an available register, alongside a substantial counter-articulation register that other conditions do not produce. This distinction matters for how platform-level and public-facing responses to hate-adjacent content are evaluated: focusing on volume alone risks missing that host framing shapes what the audience discourse becomes, not only how much of it accumulates.

The third implication concerns the explicit-versus-coded distinction and its consequences for platform moderation. Prior speaker-driven studies in this research line documented antisemitic discourse operating through coded scaffolding — plausible-deniability constructions that route the antisemitic content through hint, insinuation, and dog whistle. The present corpus documents an audience response register in which endorsement of the Nazi reference operates without such scaffolding: commenters affirm the explicit content of the song directly, in formulations that do not require interpretive reconstruction. The prior comparator differs from the present case on multiple dimensions — speaker category, artifact form, supply mode — and the research design does not isolate which difference accounts for the shift, so the association reported here is not a general claim about coded versus explicit supply. It does suggest, however, that when explicitly Nazi-referential material enters the discursive environment, the register of audience response can shift in ways that content-moderation frameworks calibrated to coded discourse may not fully capture. Moderation frameworks that classify by keyword or by structural feature — Nazi salute, Nazi vocabulary, Holocaust denial numerics — will register such content; frameworks that classify by tone or by inferential distance may not. This is not an argument for either type of framework, but for calibration to the register the audience discourse is actually operating in.

The fourth implication concerns counter-speech capacity and where it appears. Counter-speech in this corpus concentrates unevenly. The sympathetic-amplifier thread (KW-07) produces the corpus’s densest substantively engaged counter-speech alongside its second-highest antisemitism density; the progressive ban-debate thread (KW-09) matches its count through explicit pro-ban stances. The reaction-creator thread (KW-06) produces zero counter-speech, and the adversarial-religious thread (KW-05) a single instance. Where counter-speech does appear, it tends to coincide with either a substantive antisemitism register the audience can engage directly or with explicit antisemitism-defending clusters that invite rebuttal. This is a case-specific pattern, not a generalization about counter-speech dynamics on YouTube. It does suggest, however, that counter-speech is not uniformly enabled across host conditions: the conditions that produce the most antisemitism content are not necessarily the same conditions that produce the least pushback, and the conditions that produce the most pushback are not necessarily the least conducive to antisemitism content. For public-discourse interventions oriented around counter-speech promotion, this suggests that engagement strategies calibrated to a specific host condition — the discourse the audience is likely to encounter — may be more empirically informed than strategies calibrated to a generic assumption about the platform-level environment.

The fifth implication concerns the erosion of civic conditions through the trivialization of past atrocities. The findings above are empirical patterns; this implication is the civic frame within which those patterns acquire their public stakes. The song under analysis takes the historically specific vocabulary of Nazi persecution — the salute in the title, the Führer’s voice in the sample, the numerical vehicles of Holocaust denial that circulate in the audience response — and routes it through the affective architecture of pop music: a hook, a beat, a chorus designed for repetition. The register documented across the corpus’s Affirmation cluster — “I did Nazi that coming,” “Yedolf turning up the gas,” aesthetic praise of the song as “catchy” — is the audience-side continuation of the same operation: historical mass violence against an identifiable population becomes joke material, aesthetic material, meme material. This is not a marginal register in the corpus. Affirmation is the modal concept code across every one of the ten threads. What such a discursive environment corrodes is neither the memory of the historical atrocity in the archival sense — the historical record remains what it was — nor the specific safety of Jewish audiences alone, though both are at stake. What it corrodes is the shared civic capacity to recognize such an atrocity as an event that grounds present-day obligations toward the group against which it was committed. When that capacity thins across a broad public through repeated affective exposure to trivializing content, the political ground on which mutual recognition rests thins with it. This effect is not confined to antisemitism; the same structural erosion operates wherever the historical persecution of an identifiable community is metabolized into entertainment. That the specific vehicle in this case is a pop song delivered by one of the most-followed cultural producers of the past two decades, and that the song’s audience-facing circulation continued through mainstream YouTube channels after streaming-platform removal, is what makes the case relevant to civic-discourse conditions in this moment — not to make a normative claim about the artist, but to name the public stakes of the discursive environment the article documents.

The sixth implication concerns methodological and research-design directions the case indicates. The design here holds one artifact constant and varies the host condition — a variation that is possible because “Heil Hitler” was engaged as a discursive object across ten distinct host framings during a compressed reception window. Future work in this research line could productively invert the design (varying the artifact while holding the host condition constant), extend it (adding additional host conditions and additional artifacts to disentangle the confounds this article’s Limitations section documents), or apply it comparatively to non-English-language platforms and non-YouTube discursive environments. The three-annotator consensus coding procedure applied here is one option among several; formal Cohen’s κ measurement on sub-samples, planned for the parallel Digital Hate Review article, will provide a complementary reliability metric. The analytical construct that emerged in this case — audience permission structures inferred from interactional patterns rather than from frequency alone — is offered as one working framework future studies may take up, refine, or contest against additional data.

Taken together, the implications above are neither universal claims about digital antisemitism nor prescriptions for specific platform policies. They are the working conclusions the case appears to support — offered to platform-moderation practitioners, discourse-analytic scholars, and policy researchers as inputs to further inquiry, not as settled conclusions the corpus itself demands.

Conclusion

This study makes three points about audience response to an explicitly Nazi-referential cultural artifact.

First, host framing matters, but not always in the way standard content-moderation accounts predict. When an adversarial host is already treated by the audience as illegitimate, criticism is absorbed into the conspiracy frame and redirected toward the host rather than suppressing uptake of the song itself. When a sympathetic host articulates the conspiracy frame on camera, the audience is more likely to elaborate that frame directly. What changes across host conditions is which discursive moves commenters treat as ordinary and which moves draw friction.

One mechanism these patterns suggest is not amplification per se but the shaping of audience permission structures by host condition. Under adversarial framing where audience legitimacy is denied, commenters defend the song and reframe the host as the illegitimate party. Under sympathetic-amplifier framing, commenters elaborate the song’s claims directly, alongside substantial counter-articulation against them. The contrast lies less in the volume of antisemitic content commenters produce than in which register that content takes — and which moves draw challenge. Reframed this way, “sympathetic amplification” is one instance of a broader phenomenon: host-conditional shaping of permission structures in audience response to explicit content.

Second, the corpus shows that the song’s explicit Nazi reference can be accompanied by a distinct response register: direct affirmation of that reference itself. This register appears here at densities not documented in the prior speaker-driven corpora in this research line — a within-research-line observation, not a generalization about explicit-vs-coded discourse, since the available comparator differs from the present case on multiple dimensions (see Comparative Observations above).

Third, platform intervention can become part of the interpretive machinery of the conspiracy frame. Removal, demonetization, or other moderation actions are recoded by commenters as proof of hidden control when the intervention is not paired with a clear account of why the material violates policy. In this corpus, moderation without explanation was frequently interpreted by commenters as confirmation of the very narrative it was meant to interrupt.

Taken together, these findings support a more precise formulation of sympathetic amplification: sympathetic hosts do not necessarily increase antisemitic volume; they alter the discursive object and the audience response register that surfaces in the thread. The implication is that host framing shapes which form audience response takes — defense of the song, attack on the host, elaboration of the conspiracy framework, or counter-articulation against it — and that moderation strategies targeting volume alone may miss the discursive operations at issue.

Limitations

Confounding of register, speaker category, and artifact form. The descriptive cross-study observation reported in Comparative Observations — that the Affirmation register surfaces more densely in the present case than in the prior speaker-driven comparator — is not reducible to the explicit/coded axis alone. Two alternative explanations require acknowledgment. First, the prior comparator involves a political-influencer speaker; the present case is a cultural-celebrity speaker, and the two speaker categories may activate different audience dispositions independent of register. Second, the prior comparator is podcast-interview material; the present case is a song, and audience interaction with music differs from audience interaction with political commentary along dimensions (repeated consumption, lyrical memorization, aesthetic-affective engagement) that the comment data alone cannot disentangle from the explicit/coded contrast.

The article reports the cross-corpus difference as a descriptive observation rather than a causal claim for this reason. Disentangling these factors would require additional cases — a cultural celebrity producing coded antisemitic content, or a political influencer producing an explicit cultural artifact — that future work could supply.

Artifact generalizability. The article does not claim that the Affirmation register documented here generalizes to all explicitly Nazi-referential cultural-production content. The claim is bounded: where a high-reach speaker delivers a song built around explicit Nazi references through a cultural-production vehicle, the audience response is consistent with including a direct Affirmation register not prominent in prior speaker-driven studies.

Host-condition coverage. The corpus includes ten host conditions but does not exhaust the relevant configurations. Notable absences include left-mainstream news commentary (CNN, MSNBC published coverage with weaker audience engagement than the included outlets) and far-right alternative platforms where the song circulated in formats not amenable to the YouTube comment sampling protocol.

Sampling and representativeness. The 100-comment-per-video chronological sample captures early, visible participation rather than the full engagement distribution, and may overrepresent highly active or ideologically aligned users. Comments removed by platform moderation before data collection are not captured.

Platform distribution as confound. A separate caveat applies to the within-host Morgan comparison (KW-01 vs KW-04). The audience-composition note above concerns who chose to comment on each video; a logically distinct factor is who YouTube’s recommendation algorithm presented each video to in the first place. The same host channel can reach systematically different audiences across uploads depending on topic, timing, related-video co-occurrence, and the algorithm’s contextual signals (e.g., whether a video is surfaced to viewers who recently engaged with antisemitism-adjacent content). The comment data cannot distinguish a permission-structure shift produced by the song’s discursive presence from a shift produced by differential algorithmic distribution that brought a partly different viewership to KW-04. Both could be operating, in which case the comment-level pattern is the joint product of audience composition, algorithmic exposure, and discursive object. Disentangling these would require platform-side viewership data that the comment corpus alone does not provide.

Supplementary Appendix

Standing Coding Rules

The following rules emerged from three-annotator adjudication sessions and were applied retroactively across the full corpus once formalized. They are documented here to support replication and calibration for future work in the research line.

Puppet-family rule. Puppet-family metaphors (puppet, asset, agent, pawn, stooge, payroll, controlled-by, on-behalf-of) code as Conspiracy (C2-3) under the puppet-rule rather than as Power (C2-2), on the grounds that these constructions invoke string-puller logic — a hidden controlling agency behind the visible actor — rather than bare-influence attribution. Power (C2-2) is reserved for bare-influence claims that do not invoke string-puller logic (e.g., “Zionist megaphone”).

Puppet-rule quotation exception. Quoted or mocked puppet-constructions do not trigger C2-3; the rule fires only on the commenter’s own on-behalf claims. Paradigmatic case: KW-02 Comment 4-13, in which “Working on behalf is ‘isreal’” quotes and mocks another user’s formulation before the commenter’s own contribution (“What has he said that wasn’t true? He’s had Jewish people fucking with him forever”) — adjudicated to I1a + C4-3, the Affirmation carried by the truth-endorsement, with no Conspiracy code because the puppet-construction is not the commenter’s own claim.

Explicit-versus-coded rendering rule. Comments in which the target group is written out explicitly (”Jews,” “Jewish culture”) code as bare I1 (Explicit AS). Censored, coded, or euphemistic renderings that require linguistic or world knowledge to resolve — “j***s,” “(((them))),” “little hats,” “noticer,” portmanteaus — code as I1b (Contextual AS). Parent-derived readings code I1a (Co-textual AS); readings requiring cultural or historical knowledge external to the thread code I1b.

A4 default. Aesthetic praise of “Heil Hitler” codes as Affirmation (C4-3, A4 sub-register) absent an explicit distancing move within the comment itself. Formulations that qualify as distancing include “I wish the lyrics were changed,” “shame about the message,” “despite the content,” or an irony marker that distances the speaker from the Nazi referent specifically. Subsequent self-disavowals in reply chains do not retroactively change a coded A4 instance.

Schadenfreude / affective-cover rule. Schadenfreude operating as affective cover for substantive antisemitic content codes under the substantive category rather than under a stand-alone affective category, on the grounds that the affective register operates as delivery mechanism rather than as the substantive discursive move.

Humor and irony as Affirmation markers. Irony and humor count as positive markers of antisemitic content when the humorous frame is built on the antisemitic referent itself (e.g., “I did Nazi that coming”) rather than on a target external to that content. The humorous frame is treated as an Affirmation vehicle when the antisemitic referent is the joke’s structural anchor.

Humor-as-affirmation rule. The humor-as-affirmation rule extends the previous rule to comments engaging antisemitic content primarily through entertainment or humor framing that indirectly affirms it via trivialization. This includes cases in which a commenter’s Jewish self-identification does not neutralize the trivialization move — the identification itself may function as authorization for other commenters to trivialize. Paradigmatic case: KW-08 Comment 9.

Emoji-confirmation rule. Celebratory emojis (🎉, 🔥, ⭐, 💯 and equivalents) attached to an antisemitic-content anchor code as confirmation: C4-3 + MMD, with ideation I1a where the anchor is a thread parent (co-textual derivation) and I1b where the anchor is the video/song itself (contextual derivation). Laughing tokens (😂, 🤣, “lmao”) are not per se confirmation and code I0 absent affirming co-text. Established via KW-06 Comment 4 (“🎉🎉🎉”, I1b + C4-3 + MMD) and calibrated via KW-02 Comments 4-03 and 4-33 (I1a + C4-3 + MMD) and KW-01 Comment 7-1 (“Lmao 😂” under an antisemitic parent, I0).

Anti-ban rule. A bare or equivalence-based anti-ban stance toward the song (“No it should not be banned”; “Guess we should ban the Bible then”), without condemnation attached, codes as I1a + C3-4 (Relativization & Denial of Antisemitism) — opposing the removal of openly Nazi-referential content without distancing functions as relativization. Anti-ban stances with explicit condemnation attached (“Banning a deplorable song will only intensify the attention… let the market boycott it”) code I0.

Pro-ban counter-speech rule. An explicit pro-ban stance toward the song codes as I0c (Counter-speech). Generic condemnation of the artist without a ban-stance or substantive engagement with the antisemitic content codes I0, not I0c.

Threats and death wishes. Direct or implied threats — first-person intent, target-directed menace, location solicitation — code C6-2 (Threats); death wishes code C6-2 with the qualifier “death wish” appended. Violence endorsement in the abstract, without target-directed menace, is distinguished from threats.

Platform label set. Final concept labels: C4-3 “Affirmation of Antisemitism”; C1-2 “Evil / Moral Corruption”; C1-5 “Religious Antisemitism.” These supersede earlier working labels (”Affirmation of Nazis, Hitler, & the Holocaust”; “Evil/The Devil”; “Immorality”).

Full Concept-Frequency Matrix

The compressed matrix in the main text shows the twelve concept codes with the highest corpus-wide frequencies. The full distribution adds the rare-incidence codes that appear in fewer than five comments across the corpus, organized by Lexicon chapter (Becker et al. 2024):

Supplementary Figure. Concept-frequency heatmap (top 12 codes). (To be re-rendered from the matrix in the main text.) The heatmap makes visible the same patterns the matrix records — Affirmation (C4-3) as the corpus’s modal register, densest in KW-02 (27) and KW-06 (20); the Relativization / Taboo / Conspiracy / Power secondary cluster, with Relativization (47) now the clear second register, densest in KW-09 (12); KW-05’s distinctive Religious Antisemitism (C1-5) concentration (6); KW-07’s Taboo (C3-8) peak (8); KW-10’s co-dominant Affirmation/Relativization profile (8/8); and KW-09’s Nazi-Analogy (C5-1) concentration (6) — at a single glance.

All counts reflect the team’s three-annotator post-adjudication consensus, including the puppet-rule audit applied to the KW-01 host-attack cluster (Comment 21 and Comment 54 reclassified from C2-2 to C2-3 per the standing rule that puppet/asset/agent/pawn/stooge metaphors invoke string-puller logic and code as Conspiracy rather than Power). The KW-04 C2-2 count of 0 — paired with a corpus-high C6-1 (Insults) count of 6 and the corpus’s only C6-2 (Threats) codings (2) — reflects the register shift discussed in the main-text footnote on the matrix: KW-04’s host-attack content operates through the slur-and-threat register rather than the puppet-family register that organized KW-01.

References

Becker, Matthias J., 2026a. “The Intervals Are Shrinking: Antisemitic Attacks in the US and their Digital Aftermath — Reactions on YouTube after the Temple Israel Synagogue Attack.” MJB’s Substack, 13 March 2026. https://decodingantisemitism.substack.com/p/the-intervals-are-shrinking-antisemitic

Becker, Matthias J., 2026b. “’Faker Than a Three Dollar Bill’: The White House Correspondents’ Dinner Shooting and the Conspiratorial Afterlife of Political Violence — A YouTube Discourse Analysis.” MJB’s Substack, 26 April 2026. https://decodingantisemitism.substack.com/p/faker-than-a-three-dollar-bill-the

Becker, Matthias J., 2026c. “At the Bottom of the Iceberg: Dan Bilzerian, the Congressional Run, and the Mainstreaming of Antisemitism.” MJB’s Substack, 1 May 2026. https://decodingantisemitism.substack.com/p/at-the-bottom-of-the-iceberg

Becker, Matthias J., Gabrielle Beacken, and Liora Sabra, 2026. “How Political Ambiguity Transforms Across Digital Platforms: Mechanisms of Discourse Radicalization in the Case of Zohran Mamdani.” NYU Ultraviolet, 12 January 2026. https://ultraviolet.library.nyu.edu/records/6m1xt-2nf20

Becker, Matthias J., and Liora Sabra, 2025. “Joe Rogan and the Descent from Reality.” The Times of Israel — The Blogs, 18 April 2025. https://blogs.timesofisrael.com/joe-rogan-and-the-descent-from-reality

Becker, Matthias J., Hagen Troschke, Matthew Bolton, Alexis Chapelan (eds.), 2024. Decoding Antisemitism: A Guide to Identifying Antisemitism Online. Springer Nature. https://link.springer.com/book/9783031492372

Blue Square Alliance Against Hate, 2026. “Kanye West’s History of Antisemitism: From ‘Death Con 3’ to ‘Heil Hitler’.” Command Center Insights, 24 June 2026. https://www.bluesquarealliance.org/command-center-insights/kanye-west-history-of-antisemitism

France, Lisa Respers, 2026. “Does anyone really want to deal with Kanye West?” CNN, 13 April 2026. https://www.cnn.com/2026/04/13/entertainment/kanye-west-ye-return

Gilson, Grace, 2025a. “Joe Rogan defends Kanye West’s ‘Heil Hitler,’ echoes antisemitic stereotypes in song.” The Times of Israel — The Blogs, 17 May 2025.

Gilson, Grace, 2025b. “Russell Brand defends Ye’s ‘Heil Hitler’ music video.” Forward, 13 May 2025.

AddressHate Research Scholar at NYU’s Center for the Study of Antisemitism | Lead, Decoding Antisemitism — PI, Decoding Hate | Research Advisor, AddressHate | Editor-in-Chief, Digital Hate Review

Previously in this series:

Part I — Antisemitism in Real Time, Part I: U.S. YouTube Reactions to Israel’s Strike on Iran (June 2025)
Part II — Antisemitism in Real Time, Part II: U.S. YouTube Reactions to Iran’s Retaliatory Strike on Israel (June 2025)
Part III — Antisemitism in Real Time, Part III: “Israel Made America Do It” — YouTube Reactions to the US–Israel Campaign Against Iran (March 2026)

Executive Summary

On June 14–15, 2026, the United States and Iran announced an initial agreement after more than three months of war, signed by President Trump on June 17. Israel was not a party to the agreement, and Israeli reactions ranged from cautious skepticism to outright opposition. The announcement also exposed visible tensions between Trump and Netanyahu, including Trump’s public criticism of Israeli strikes in Beirut that he said risked the deal. The agreement’s stability came into question in the days that followed.

This article asks how YouTube comment sections under U.S. news coverage of these events respond, and what happens to the antisemitic discursive register documented in Parts I–III when the political conditions that underpinned Part III’s interpretation had substantially changed.

For several weeks in mid-June 2026, the specific condition of U.S.–Israeli co-belligerence — which Part III argued was responsible for the subordination register — no longer held: Trump signed a deal Israel publicly opposed, exposing visible tension in the U.S.–Israel relationship. Part IV tests how the register responds across three temporal phases — Phase A (rift buildup, May 20 – June 10), Phase B (deal-period rupture, June 12 – June 16), and Phase C (post-signing destabilization, June 17 – June 21, 2026) — sampling the same eight outlets as Parts I and III (CNN, MS NOW [formerly MSNBC, rebranded November 2025], NBC, CBS, ABC, AP, Fox News, Forbes Breaking News) with 4,598 coded comments across 23 threads.

The register does not disappear under apparent contradiction; it reorganizes. The base “Trump-is-Bibi’s-puppet” register stays roughly flat per thread across all three phases. What grows most sharply is the blackmail and hidden-control sub-form, in which Trump’s compliance with Israeli interests is explained not by overt political alliance but by Mossad’s covert leverage. The clearest single instance, from an MS NOW comment section, is the article’s title quote:

“A fake feud. We wouldn’t be in Iran in the first place if Netanyahu didn’t own Trump.” (MS NOW T2 #0016)

A real Trump–Netanyahu rupture — Trump publicly criticizing Netanyahu over the Beirut strikes, Vance rebuking the Israeli Cabinet on NBC — is neutralized by declaring the rupture itself fake. The findings identify a mechanism through which this conspiracist register remains stable despite contradictory evidence: rather than abandoning the parent narrative, speakers replace one explanatory sub-form with another that better accommodates the new political reality while preserving the underlying interpretive structure. This article terms the mechanism compositional mutation.

Scale and design. Part IV analyzes 4,598 coded comments across 23 YouTube threads — considerably larger than the previous studies — and tracks the same eight outlets across three successive political phases. A second finding emerges from the design: observed antisemitism density is strongly constrained by video framing (more Netanyahu-focused → more antisemitism), with outlet-audience composition modulating density inside that constraint.

Parts I–III asked how antisemitism shows up during Israel-related wars. Part IV asks whether it survives apparent contradiction, and finds that it does not collapse — it mutates.

Introduction

On June 14, 2026, after more than three months of war that began with the joint U.S.–Israeli Operation Epic Fury on February 28, the United States and Iran announced an initial agreement to end the conflict and reopen the Strait of Hormuz. Mediated by Qatar and Pakistan, the memorandum of understanding was signed in person by President Trump at a state dinner at the Palace of Versailles on June 17 following the G7 summit in Évian-les-Bains, France, and signed remotely by Iranian President Masoud Pezeshkian; the document was finalized in both English and Farsi at Iran’s insistence. Follow-up technical talks scheduled for Bürgenstock, Switzerland on June 19 were postponed after Iran demanded guarantees that Israel would cease its operations in Lebanon. The memorandum extends the Iran ceasefire to Lebanon, though Israeli officials have stated that Israel reserves the right to continue operating against Hezbollah and will not withdraw from the southern Lebanon security zone. Israel was not a party to the agreement and was not directly involved in the negotiations.

Netanyahu’s first public response, delivered at a press conference in Jerusalem on June 15, was measured: he acknowledged that, with Trump, “many times we see eye to eye, and there are also cases in which we see less eye to eye”, while making clear that Israel does not intend to withdraw from southern Lebanon, Gaza, or Syria. Israeli opposition figures and former senior security officials were considerably sharper: Yair Lapid said Netanyahu “can no longer fix” the situation and committed to defeating him at the upcoming election; Yair Golan and former Prime Minister Ehud Barak framed the agreement as a strategic failure attributable to Netanyahu’s leadership. The U.S.–Israel relationship surrounding the announcement was visibly strained: Trump publicly criticized Netanyahu over Israeli strikes in Beirut, which he said risked the deal. Vice President Vance, defending the deal, rebuked Israeli cabinet officials who had criticized it, saying he “might not be attacking the only powerful ally” Israel has “anywhere left in the entire world.”

This article examines what happens to the antisemitic discursive register documented in Parts I–III of the Antisemitism in Real Time series under the materially changed political condition this episode represents. The empirical material consists of YouTube comment sections under U.S. mainstream-news coverage of the deal, sampled across three phases of the surrounding Trump–Netanyahu rift. The specific question is whether the subordination frame documented in Part III — discourse portraying American political agency as covertly controlled by Israel or Jewish interests — changes when the political configuration that prompted its consolidation in Part III no longer holds.

Two findings structure the analysis. First, the subordination frame does not disappear — it reorganizes. The base subordination idiom (dog/master, puppet, owns-the-party) declines as a share of the Power and Conspiracy register while the blackmail and hidden-control sub-form rises from approximately 18 % to approximately 57 % of all Power-and-Conspiracy content across the three phases. Second, observed antisemitism density across the 23-thread corpus is strongly constrained by how directly the video centers the Trump–Netanyahu confrontation versus the Iran-deal mechanics, with outlet-audience composition modulating density inside that constraint. The first finding bears on the trajectory of one discursive register under empirical contradiction; the second, on how density figures should be read in cross-platform and cross-period comparison.

This article is the fourth installment in the Antisemitism in Real Time series, which tracks how antisemitic discourse is activated by geopolitical events and adapts to them as they unfold in YouTube comment sections beneath major U.S. news coverage. Across all four installments, the same protocol and analytical instrument are applied; Parts I, III, and IV sample the same eight outlets.

Part IV differs from the prior installments in one substantive respect: its triggering event is diplomatic rather than military. Comment-section dynamics in response to a settlement need not mirror dynamics in response to strikes or escalation, and raw rate comparisons between Parts I–III and Part IV accordingly require interpretive care.

The Four-Study Arc: Testing a Structural Claim

The first three installments produced a cumulative argument: the antisemitic register does not simply recur across events — it adapts to them. Part I documented predominantly religious and demonological registers. Part II documented a pivot toward emotional celebration of Israeli suffering. Part III documented the emergence of a consolidated subordination frame and advanced a conditional claim: when the United States became a full participant in a war publicly framed as aimed at regime change alongside Israel, conspiratorial discourse about Jewish control of American foreign policy acquired what Part III called a political scaffold. Part IV makes that claim testable: if the scaffold changes substantially, the discourse should change in one of three ways:

The three outcomes are not mutually exclusive; persistence and mutation in particular can co-occur, with the parent category persisting while the dominant sub-form changes. The analytical task is to identify which patterns predominate at which outlets, and across which subtypes of antisemitic language.

The hypotheses can also be read as a specific test case for a broader question about conspiracist discourse. When political events appear to contradict the core claim of a conspiracy theory, what happens to the theory? Does the framework collapse, attenuate, or absorb the contradiction by reinterpreting it within its own terms? The literature on conspiracy belief — going back to Festinger, Riecken, and Schachter’s When Prophecy Fails (1956), and continuing through Barkun’s A Culture of Conspiracy (2003) on the epistemic structure of conspiracist worldviews — suggests the third outcome, but the mechanism has typically been documented in cult or fringe settings rather than in mainstream-news comment sections at scale. The Part IV corpus, where the political condition Part III treated as causal no longer holds, is a test case for whether the mechanism operates in a mainstream venue.

A second analytical question concerns the stability of cross-ideological symmetry. Part III’s most striking finding on this axis was the convergence between CNN and Fox News — opposite ends of the political spectrum producing near-identical rates (21 % and 20 %). The Trump–Iran deal places particular strain on this symmetry on the right, where the America-First wing should welcome a deal that ends U.S. involvement while the pro-Israel right faces a bind between defending Trump and defending Netanyahu. Whether the Fox News comment section continues to track CNN’s antisemitic content rate, or whether the two diverge, is an empirical question this study is positioned to answer.

A narrower follow-on question concerns the Trumpstein thread documented in Part III — the suffix-Jewification of Trump’s name within a cross-platform conspiratorial formation linking the U.S. war alongside Israel to the alleged suppression of Epstein-related files said to be controlled by Jewish elites (the broader online suffix-Jewification pattern is documented in Becker 2024). In the Part III corpus, the Trumpstein formation depended on the Trump administration actively prosecuting the war alongside Israel; the deal removes that political fact. Whether the thread attenuates, mutates, or persists is the narrower empirical sub-question this study addresses alongside the main hypotheses.

Dataset & Methodology

Corpus overview

The Part IV corpus comprises comments drawn from 8 U.S. news outlets across three analytically distinct phases of the Trump–Netanyahu rift: Phase A (rift buildup, May 20 – June 10), Phase B (deal-period rupture, June 12 – June 16), and Phase C (post-signing destabilization, June 17 – June 21, 2026). Each phase corresponds to a discursive event window in which the rift and the deal pass through a distinct political condition; the cross-phase comparison enables the three-hypothesis test set out in the Four-Study Arc section above.

Within each phase, each outlet is represented by exactly one thread — the single video in that outlet × phase cell with the strongest combination of engagement and rift-centrality. This one-video-per-cell selection rule produces a corpus of 23 videos across 24 outlet × phase cells, with one structural cell absence (CBS Phase B) and one documented sub-threshold inclusion (NBC Phase A, at 198 comments). The coded sample applies the Parts I–III standard of 200 comments per outlet, but per phase rather than pooled across phases, enabling per-outlet per-phase comparison. The total coded sample is 4,598 comments (200 × 22 filled cells at threshold + 198 from the NBC Phase A sub-threshold cell).

Outlet selection follows Parts I and III: the same eight outlets are sampled in this study.

Operational definition of rift-centrality

The one-video-per-cell rule invokes “rift-centrality” as a selection axis alongside engagement. To prevent the term from functioning as post-hoc justification, the study operationalizes it through an explicit six-criterion definition applied uniformly to the candidate corpus.

A video qualifies as rift-central if its title, description, or primary news hook explicitly references at least one of: (1) Trump–Netanyahu disagreement, (2) Israel as non-party to the deal, (3) Israeli strikes threatening the deal, (4) U.S. administration criticism of Israeli officials, (5) Lebanon–Hezbollah as deal-violation site, (6) phase-anchor deal-events (signing, Strait of Hormuz closures and reopenings, postponement of follow-up talks).

A single match qualifies a thread as rift-central. Criteria 1–4 target the bilateral relationship directly; Criterion 5 captures Lebanon–Hezbollah as a specific deal-violation vector; Criterion 6 captures phase-anchor deal-events. Selection follows a two-stage rule: threads are first filtered for rift-centrality and the 200-comment engagement minimum, then the selected thread combines strongest engagement with the most directly bilateral criterion match. The criteria were specified before final thread selection and applied uniformly to the candidate corpus. Of the 23 selected threads, all 23 satisfy at least one criterion, with three within-criterion deviations at ABC Phase A, Fox News Phase B, and AP Phase C.

Video framing-position framework

Within the rift-centrality criteria, threads are further classified along a five-point framing-position axis that captures how directly the video centers the Trump–Netanyahu confrontation versus the Iran-deal mechanics:

• Position 1 — Direct Netanyahu-focused (e.g., “Netanyahu will do whatever I want him to do,” “Israel defies demand from Trump”).

• Position 2 — Israel-as-deal-breaker or deal-survival-conditional-on-Israel.

• Position 3 — Trump or VP versus Netanyahu confrontation (e.g., “Vance issues blunt message to Israeli Cabinet”).

• Position 4 — Deal mechanics with Israel as secondary actor (e.g., both-sides stop-shooting framings).

• Position 5 — Iran-deal-with-Netanyahu-minimal (e.g., Strait of Hormuz closure as primary focus).

Framing-position serves as the study’s principal structural variable in the Section 5.D quantitative apparatus and provides the analytical bridge between news production and comment-section discourse. Analysis of the sampled threads indicates that, in this corpus, framing-position substantially constrains observed antisemitism density, with outlet-audience composition operating as a second-order modulator inside that constraint.

The five-position scale is derived inductively from the 23 selected threads themselves. The first sweep across the candidate corpus identified the visible spectrum of video framings — direct Netanyahu-focused content at one end, Iran-deal-with-Netanyahu-minimal content at the other — and the intermediate positions were operationalized to capture the structurally distinct mid-range types observed: Israel-as-deal-breaker (Position 2), Trump or VP versus Netanyahu confrontation (Position 3), and deal-mechanics-with-Israel-secondary (Position 4). Five positions represented the smallest number of categories that preserved the empirically distinct framing types observed across the candidate corpus without collapsing conceptually different news framings. The scale’s discriminating power was therefore developed against the same corpus on which it is reported; this is acknowledged as a methodological constraint, and the regression line in Figure D2 is presented as descriptive of this corpus rather than as a generalization beyond it.

Coding decisions on framing-position were made by the PI on the basis of each video’s title, framing, and opening content. Boundary cases — particularly at the Position 2 / Position 3 split, where a video may mention Netanyahu without foregrounding the Trump/VP confrontation — were reviewed twice. Formal inter-coder reliability has not been computed for Part IV and is flagged here as a methodological limitation.

Corpus structure at a glance

The 24 outlet × phase cells of the Part IV corpus are structured as follows:

Legend: — = structural absence (CBS Phase B, no in-period video met the rift-centrality criterion). The NBC Phase A cell (198 comments) is a documented sub-threshold inclusion. Corpus total: 27,382 comments across 23 selected threads in 8 outlets.

Sampling procedure

The sampling target follows Parts I and III, with one design difference: 200 comments per outlet per phase, rather than 200 per outlet pooled across the study. For each filled outlet × phase cell, the first 200 comments of the selected thread are imported chronologically into the analytical software. Within each thread, top-level comments and replies are counted equally as discrete comments, in the order YouTube presents them when sorted chronologically. This research line treats antisemitic discourse as a relational and sequential phenomenon in which comments respond to, escalate, qualify, or contest one another; for a design focused on these dynamics, chronological sampling is preferable to random or interval-based sampling.

Five-week window: justification

Parts I, II, and III each sampled discrete events within tight 24- to 72-hour windows. The Trump–Iran deal did not arrive as a discrete event but as the culmination of a multi-event diplomatic rift spanning approximately five weeks, from Trump’s May 20 claim that Netanyahu “will do whatever I want him to do” through Iran’s June 20 closure of the Strait of Hormuz. Testing whether the subordination frame documented in Part III attenuates, mutates, or persists requires comments from across this entire rift rather than from any single moment, because the frame’s grammar may transform across the buildup, deal-announcement, and post-signing periods.

Annotation protocol

Annotation follows the linguistic and multimodal close-reading protocol used in Parts I, II, and III, supplemented by sequence and relational analysis, and guided throughout by the Decoding Antisemitism (DA) Lexicon (Becker et al. 2024).

The DA Lexicon’s analytical inventory is carried forward without modification. The Part III five-category structure (Power and Conspiracy Narratives; Demonology and Dehumanization; Schadenfreude, Death Wishes, and Glorification of Violence; Nazi Analogy; Classic Financial and Resource Exploitation Tropes) plus two Part IV additions (Denial of Israel’s Legitimacy; Discursive Strategies) supplies the coding spine. The final qualitative organization — five sections in Section 5.B plus a separate Discourse Dynamics section — reflects the cross-phase observations rather than the initial coding spine; the relationship between the two structures is set out in the Section 5.B opening, which maps each section to its Lexicon source and notes where the Part IV reorganization sits relative to the Part III coding categories. Whether and to what extent these categories actually dominate the Part IV corpus is an empirical question, not a methodological commitment.

Coding workflow

Comments were imported into the analytical software and annotated by trained team annotators using the DA Lexicon as the operational codebook. Each coding decision was reviewed by the Principal Investigator (PI); disagreements between annotator and PI were adjudicated against the DA Lexicon’s defined criteria for each sub-register, and the resolution was recorded for codebook calibration. Where a comment’s coding turned on a contextual reading not directly resolvable by lexical criteria — for example, irony readings or indirect-activation readings in which the antisemitic content depends on the surrounding thread context rather than on the comment’s surface text — the coding decision was made by the PI on a per-comment basis.

Category boundaries within Section 5.B carried recurring adjudication patterns rather than one-off difficulties. Three are worth naming because each affects multiple comments. The Power-and-Conspiracy / Financial boundary concerns AIPAC and donor-network framings, which can read primarily as control-mechanism (Cat 1) or primarily as financial-extraction stereotype (Cat 5); the operational rule applied here is that comments framing Jewish money as the mechanism of policy capture code under Cat 1, while comments framing the financial register as the content of the stereotype code under Cat 5. The Demonology / Financial boundary concerns parasite metaphors, which can function dehumanizingly (Cat 2) or as financial-extraction stereotype (Cat 5); the boundary is determined by the surrounding content (see Section 5.B.5 disambiguation paragraph). The Dual-Loyalty / Legitimacy-Denial boundary concerns “Jews-belong-elsewhere” framings, which can function as essentialized exclusion (Cat 2, Lexicon 1.8 Disloyalty) or as a denial-of-sovereignty argument applied to Israel (Cat 4, Lexicon 5.7 Denial of Right to Exist); comments framing the predicate as a character claim about Jews code under Cat 2, while comments framing it as a sovereignty argument code under Cat 4.

Sampling rationale and scope

The choice of YouTube comment sections as the data source is continuous with Parts I–III of this series and reflects three properties of the platform: its position as a mass-readership engagement venue under mainstream-news content, its hosting of long-form threaded discussion that enables observation of reply-context dynamics, and the public availability of comment text without authentication. The eight outlets selected — CNN, MS NOW, NBC, CBS, ABC, AP, Fox News, Forbes Breaking News — preserve cross-Part comparability and span the visible spectrum of U.S. mainstream news positioning: three centrist legacy networks (NBC, CBS, ABC), one wire service (AP), two left-leaning outlets (CNN, MS NOW), and two right-leaning outlets (Fox News, Forbes Breaking News). The 200-comments-per-outlet-per-phase target is calibrated to the corpus density observed in Parts I and III; beyond roughly the first 200 comments of a thread, the marginal register-content yield declines and the additional comments are dominated by reaction-emoji and short rejoinder content.

These choices carry limitations that should be flagged before the empirical sections. Formal inter-annotator reliability statistics across the team were not computed for Part IV and are flagged here as a methodological limitation; the Part V design will incorporate explicit inter-annotator reliability calculation per sub-section of Section 5.B. YouTube comment sections are not representative of antisemitic discourse on social media as a whole; the audience is self-selected to interact with mainstream-news content and differs from message-board, X/Twitter, or TikTok populations. The English-language restriction limits the corpus to one linguistic community of antisemitic discourse, and the registers documented in non-English contexts may differ from those reported here. The 200-comment threshold is a sampling-economy choice rather than a representativeness claim; on threads where the antisemitic register concentrates in the first 50 comments, the threshold captures the bulk of the register, whereas on threads where the register diffuses across the first 500, the threshold may understate it. These limitations bear on how the article’s findings should be read: as descriptive of this corpus and the conditions it samples, rather than as generalizations to the broader landscape of antisemitic online discourse.

Summary of Comment Patterns by Outlet

The corpus comprises 4,598 user comments across 23 selected threads in 8 outlets across three temporal phases. The per-thread density of antisemitism-coded content is reported in the table below; the three substantive findings that emerge from the cross-phase comparison are summarized in plain prose immediately after. The full quantitative apparatus — figures, regression statistics, two-stage determination evidence, per-phase compositional table — appears in Section 5.D Quantitative apparatus, after the qualitative chapter, for readers who want the technical grounding.

Cross-phase density

The two density columns report the same threads under two coding standards used throughout the AiRT series: the explicit count includes only comments that meet the DA Lexicon’s strict coding criteria on the comment’s surface text; the context-dependent count additionally includes comments whose antisemitic reading depends on the surrounding thread context (irony, indirect activation, reply-context disambiguation). The narrative analysis below uses the explicit count as the conservative reference figure; the context-dependent figures are reported for transparency.

What the corpus shows

The headline numbers, read naively, would mislead. On their own, the raw percentages suggest that antisemitism simply faded once Trump and Netanyahu appeared to split. Phase C — the post-signing period in which Trump publicly broke with Netanyahu — produces a lower peak density (19 % at AP T21) than Phase A (26 % at AP T6) or Phase B (33.5 % at AP T13). On its face, this looks like attenuation: the political conditions shifted, the antisemitic register quieted. That interpretation is incomplete, for two reasons. Attenuation is visible in the raw numbers; what cannot be inferred directly from those numbers is that the antisemitism register itself attenuated, because the framing-position mix of the news cycle also changed.

The first is that the three phases are not sampled at comparable video framings. Phase A and Phase B both contained “direct Netanyahu-focused” videos — the framing class that produces the corpus’s highest densities. The Phase C news cycle (June 17–21) did not produce videos in that class; coverage shifted to “Trump-vs-Netanyahu” or “Iran-deal-with-Netanyahu-minimal” framings, which produce lower densities by structural design across all three phases. The density numbers cannot be read as straightforward attenuation, because the news cycle’s framing mix also changed. (The full framing-position regression and its statistical apparatus are in Section 5.D.)

The second is the substantive Phase C finding, which is the article’s central result: what changed at Phase C is not the amount of antisemitism but the dominant sub-form within the subordination register. This compositional finding is methodologically distinct from the raw-density question disqualified above: the comparison here is within Cat 1.1 — base-form versus blackmail sub-form — rather than across thread-level antisemitism densities, so it is not subject to the same framing-position confound that disqualifies the cross-density comparison. The base “Trump-is-Bibi’s-puppet” register stays roughly flat per thread across all three phases. The proportion of comments expressing the blackmail and hidden-control sub-form rises markedly across the three phases — Mossad has Epstein files on Trump, “Trumpstein,” golem-Trump, “Bibi has Trump by the curlies,” Israel-first / Epstein-administration. This sub-form rises from approximately 1.2 instances per thread in Phase A to approximately 5.4 in Phase C, with Forbes T23 alone containing 13 instances in a single 200-comment sample. Its share of all subordination-register content rises from approximately 18 % in Phase A to approximately 57 % in Phase C: by the post-signing window, the blackmail sub-form produces more than half of all subordination-frame content. Under the changed political conditions — Trump signing a deal Israel publicly opposes, Vance rebuking the Israeli Cabinet on NBC — the within-frame composition shifts toward a sub-form that can absorb apparent counter-evidence. Among the three hypotheses set out in the Four-Study Arc section, the Phase C composition is most consistent with mutation over attenuation or stable persistence.

Two further cross-phase observations The outlet-audience composition produces large variation in density at the same framing-position throughout the corpus — Fox News and Forbes Breaking News stay quieter under the same framing than AP, ABC, and MS NOW. The methodological implication: no single “X % of comments are antisemitic” number is meaningfully interpretable for cross-period or cross-platform comparison without specifying outlet and framing. Separately, the religious antijudaism register (Christ-killer, Jews-as-usurers, Jews-threaten-Christians) and the Nazi analogy — concentrated at right-leaning outlets in Phase A — recede in Phase B and re-surface at higher cross-outlet concentration in Phase C, accompanied by new lexical-allusion vehicles entering the selected mainstream-news YouTube comment sections (yarmulke-coded allusions, neo-Nazi numeric codes in usernames, Holocaust-denial book recommendations).

The qualitative material that grounds these claims — the actual comment text, organized by register — is the subject of Section 5.B below, which is where the article’s evidentiary weight sits.

Breaking Down the Predominant Concepts of Antisemitism in YouTube Responses

The qualitative analysis of the 23-thread corpus organizes the findings into five sections aligned with the DA Lexicon’s analytical architecture: 1 Classic Antisemitism, 2 Concepts of Power, 5 Attacks on Israel’s Legitimacy, 6 Aggressive Speech Acts, and 2.1 Greed/Exploitation. The register coded as Lexicon 3 Secondary Antisemitism in the prior installments — the treatment of antisemitism accusations themselves as bad-faith manipulation deployed by Jewish actors to silence criticism, here termed antisemitism-instrumentalization — is incorporated under Section 5.B.2 (sub-section 2.7) because it applies the same deceit stereotype to claims of antisemitism. A separate Discourse Dynamics section, placed after the Quantitative Apparatus, treats the production-level patterns through which these registers circulate — same-speaker template deployment, cross-thread amplifier patterns, affirmation cascades — at a different analytical level than the content-register sections.

Empirical volume distribution The five Section 5.B sections are not equal in size; the separate Discourse Dynamics section is included in the final row for completeness. Approximate proportions across the corpus:

Relationship to the Part III five-category structure

Each section below is illustrated with representative items from across the three phases. The cross-phase trajectory of each unit is flagged where present. A separate Section 5.C addresses forms of rejection that are not antisemitism — the boundary cases that the conservative-coding standard excludes.

1. Power and Conspiracy Narratives

Power and Conspiracy Narratives constitute the dominant antisemitic register across all 23 threads of the Part IV corpus, accounting for roughly two-thirds of coded instances. As Part III documented in March 2026, when the geopolitical configuration places the United States in a posture of cooperation or co-action with Israel, the discourse that emerges in mainstream-news comment sections is overwhelmingly conspiratorial rather than religious or affective. Part IV asks whether this register weakens when the configuration changes. It does not. Its internal composition changes instead.

The central organizing formation across all three phase windows remains the subordination frame — discourse rendering the United States as Israel’s instrument, the U.S. president as Netanyahu’s functionary, and the U.S. military as a mercenary force deployed in Jewish interests. What distinguishes the present corpus from Part III is not the persistence of this frame but the shift in its internal composition. The base subordination idiom — the dog/master, puppet, and ownership metaphors that dominated Part III’s documentation — holds steady in per-thread frequency across all three phases (≈5.8 → 5.1 → 4.1 instances per thread), with only a modest decline. What grows most sharply is a related sub-form, blackmail and hidden-control, in which Trump’s compliance with Israeli interests is explained not by overt political alliance but by Mossad’s covert leverage. The sub-form’s share of total Power and Conspiracy content rises from approximately 18 % in Phase A to approximately 57 % in Phase C: by the post-signing window, it produces more than half of all subordination-frame content.

The register also contains two adjacent sub-registers operating within the same conspiracy logic: a False-Flag / JFK-Precedent formation positioning Trump or American interests as the target of Jewish or Israeli action rather than its agent, and a Jewish Elite Networks formation naming Rothschild finance, Chabad-Lubavitch leadership, and adjacent elite actors as controlling forces.

Subordination The base subordination idiom — Israel-controls-America in the dog/master, puppet, owns-the-party, dictate-US-policy, tail-wagging-the-dog form — is documented across all 23 threads. Nine representative instances, drawn from across the corpus to span outlet, phase, and metaphor sub-class:

“Bibi owns Trump it’s clear.” (CNN T1 #0015)

“Dump is Israel’s personal dog!” (CNN T1 #0047)

“BiBi has a short leash on Trump!” (Fox T7 #0009)

“Mileikowsky is Trump’s daddy.” (ABC T5 #027-17)

“He is Israels puppet. PUPPET PRESIDENT” (CBS T4 #0014-13)

“Israel is entirely in control of US domestic and foreign policy. As well as our military.” (CNN T1 #0035-23)

“AIPAC has captured America.” (ABC T5 #011; 204 likes)

“The biggest enemy for America is Netanyahu. He has dog walked America to 3 wars: One with Iraq, and twice with Iran.” (NBC T3 #041)

“Trump writes Netanyahu strongly worded check” (Forbes T23 #053)

NBC T3 #041 supplies the canonical compression — America as the dog being walked through three wars on a leash held by Netanyahu personally. The ABC T5 #011 “AIPAC has captured America” comment works the same Subordination register through a different mechanism: institutional proxy rather than personal submission, with AIPAC as the named conduit through which Jewish capture of US policy is alleged to operate. The Forbes T23 #053 “strongly worded check” comment is the corpus’s clearest ridicule-class instance: rather than asserting Israeli control through metaphor, the comment mocks Trump’s ineffectuality against Netanyahu via the “strongly worded letter” trope, with the check substitution converting Trump’s response into transparently inadequate payment. The mockery presupposes the subordination — only a subordinate would reduce its protest to a check — and works the register from the inverse angle. T6 AP shows the strongest single-thread base-form saturation in the corpus (~26 %) under the “Netanyahu will do whatever I want him to do” video framing — Trump’s May 20 assertion of dominance over Netanyahu. Rather than aligning with Trump’s framing, the comment section overwhelmingly reverses it — Netanyahu controls Trump, not the other way around — across six metaphor classes in a single 200-comment sample (master, puppet, canine, sexualized-submission, daddy, owned-and-controlled), including a multi-presidential-orchestration variant absent in Phases A T1 and T2. The thread is the Phase A precursor of the mutation pattern that develops at Phase C: even direct counter-evidence to the subordination claim is reframed as confirmation of it. The full inventory spans six metaphor sub-classes: canine, puppet, ownership, parent/child, ridicule, and sexual-domination.

Blackmail and Hidden Control The defining feature of the hidden-control sub-form is that observable political disagreement is reinterpreted as evidence of deeper covert coordination. The political proposition “Trump is influenced by Israel” is routine political analysis; what makes the Phase C material antisemitic is the mechanism speakers reach for to re-narrate the relationship under the changed political conditions:

“A fake feud. We wouldn’t be in Iran in the first place if Netanyahu didn’t own Trump.” (MS NOW T2 #0016)

“Trump is blackmailed by Israel over the Epstein files.” (Fox T22 #0012-1)

“Ireasl [sic] said to Trump do it, or we will release the sick PDF video of you” (Forbes T23 #0051-2)

“Donald Judas Trump eventually betrays everyone” (Forbes T23 #005)

“Golem Trump, turned on its creator” (Forbes T23 #006)

The MS NOW T2 #0016 “fake feud” comment is the article’s title quote and the corpus’s most compressed instance of the mutation: a real Trump–Netanyahu rupture — public criticism, Cabinet rebuke, Lebanon refusal — dissolved by declaring the rupture itself fake, with the hidden-control premise (Netanyahu owns Trump) preserved through the dissolution. The Fox T22 “blackmailed over the Epstein files” comment is the lexically explicit categorical attribution of the mechanism. The Forbes T23 PDF-video and the two adjacent Forbes T23 cultural-vocabulary comments — “Donald Judas Trump” and “Golem Trump, turned on its creator,” posted as #005 and #006 in the same thread — frame the Trump–Netanyahu rift through name-vocabulary drawn from religious antijudaism (Judas the betrayer-disciple; golem the creature-of-its-creator). In context, both comments contribute to the hidden-control narrative by treating Trump’s apparent break with Netanyahu as a betrayal or malfunction inside an assumed prior relationship of control — the Judas-betrayer trope frames the rupture as collaborator-betrayal internal to a Jewish-patron relationship, the Golem-turned-on-its-creator trope as creature-turn internal to a Jewish-creator relationship. Both convert apparent counter-evidence into further evidence of the framework’s core premise.

The blackmail and hidden-control sub-form appears as Phase A precursors at CNN T1, Fox T7, NBC T3, and Forbes T8; Phase B Trumpstein instances at ABC T12; and Phase C instances at Fox T22, CBS T19, and Forbes T23.

The sub-form grows from approximately 1.2 instances per thread in Phase A to approximately 5.4 in Phase C — a roughly 4.5× increase. The Trumpstein thread documented in Part III does not attenuate under policy reversal; its within-frame composition shifts toward the blackmail-leverage sub-form that can absorb the political contradiction. The headline share-shift (18 % → 57 %) is documented in the Section 5.D Phase C compositional table.

False-Flag and JFK-Precedent Conspiratorial framings positioning Trump or American interests as the target of Jewish or Israeli action — a related but inverse variant of the subordination frame, operating within the same conspiracy register. The Phase C T21 AP comment is the corpus’s lexically explicit realization:

“False flag attack by Israel incoming” (AP T21 #106)

The comment activates the broader register in which Israeli covert action is expected when U.S. policy diverges from Israeli interests. In this corpus, the JFK register functions as the paradigm case: a president who diverged from Israeli interests and was punished by Mossad-orchestrated assassination. Trump’s current divergence — signing a deal Israel publicly opposes — invites the inference that the same sequence is now in motion, with the false-flag prediction at T21 #106 operating as the anticipatory move that places the predicted operation in continuity with the JFK precedent. The register also operates in elaborated forms across the corpus, with the JFK and October-7 precedents made fully explicit:

“They’re going to false flag nuke an American city and blame it on Iran to drum up war support.” (ABC T5 #012-2)

“You mean the attack Israel knew about, had ties to, provided material support, and chose not to warn its own people” (ABC T20 #027-05)

“Only Kennedy did not cooperate with Jews and Israel, so they killed him.” (MS NOW T2 #0021-04)

The JFK precedent (MS NOW T2) and the October-7-as-false-flag attribution (ABC T20, framing the October 7 attack as one Israel “knew about, had ties to, provided material support”) share a structural feature with the T21 “false flag attack by Israel incoming” prediction: all three treat documented or anticipated violence against Jewish, Israeli, or American actors as evidence of Jewish-orchestrated cover, with the violence itself reframed as an Israeli operation. The corpus contains five instances of the register across three Phase C threads (T19, T21, T23) plus a Phase A precursor at T2.

Jewish Elite Networks The classical conspiracy register naming Jewish or Jewish-associated elite networks as the controlling force behind Israeli or U.S. policy. The register appears in three recurrent forms in this corpus: (1) historical-elite-conspiracy framings naming Rothschild finance and Chabad-Lubavitch leadership; (2) named-billionaire and institutional-listing framings naming Miriam Adelson, Larry Ellison, AIPAC, ADL, and adjacent entities; and (3) interlocutor-directed accusations that apply the same control-through-Jewish-money logic to comment-section opponents themselves.

“Isreal [sic] has been owned by the Rothschild’s since 1917 through the signing of the Balfour declaration” (AP T13 #0010-18)

“Kushner and Bibi are both beholden to Chabad Lubavitch.” (AP T6 #017-3)

The T13 AP Rothschilds-1917 comment is the most lexically explicit Rothschilds-controls-Israel claim in the corpus and supplies the Phase B anchor for this sub-register. T6 AP supplies the Phase A Chabad-Lubavitch sub-register, with the Phase C continuation at T23 Forbes. A more inferential variant of the register — in which the conspiratorial implication is carried by the structural juxtaposition of Jewish names without explicit “Jewish networks” naming — also appears in the Phase C corpus, notably CBS T19 #014-5 (juxtaposing Jeffrey Epstein and the current US Commerce Secretary as long-term neighbors).

The register also surfaces in named-billionaire and institutional-listing variants. A Phase A CNN T1 instance personalizes the framework through a single named Jewish billionaire framed as Trump’s financial controller:

“There’s NO way Trump is going to jeopardize his ATM Miriam Adelson. He knows if he let Israel fall that evil woman will shut off the money stream.” (CNN T1 #0035-46)

“ATM” reduces a named Jewish person to a financial instrument; “money stream” frames political support as transactional cash-flow; “evil woman” essentializes the personal-character predicate. The construction sits within the Jewish Elite Networks register — a named Jewish actor as the controlling force behind Trump — rather than within the cross-phase taxpayer-extraction register treated at Section 5.B.5. The same speaker carries the “USA keeps footing the bill” framing elsewhere in the same thread, producing both the institutional-extraction and the named-billionaire variants within a single thread of approximately 200 comments.

A Phase A T3 NBC instance supplies a compressed list of named Jewish-funding institutions and individuals:

“Pro Israel pro Trump rightwing Israel government lobbies like AIPAC, ADL, AI, Crypto, Miriam Adelson, Larry Ellison funded pro Israel corporate moderate Democrats” (NBC T3 #036)

Here the speaker invokes the institutional-Jewish-funding-controls-politics register by listing AIPAC (lobby), ADL (advocacy), AI (industry sector), crypto (industry sector), two named individuals (Adelson, Ellison), and naming the political-actor-category being controlled (corporate moderate Democrats). ADL appears here not in its institutional capacity — it is a civil rights organization, not a funding body — but as a Jewish-associated organization that the speaker recruits into the financial-control frame alongside actual lobbying and donor entities; the comment’s persuasive structure depends on the reader not distinguishing among the entities listed. The construction is methodologically distinctive — naming-density itself becomes the persuasive vehicle, on the implicit premise that the list’s length is its argument.

The register also operates at the level of discourse interlocutors rather than at the level of state policy, applying the same control-through-Jewish-money logic to comment-section opponents themselves. The “shekelstein” coinage at MS NOW T2 #0014-13 (“Says the shekelstein... you get those illegal shekels from tel aviv”) applies the Trumpstein “-stein” suffix (documented in the Blackmail and Hidden Control sub-section earlier) to the interlocutor and pairs it with a categorical Israeli-payment claim. “ZIONIST BOT” (Fox T22 #0004-3, recurrent label) functions in the same way through a different lexical vehicle. Both register-instances belong analytically with the Power and Conspiracy register — Jewish money as the imputed mechanism through which dissent is bought — rather than with the financial-exploitation stereotypes treated in Section 5.B.5.

2. Classic Antisemitism: Demonology, Dehumanization, and Character Stereotypes

Consolidates the Part III Demonology and Dehumanization material under the DA Lexicon’s 1 Classic Antisemitism architecture, with sub-sections following the Lexicon’s own sub-categories. The Terrorist-State and Genocide registers that were grouped under Part III 2 are reclassified under Attacks on Israel’s Legitimacy (Section 5.B.4) per the Lexicon 5 architecture, where they sit more naturally with Nazi Analogy and Legitimacy Denial.

Three cross-phase developments distinguish the Part IV corpus from Part III. First, dual-loyalty / Jews-belong-elsewhere exclusion appears at sufficient cross-thread density to warrant explicit treatment. Second, the religious antijudaism register (Christ-killer + Jews-as-usurers + Jews-threaten-Christians) resurges in concentrated form at Phase C. Third, the AS-instrumentalization register surfaces at sufficient cross-phase density to warrant explicit treatment at 2.7, classified under Lexicon 1 because it applies the same deceit stereotype to claims of antisemitism.

2.1 Evil

The sub-section corresponds to DA Lexicon 1.2 Evil/The Devil. Israel or Jews framed as ontologically evil. The sub-register operates both at the level of the state (Israel-as-categorical-evil) and at the level of Jews as a collective (Jews-as-categorical-agent of obstruction or threat), with the corpus instances concentrating at the state level, and the collective-Jews framing surfacing in the MS NOW T10 instance below:

“Israel is THE evil that brings death and destruction to west Asia. It must be stopped before it will bring the same to the rest of the world” (ABC T20 #0037)

“Israel will be what brings the world to its knees” (ABC T5 #008-03)

“a threat to everyone in the world really. Because Israel is” (Fox T22 #0008-20)

“biggest threat to humanity than anything Iran wants to do” (Forbes T23 #0013-14)

“the entire gdp is propped up by endless war and conflict and occupation their stock market booming because all the defense industries is just more proof that the country is done for the second they stop killing their neighbors” (CBS T19 #005-17)

“Israel will not stop. Israel can’t stop. Wake up people. There will be no ceasefire. For Israel has spoken” (Forbes T23 #0026)

“Jews will never allow peace” (MS NOW T10 #031)

Categorical attribution of evil to Israel — the shift from “Israel pursues X policy” to “Israel IS evil / threat” — is the coded element within this register when embedded in the broader antisemitic predicates documented across this and the surrounding sub-sections; bounded political criticism of specific actions sits outside coding (see Section 5.C). CBS T19 supplies the analytically richest evil-essentialization sub-variant of the register: Israel is framed as a state that cannot exist except by killing its neighbors — “the country is done for the second they stop.” Economic prosperity and ontological violence are made co-extensive. The Forbes T23 “Israel has spoken” sequence performs the operation through a different mechanism: the four-clause incantatory structure (”will not stop / can’t stop / no ceasefire / has spoken”) elevates a state to the grammatical position of a sovereign-religious-actor whose word is law, with violence as its constitutive expression. The MS NOW T10 #031 “Jews will never allow peace” instance is the corpus’s collective-Jews variant — categorical predicate attributed not to Israel-as-state but to Jews-as-collective — and also activates the obstruction/control register at Section 5.B.1; the comment is preserved under Lexicon 1.2 here because the collective-essentialization carries the predicate.

2.2 Repulsiveness and Dehumanization

The Lexicon 1.4 dehumanization register in this corpus surfaces through vermin/extermination-coded imagery and through the parasite-host trope, the latter straddling Lexicon 1.4 (vermin-register) and Section 5.B.5 (financial-extractor-coded) depending on imposed-character context:

“isreal is literally nothing but a cockaroach without American support” (Forbes T23 #0051-1)

“Their biggest problem are themselves. They don’t care about doing what God has commanded them. Having those filthy parades in Jerusalem, and then the rumors, making dogs rape captives. They are wicked! They should really believe in God, and in Jesus Christ the Messiah, so that they will be well. If they don’t, they will never be well.” (Forbes T23 #0031)

“Israel has no support anymore. They are only relevant bc they leach off the U.S.” (ABC T5 #016-2)

The T23 #0051-1 cockroach instance carries the classical vermin-trope: “cockroach” is a compact vermin-dehumanization label deployed against Israel-as-state, and its appearance in a mainstream-news YouTube comment section is one of the cross-platform migration markers documented in this study. The T23 #0031 instance compresses three Lexicon 1.4 sub-registers — moral-condemnation (”They are wicked”), bodily-repulsion (”filthy parades”), and animalized sexual-violence imagery (”making dogs rape captives”) — into a single comment, concentrating the dehumanization through accumulated lexical density rather than through a single sustained metaphor; the Christian-supremacist coda (”believe in God, and in Jesus Christ the Messiah”) activates the religious antijudaism register at 2.3. The ABC T5 #016-2 leach/parasite predicate operates in Lexicon 1.4 dehumanization-coded mode (parasitism as character-essentialization, vermin-coded register) and continues the Part III pattern: the Part III Zionist-rats and parasite-as-character registers both recur in the Part IV corpus. The same predicate also activates the financial-extraction trope treated at Section 5.B.5; the boundary placement here reflects the dehumanizing vermin-register lexicalization (”leech”) as the surface marker, with the financial relation framed as its content. The corpus’s most lexically explicit parasite-host instance — Forbes T23 #024-05 “Israel is a parasite, a parasite who has the audacity of thinking that the host is indebted to them” — operates in financial-extractor mode and is treated at Section 5.B.5 (Parasite-Host) accordingly.

2.3 Religious Antijudaism

The sub-section covers the Christ-killer, Jews-as-usurers, Jews-threaten-Christians, and goyim-as-cattle sub-registers. The term religious antijudaism is used here to refer to the Christian-theological anti-Jewish tradition — predominantly medieval-European in origin — that frames Jews as deicidal, usurious, and a threat to Christian civilization. The category sits within Lexicon 1.5 Immorality, which also covers modern character-morality attributions; in this corpus, the Lexicon 1.5 instances concentrate at the theological end of the spectrum. As with Part III’s Blood Libel sub-section, this material is a secondary Phase C finding rather than the primary one (the Cat 1.1 mutation treated in Section 5.B.1). It warrants explicit treatment because Phase C surfaces these registers in concentrated form for the first time across the four-study series, but the analytical center of gravity of the article remains the conspiracy mutation, not the religious antijudaism resurgence.

The T23 Forbes thread carries the most explicit single instance:

“WAKE UP CHRISTIANS YOU DISHONOR YOUR SAVIOR BY SUPPORTING ISRAEL. CHRIST WAS CRUSIFIED BECAUSE HE SPOKE OUT AGAINST THE JEWISH PRACTICES OF USERY AND PROFITEERING IN GODS NAME.” (Forbes T23 #0042)

The comment combines two foundational Christian anti-Jewish motifs: the deicide accusation and the usury stereotype.

The register is not a Phase C novelty. Three Phase A instances anchor it across both right-leaning and centrist outlets and across both Christian and Islamic religious framings:

“the most bitter enemies of God Almighty, they are the Jews, the killers of prophets and messengers, and those who distort the words of the Torah. They are the most bitter enemies of both Christianity and Islam.” (CNN T1 #0021)

“Why do you support murdering children? Are you a Ba’al worshipper?” (AP T6 #015-4)

“Read the TALMUD and see what they actually think of non-jews, and what they’re supposed to do to them. That is their book of laws. They have everybody brainwashed” (Forbes T8 #107-3)

The CNN T1 instance is the corpus’s lexically densest religious-antijudaism statement, compressing three registers into a single comment: killers-of-prophets (a deicide-adjacent variant that travels beyond Christianity), Torah-distortion, and a dual-religion enemy-framing — Jews as enemies of Christianity and Islam, both named. The comment validates the umbrella term religious antijudaism by surfacing the Islamic-religious variant of the frame alongside the Christian one. The AP T6 Ba’al-worshipper accusation operates at a centrist outlet through transferred-worship logic: support for Israel is reframed as idolatry, with the supporter assigned the false-god-worship predicate by association. The Forbes T8 Talmud-anti-gentile instance is the cross-phase predecessor of the T23 #116 goyim-as-cattle trope — both attribute to the Talmud the claim that Jewish religious teaching sanctions contempt for non-Jews, and both circulate as cross-platform migration markers from white-nationalist vocabulary into mainstream comment sections. The Forbes T8 thread also carries a god-of-Israel-as-Baal accusation at #084 (”The god of Israel is baal”) and an anti-Christian-Israel framing at #066-2 (”Israel killing christians in Lebanon... destroy statues of Jesus”), confirming the right-leaning Forbes concentration of the theological register in Phase A.

The Jews-threaten-Christians register concentrates in Phase C through a single speaker’s three-template deployment in T22:

“Ben Shapiro is a threat to American Christian peaceful values, USA would be better off if he fled to Tel Aviv instead of Tennessee” (Fox T22 #0004-3)

The same speaker posts two further variants in T22 — one substituting “Israel” for “Tel Aviv” and “Nashville” for “Tennessee,” a third with a “Zionist Bot” opener — both treated as instances of same-speaker template deployment in Section 5.E. The register also surfaces at T23 #0031 (”they don’t follow what God commanded them, they should believe in Jesus”), combining Christian-supremacist framing with the wicked-Jew character essentialization above.

The “goyim-as-cattle” trope — an antisemitic attribution to the Talmud holding that Jewish religious teaching itself sanctions contempt for non-Jews, a misrepresentation of certain Talmudic passages with long-standing currency in white-nationalist message-board vocabulary — surfaces in the Phase C corpus through ironic ventriloquism:

“Cloooowwn..mid term panadering to the goyims..headless cattle” (Forbes T23 #116)

Here the speaker attributes the statement to an imagined Jewish voice rather than asserting it directly, positioning themself as critically reporting on a presumed Jewish view of non-Jews (”goyims..headless cattle”). The discursive structure is parallel to the stereotype-confirmation register at CBS T19 #007-16: an antisemitic claim is asserted not as the speaker’s own belief but as an empirically verified description of Jewish thought. Both forms operate by reframing antisemitic content as observation rather than accusation. The trope’s appearance at a mainstream-news YouTube comment section is one of the cross-platform migration markers documented in this study.

Together with the Disloyalty and Non-Belonging register at 2.5 and the eliminationist 109/110-countries variant at 2.6, the religious antijudaism resurgence at Phase C constitutes a complete classical Christian-anti-Jewish framework: Jews threaten Christians + Jews betray Christ + Jews don’t belong here + Jews see non-Jews as cattle. Whether this resurgence is specific to the Trump–Iran deal context or reflects a broader pattern is a question for subsequent studies.

2.4 Lie and Deceit

“Lol if only Israel didnt lie to Trump had him thinking he could take Iran like Venezuela” (ABC T20 #0042-4)

“Netanyahu sabotaging the person who started a war for him is the most Israel-like thing I’ve seen” (CBS T19 #007, 125 likes)

“Israelsucker. Ben Shitpiro is lying...AGAIN. If the US isn’t bending over backwards for Israel, the tiny hats claim its failure. Its good for everyone, but Israel is a untrustworthy lying LACKEY and they should be treated as such” (Fox T22 #0008-12)

“😂😂 they love to play into the stereotypes” (CBS T19 #007-16)

The essentialization is the coded element: deceitfulness is framed as essentially-Israeli, not as Netanyahu’s tactical choice. The comparative “most Israel-like thing I’ve seen” presupposes a baseline of Israel-typical deceit, and 125 likes on the CBS T19 comment — second-highest engagement in the thread — mark the formulation as canonical rather than marginal. The Fox T22 comment fuses three antisemitic moves in a single sentence: “untrustworthy lying LACKEY” essentializes deceit as Israeli character; “tiny hats” supplies a yarmulke-coded reference to Jews (cross-outlet companion to CBS T19’s “small hats”); and “bending over backwards for Israel” supplies the power-control framing. The reply “they love to play into the stereotypes” closes the loop — an antisemitic stereotype is treated as empirically confirmed by Israeli behavior. The distinctive move here is not the deployment of a stereotype but the citation of one as vindicated. In this thread context, the ratified stereotype reads as deceit — that is what the news story is about — but the wording “stereotypes” is open: the same construction can activate any of the classical AS character-stereotypes (greed, dual-loyalty, control) depending on framing. The same register surfaces cross-thread at MS NOW T17 (”reinforce all the stereotypes they insist are false”), confirming the move’s stability.

2.5 Disloyalty and Non-Belonging

“we need to distance ourselves from that dump. All the special people here need to go over there and fight, if it is something they believe in so much” (Forbes T23 #011)

“Shapiro is a traitor who isn’t even Israel first, he is Israel ONLY” (Fox T22 #0004)

“USA would be better off if [Ben Shapiro] fled to Tel Aviv instead of Tennessee” (Fox T22 #0004-3) [keving5711 template, deployed 3× in T22]

“Special people” in the Forbes T23 #011 instance functions as a coded reference to American Jews; the imperative to “go over there and fight” activates the classic dual-loyalty exclusion frame — Jews-belong-elsewhere coupled with the demand that Jewish Americans prove their commitment by relocating to Israel. In this corpus the Lexicon 1.8 register concentrates around two foci — Ben Shapiro as the named-public-figure case and the Forbes T23 “special people” framing — rather than appearing as a broad cross-outlet pattern.

Dual-loyalty assigned to Jewish public figures is the Lexicon 1.8 register treated here; the discursive complement — Hasbara-accusation against discourse interlocutors — sits under the control-through-money logic at Section 5.B.1 (Hasbara Accusation and the Shekel-Payment Slur).

2.6 Blame for Antisemitism

Antisemitism framed as the natural consequence of essentialized Jewish or Israeli action — antisemitism is Jews’ own fault — paired with the complementary register in which Jews are framed as falsely claiming victimhood to evade responsibility for their own behavior:

“’It’s never our own fault’ -small hats” (CBS T19 #012-1)

“that’s what happens when you steal peoples land and terrorize the native population” (CBS T19 #027-20)

The CBS T19 #012-1 “small hats” comment supplies the corpus’s clearest Lexicon 1.9 instance: “small hats” is a coded reference to kippot or yarmulkes — i.e., to Jews — that has migrated from white-nationalist message-board vocabulary into mainstream YouTube comment sections, and the construction ironically attributes to “small hats” the self-exculpating utterance “It’s never our own fault,” with the irony inverting the attribution: the self-exculpation is framed as the lie. The key discursive move is the reframing of Jewish victimhood as manipulative self-exculpation. The CBS T19 #027-20 “steal peoples land” comment carries the paired deserves-the-hate move within the same thread context — antisemitism reframed as deserved consequence of essentialized Israeli action; the comment falls within Lexicon 1.9 because of the surrounding thread’s broader essentialization of Israel rather than the bounded political claim about land alone. Both moves sit within the same Lexicon 1.9 category: the first reframes Jewish victimhood claims as manipulative, the second converts antisemitism into deserved consequence. The same logic overlaps with the AS-instrumentalization register at 2.7, where the same stereotype operates through accusations of Jewish bad-faith deployment of victimhood claims.

The 109/110-countries variant. The variant is not merely legitimacy denial; it explains persecution as deserved. The most heavily-loaded form of the Blame-for-Antisemitism register in the corpus — the antisemitic claim that Jews have been expelled from 109 countries, combined with an invitation for the US, or in T19’s variant for Israel itself, to join the expulsion list:

“America becomes the 110th” (NBC T3 #073 — Phase A first instance)

“So Israel attacked itself to keep the war going.. Big surprise. Let’s make it 110 countries” (CBS T19 #012, 70 likes)

The Blame-for-AS premise (Lexicon 1.9) is the coded element: the historical claim that Jews have been expelled from 109 countries frames the recurrence of Jewish persecution as Jews’ own deserved consequence — if 109 polities expelled the same population, the speaker invites the inference that the cause is inherent to the expelled. The Phase A NBC T3 instance applies the trope to the United States as the next country to expel its Jews; the Phase C CBS T19 instance applies the same arithmetic to Israel itself, framing Israel-as-state as the 110th eliminationist target. Both versions are coded at the highest severity level. The Phase C variant is more pointed because it pairs the expulsion-arithmetic with a false-flag accusation in the same comment (”Israel attacked itself”): Israel is essentialized as a state that fabricates its own attacks, then proposed as the next entry in the expulsion list. The “make it 110” move converts the Blame-for-AS premise into a call to action that crosses into Section 5.B.4 (4.4 Legitimacy and Indigeneity Denial) and Section 5.B.3 (Aggressive Speech Acts), but the primary code remains Lexicon 1.9 — without the sole-blame premise, the call would not carry its eliminationist force.

2.7 Instrumentalization of Antisemitism

The register is a specific application of the deceit-stereotype documented at 2.4: speakers position themselves as critics of an allegedly improper use of antisemitism-as-shield, framing Jewish actors’ invocation of antisemitism as bad-faith manipulation. It is treated within Lexicon 1 Classic Antisemitism because it applies the same deceit stereotype to the antisemitism-accusation itself. The placement is a classificatory adjustment for Part IV — reflecting the empirical fact that the corpus’s instances operate through the deceit-stereotype mechanism — rather than a rejection of the Lexicon 3 secondary-antisemitism reading carried in the prior installments.

“ungrateful Zionists… using the excuse of being in danger of survival to do the killings” (Forbes T23 #0039)

“ahh yes, the victim excuse for bad behavior. They’ve cried that tired script to the world for a few thousand years too many” (ABC T20 #0026-8)

“What will those POOR Professional Victims Here in the US and those in Isreal [sic] do without MAGAS TEMU president” (Forbes T23 #017)

The strategy frames antisemitism itself as an instrument Jewish actors deploy in bad faith to silence criticism — part of the antisemitic registers, not a counter to them. The Phase C ABC T20 variant adds historical essentialization — 'few thousand years too many' — that converts the antisemitism-shield critique into a victim-blaming claim with millennia-scale duration. The Forbes T23 “Professional Victims” instance applies the same predicate to U.S.-and-Israeli Jews collectively, framing organized Jewish self-defense as a professional category. (Note: the surrounding “TEMU president” phrasing — TEMU being the Chinese discount-retail platform known for cheap knockoffs — is anti-Trump rather than antisemitic; the coded register is the “Professional Victims” predicate.) The ABC T20 speaker also surfaces at Phase B MS NOW T17 carrying the same stereotype-reinforce register, an instance of the cross-thread author pattern documented in Section 5.E.

3. Aggressive Speech Acts: Schadenfreude, Death Wishes, and Glorification of Violence

The Aggressive Speech Acts register in this corpus concentrates in two distinct conditions. The first is the perceived Israeli defeat at the immediate post-signing window (Phase C T22 Fox, T21 AP), where speakers celebrate Trump’s apparent break with Netanyahu as Israeli defeat and call for that defeat to be extended. The second is the Phase B density peak at T13 AP, where the deal-announcement itself generates the register alongside the broader Power and Conspiracy register. The cross-phase pattern carries forward Part III 3 in conceptual treatment and expands it with the Phase C Schadenfreude concentration absent in the prior installments.

The three sub-registers below operate at different temporal positions relative to the act of harm — Schadenfreude celebrates a loss that has occurred, Death Wishes call for one that has not, and the boundary between them is structurally important because it tracks the target of aggression. The Phase A and Phase B instances concentrate on Israel-as-state (a defeat in war, the loss of a particular leader, the collapse of a policy); the Phase C instances expand the target to Israel-as-collective-existence (”Erase Israel off the map”) and, at the eliminationist edge, to Jews-as-population (the 109/110-countries register treated at Section 5.B.2 2.6). The examples below suggest a cross-phase target expansion: from defeat or loss, toward eliminationist fantasies of Israel’s collective existence.

Schadenfreude

Distinct from death-wish: schadenfreude celebrates an already-occurring loss, while the death-wish register calls for a future one:

“HERES ISRAEL’S NEW FLAG. 🏳️” (Fox T22 #0015-04)

The T22 Fox News white-flag comment is the most schematically compressed instance in the corpus: a single image — Israel’s new flag — and a white-flag emoji do the entire framing work. The username under which the comment is posted contains “1488,” the explicit neo-Nazi numeric self-identification (14 = the fourteen-word slogan; 88 = HH, Heil Hitler), pairing the schadenfreude register with overt extremist signaling. The mainstream-news YouTube comment section here carries content originating in white-nationalist message-board vocabulary.

Death Wishes

Calls for future destruction of Israel or Israelis. Phase A T2 supplies the first corpus instance — a mutual-destruction wish framed as gratification:

“Wouldn’t it be absolutely delicious if Israel and Murka went to war - against each other - and ended up destroying each other.” (MS NOW T2 #0010)

“Murka” is a mocking phonetic rendering of “America” — dismissive where “’Merica” is affectionate; the construction frames mutual annihilation between Israel and the United States as a desired outcome. Phase B at CNN T9:

“If America wants to end hostilities for good and save some taxpayer money they should bomb Israel!” (CNN T9 #0005)

Phase C contains the most explicit eliminationist instances:

“Erase Israel off the map 🎉🎉🎉🎉” (AP T21 #049)

“Zoomers will nuke Israel” (AP T21 #013-3)

These extend the Part III “I wish Israel to turn to glass” formulation. Two cross-phase shifts are visible. First, the target expands: Phase A’s MS NOW T2 instance imagines reciprocal destruction of Israel and the United States, retaining symmetry between belligerents; the Phase B CNN instance directs U.S. weapons at Israel; the Phase C AP instances move to elimination of Israel as a state (”erase…off the map”) with no reciprocal harm to the U.S. proposed. Second, the affect-marking shifts: Phase A frames the death-wish as gratification (”absolutely delicious”); Phase C pairs it with celebration emojis (🎉🎉🎉🎉), fusing the call for destruction with schadenfreude-style affirmation. The celebration-emoji modifier converts what is structurally a future-oriented wish into something closer to a present-tense cheer — collapsing the temporal distance between Schadenfreude and Death Wishes that the sub-section headers preserve. At Phase C the two registers begin to behave as one.

The glorification-of-violence sub-register sits at the boundary with death-wishes — Forbes T23 #0007 “Only military force will stop these Israeli war criminals” frames future force against Israelis as the necessary instrument rather than as a wish-fulfillment, paired with the T8 Forbes Phase A nuclear-destruction-affirmation pair (#028 + #054-6) — and is treated here as a variant of the death-wish register operating at one register-step further toward instrumentalization. The instrumentalization move is the third position on the same target-expansion vector: from celebration of a past loss, to wish for a future one, to designation of the means by which the future loss should be brought about.

4. Attacks on Israel’s Legitimacy

Legitimacy attacks remained a stable component of the corpus but shifted internally across phases. Nazi analogies concentrated in Phase A under Israel-strikes-civilians framings; legitimacy-denial expanded in Phase C through authenticity, expansionism, and right-to-exist registers. This section consolidates the Part III 4 Nazi Analogy category with the cross-phase legitimacy-denial registers documented across the Part IV corpus, under the DA Lexicon’s 5 Attacks on Israel’s Legitimacy architecture. The four sub-sections correspond to Lexicon 5.1 (Nazi Analogy), 5.4 (Terrorist State), 5.5 (Genocide), and 5.7/5.9 (Denial of Right to Exist + Sole Guilt). The Terrorist-State and Genocide sub-sections move here from their previous treatment under Part III 2 Demonology: both registers operate at Lexicon 5 (Israel-as-illegitimate-state) rather than at Lexicon 1 (Jews-as-ontologically-evil), and the reclassification reflects the analytical distinction between essentialization-of-Jews-as-people and essentialization-of-Israel-as-state.

4.1 Nazi Analogy

The sub-section carries forward Part III 4 unchanged in conceptual treatment. Israel-as-Nazi, Netanyahu-as-Hitler, Holocaust inversion. Analytically distinct from religious antijudaism (Section 5.B.2, 2.3 Religious Antijudaism) — operates by reversing Holocaust victim/perpetrator roles rather than by invoking medieval Christian theology — and treated as Cat 4 in the DA Lexicon, its own category.

Phase A surfaces the register most elaborately at T1 CNN and T8 Forbes:

“Two decades ago the leader of Jewish fighters in the Warsaw Ghetto wrote a letter to Palestinian fighters and activists supporting their fight against oppression — a fight similar to the one Jewish fighters in the Warsaw Ghetto fought against imperialist Nazi invaders/occupiers/oppressors.” (CNN T1 #0035-10)

“Israel’s casus belli is Germany’s. Their methods of waging war is identical” (Forbes T8 #035-5)

The Warsaw Ghetto reversal performs the inversion at maximum historical resonance: Jewish resistance fighters from the most iconic site of anti-Nazi resistance are appropriated to authorize the framing of Israelis as Nazi-equivalent occupiers, with a real or claimed letter invoked to confer historical authority on the inversion rather than asserting the inversion directly. The Forbes T8 instance performs the equation more directly. T7 Fox supplies the broader Phase A right-leaning Cat 4 grouping; T8 Forbes alone supplies five Cat 4 instances in a single thread, consistent with the higher Nazi-analogy concentration at right-leaning outlets observed in Part III. Phase B carries the register in moderate concentration; Phase C does not produce a comparable concentration. The multi-tyrant equivalence at T23 Forbes (Mao, Stalin, Idi Amin) operates as broad authoritarian-lineage delegitimization rather than as Holocaust inversion, and is treated in Section 5.B.4 4.4 (Legitimacy and Indigeneity Denial) where its Mileikowsky-anchored delegitimization frame belongs.

The cross-phase distribution suggests the Nazi analogy attaches to moments of perceived Israeli moral failure (Phase A: Israel-strikes-civilians-in-Lebanon framing) more than to the moment of U.S.–Israel friction itself. Phase B’s relative rarity of the register and Phase C’s absence of a concentration-anchoring instance support this reading: the deal-period rupture itself does not activate the inversion frame.

4.2 Terrorist State

The Terrorist-State register attributes to Israel as a whole the ontological status of a terrorist entity — distinct from the allegation that specific Israeli operations constitute terrorism.

“Israel is the gold standard if terrorism...they literally invented it” (Fox T7 #0117)

“Israel is a far right terrorist regime no different than Iran. They’ve killed tens of thousands of their own people and are now invading a sovereign nation in Lebanon.” (CNN T1 #0046)

Categorical attribution of terror-state status to Israel as a whole. The two Phase A instances operate through different rhetorical moves: Fox T7 #0117 makes the origin-essentialization move — Israel as the historical author of terrorism as a concept (”they literally invented it”) — while CNN T1 #0046 makes the categorical-equation move that places Israel within the same ontological category as the U.S. State Department’s designated state sponsor of terrorism (”no different than Iran”). The two registers regularly co-occur with the ontological-evil register treated in Section 5.B.2 (2.1 Evil) and contribute jointly to the Section 5.B thread densities reported in Figure D4.

4.3 Genocide

“We pay for Israel’s weapons to commit genocide” (Forbes T23 #0029-1)

The corpus contains fewer instances of the categorical Genocide register than of the Nazi Analogy or Legitimacy Denial sub-sections; the analysis below treats the most lexically explicit instance and the coding boundary that distinguishes the antisemitic predicate from policy critique. What makes the move antisemitic, rather than policy critique, is the categorical predicate — Israel-as-genocide-state rather than Israeli-policy-X-causes-civilian-harm. The Forbes instance does this directly: Israel is the entity committing genocide, and the construction places U.S. funding in grammatical proximity to the genocide predicate (”we pay for Israel’s weapons to commit”). The cross-register coupling — Genocide predicate plus Cat 5 financial-extraction — supplies the essentializing force that elevates the term above policy-critique. Genocide allegations are not coded as antisemitic merely because they use the term; they enter coding only when embedded in categorical essentialization, sole-guilt attribution, demonology, or another antisemitic predicate. A comment that argues Israeli military conduct in Lebanon amounts to genocide on the basis of specific civilian-casualty figures, international-law definitions, or documented policy, and that does not impose an essentializing character claim on Israel or Jews, falls outside coding (see Section 5.C). The boundary is methodologically important.

4.4 Legitimacy and Indigeneity Denial

The common denominator across this sub-section is not that all comments deny Israel’s existence in the same way, but that each weakens Israel’s claim to sovereign legitimacy: by portraying it as inherently cursed, expansionist, inauthentic, non-indigenous, or too recent to count as fully sovereign. The four sub-registers below — Right-to-Exist Denial, Greater Israel / Expansionist Conspiracy, Authenticity and Indigeneity Denial, and Temporal-Recency Delegitimization — each carry one of these moves.

Right-to-Exist Denial

“Israel having nuclear weapons is disastrous. Infact the country itself is a curse to humanity” (MS NOW T2 #0007-02)

Greater Israel / Expansionist Conspiracy

“Israel using terrorism as excuse to grab more and more land... Greater Israel project. Syria. Lebanon. Palestine” (Forbes T15 #004-08)

Phase A T3 NBC first surfaces this sub-register at four instances in a single thread; recurs at T13 AP (Greater-Israel + Chabad-Trump + Euphrates/Nile flag-conspiracy bundle) and T15 Forbes (BDS-context). The “Nile to the Euphrates” framing continues Part III’s documented instance at Forbes (”they want the Nile to the Euphrates”), now elaborated across phases as a delegitimization register in its own right.

Authenticity and Indigeneity Denial

“BiBi is American he is not from Israel not even his real name” (CNN T1 #0019-2)

“Mileikowsky is Trump’s daddy” (ABC T5 #027-17)

“He sits alongside Mao, Idi Amin, Joseph Stalin and Benjamin Mileikowsky the Polish citizen who now rules Israel” (Forbes T23 #0015-24)

T23 Forbes uses Netanyahu’s original Polish family name to frame Netanyahu as a non-Israeli imposter ruling Israel, with the surrounding authoritarian-lineage construction (Mao, Stalin, Idi Amin) placing Netanyahu within a broader category of illegitimate rulers. The Mileikowsky move reactivates the Netanyahu-impersonation and Netanyahu-body-double formations documented at T9 and T13 in Phase B. T5 ABC Phase A supplies the earliest instance of the Mileikowsky pre-Hebraization name-recovery variant — deploying an Ashkenazi origin-name as a marker of non-indigenous Jewish identity, which then grounds the delegitimization of Jewish sovereignty. This sub-register also draws on the discredited Khazar theory documented in Part III at AP (”KHAZARIAN MAFIA BIG BO$$ BATTLE”) — both perform the move of denying Jewish indigeneity to delegitimize Jewish sovereignty.

Temporal-Recency Delegitimization

Phase C T23 Forbes surfaces a sub-register that operates not by denying Jewish historical presence (the Khazar move) but by framing Israel’s 1948 statehood as too recent to ground full sovereign standing:

“1948 isn’t that long ago. Someone just got too big for their britches” (Forbes T23 #038-1)

The argument structure — Israel-only-since-1948, therefore Israel is not yet entitled to act as a full sovereign — is paired with the idiomatic “too big for their britches,” which frames the state as an arrogant child overstepping its position. The combined move converts a contested historical claim about state legitimacy into a character description: Israel as a recently-formed entity behaving above its station. The register sits at the boundary with Section 5.B.2 character-essentialization and is coded here for its primary delegitimization work.

5. Classic Financial and Resource Exploitation Tropes

Unlike every other major register documented in Part IV, financial stereotypes exhibit structural stability across the three phases. The conspiracy register mutates substantially after the Trump–Netanyahu rupture (see Section 5.B.1); the financial register remains stable, reproducing longstanding antisemitic narratives of extraction, dependency, and parasitism with little internal innovation. The persistence is itself the finding: Classic Financial density sits roughly flat across phases (approximately 1.4 instances per thread in Phase A, 1.4 in Phase B, and 1.5 in Phase C, per Figure D4) while every other register category changes in density or internal composition under the changed political conditions.

The material moves between two levels of abstraction. The first — Financial Extraction — explains political relations through a moral economy of one-sided dependence: taxpayer-as-host, Israeli action as the perpetually extractive agent, Israeli theft as ongoing and multi-dimensional. The second — Greed and Parasitism — transforms that relationship into a Jewish or Israeli character trait, essentializing Israel or Jews as constitutively greedy, parasitic, or oriented toward consumption of others. Two cross-section boundaries apply throughout: comments framing Jewish money or Jewish-elite networks as the mechanism of U.S.-policy capture are treated under Power and Conspiracy at Section 5.B.1 (Jewish Elite Networks), and the parasite-as-dehumanization variant is treated under Section 5.B.2.

5.1 Financial Extraction

The extraction register constructs a moral economy: America produces, Israel consumes; America sacrifices, Israel profits; America is the host, Israel is the dependent. The register is not bounded by any particular budget item, allocation, or appropriation. It recodes the entire fiscal relationship as essentialized one-way drain. The categorical framing — Israel as the perpetually extractive agent, the American taxpayer as the perpetually drained host — is what makes the construction antisemitic rather than ordinary fiscal critique. The corpus’s most cross-phase-stable sub-register names the American taxpayer as the financial host of Israeli action:

“Netanyahu will not allow peace to happen. USA keep footing the bill.” (CNN T1 #0035)

“All paid for with our tax dollars. Don’t be a sheep your whole life.” (CNN T1 #0015-22)

“The bill for US taxpayers is $2 to $3 billion dollars per day for war, but $0 dollars are available for tuition-free college in America.” (CNN T1 #0035-45)

“Israel still has nuclear weapons and continues sucks [sic] our tax dollars. That is a daily loss.” (NBC T16 #027-1)

“Cut off Israel we don’t want any relationship with them give us our tax money back” (Forbes T15 #036)

A Phase B Forbes T15 variant extends the same categorical framing to perpetuity:

“Israel will never be able to repay us for their forever holy war that they constantly drag us into” (Forbes T15 #036-03)

“Forever holy war” recodes the financial relationship as a religious-historical predicate (Israel’s wars are essentially perpetual, essentially theological), and “never be able to repay” forecloses the possibility of resolution. The operative move is not the financial imbalance itself but its representation as permanent and inherent: in this construction Israel cannot stop extracting and cannot reciprocate, because extraction is presented as Israel’s permanent defining relationship to the United States.

The extraction logic extends from financial to multi-dimensional when the same categorical structure is applied across domains simultaneously:

“how much more land lives and money must they steal” (MS NOW T2 #0025-2)

The three-term enumeration — land, lives, money — collapses territorial, human, and financial harm into a single categorical predicate of Israeli essential character. The “how much more” framing presupposes the extraction has been ongoing and quantifiable, grounding the categorical claim in an implied empirical baseline. The triadic structure also merges three otherwise distinct registers into a single essentializing predicate: the financial term (”money”) is flanked and authorized by the more affect-laden terms (”land,” “lives”), so that the financial-extraction claim acquires the moral weight of the territorial and human-harm claims it travels with.

5.2 Greed and Parasitism

Where Financial Extraction frames a relationship, the character-stereotype register frames a property: Israel as essentially greedy, essentially parasitic, essentially constituted by consumption of others. The Part III “bleed you of resources to the tune of billions annually” and “stir up forever wars for your children” formulations recur as continuing cross-phase registers, and the comparative-affluence frame appears across multiple outlets in Part IV with little variation in the underlying move from its Part III instances:

“We’re just Israel’s final round of Squid Game” (CNN T9 #0003-22)

“Isreal is A very wealthy country the American people struggle to even buy groceries” (Forbes T23 #0029)

The Squid Game framing positions American taxpayers as expendable contestants playing for Israel’s benefit; the wealthy-vs-struggling framing performs the classical greed-attribution by contrasting Israeli affluence with American hardship as if the two were causally linked. Both share the structural move that distinguishes the character-stereotype from the extraction register proper: the comment does not describe an action Israel performs but a quality Israel possesses.

The parasite-host register supplies the corpus’s most fully elaborated character-stereotype:

“Israel is a parasite, a parasite who has the audacity of thinking that the host is indebted to them” (Forbes T23 #024-05)

The trope unifies three operations the surrounding registers perform separately. The parasite feeds on the host (extraction); the parasite cannot survive without the host (dependency); the parasite demands gratitude from the host (inversion) — with the audacity-modifier marking the inversion as itself a further offense. The repetition (”a parasite, a parasite”) shifts the figure from analogy to constitutive identity, completing the logic of the moral economy of Section 5.1: the host has obligations to the parasite, the parasite has none to the host, and the parasite demands gratitude on top.

The parasite metaphor straddles the boundary between Section 5.B.2 (where it functions dehumanizingly, as in Part III’s pairing of “parasites” with “rats”) and Section 5.B.5 (where it functions as a financial-extraction stereotype). The boundary is determined by the surrounding content: where the imposed character is vermin-coded, the instance enters Section 5.B.2; where the imposed character is financial-extractor-coded, the instance enters Section 5.B.5. T23 #024-05 reads as financial-extractor and is treated here.

Forms of Rejection That Are Not Antisemitism

A substantial share of comments rejected U.S. support for Israel, criticized Netanyahu, or advocated ending the alliance without recruiting antisemitic discourse. Because the present study examines antisemitic discourse rather than anti-Israel sentiment per se, these comments were excluded. In several cases, individual lexical items or formulations closely resembled those found in coded antisemitic discourse. In isolation, these formulations might have suggested an antisemitic reading; in context, the comments lacked the essentializing, conspiratorial, or delegitimizing predicates required for coding. The distinction is methodologically important: coding is driven by the overall discursive construction of the statement rather than by individual words or by the political position expressed.

The distinction is not always clear at the level of individual comments. Three recurring threshold patterns in the corpus warrant explicit discussion.

Political criticism and legitimacy denial. The first concerns the line between political rejection of U.S.–Israel alignment and delegitimization of Israel as a state. The ABC T20 corpus contains the following America-First cut-ties comment, coded at Section 5.B.4 4.4 (Legitimacy and Indigeneity Denial) and Section 5.B.1 (Subordination):

“Tucker said it best. Israel is a Junior Partner while Iran is a real country” (ABC T20 #0008)

The comment illustrates how an apparently boilerplate cut-ties citation can carry coded register without using any of the standard antisemitism vocabulary. The predicate “Iran is a real country” presupposes that Israel is not a real country — the comparative structure makes the delegitimizing claim only by implication. “Junior Partner” performs the subordination move from the inverse angle: Israel as a state lacking standing or sovereignty — a framing with antecedents in U.S. foreign-policy discourse, where Israel has long been characterized variously as a strategic asset or client state of the U.S. in the region. The comment is included in coding not because it advocates ending the alliance — that position is bounded — but because the framing of why recruits a denial-of-Israeli-legitimacy register and a subordination register in the same sentence. The coding decision turns on whether the speaker frames Israel as politically inadvisable to support (bounded) or as denying Israel’s legitimacy as a sovereign state (coded).

Personification metaphors. The second concerns personification metaphors applied to Israel within cut-ties discourse. Personification metaphors constituted a recurring source of coding uncertainty in this corpus; the following cases are preserved here for transparency about the boundary-coding decisions.

“Anyways you supporting isreal is like a woman laying down with addicts” (ABC T20 #0026-5)

“Israel is the girlfriend who walks into a bar, picks a fight with the locals, and then expect her boyfriend to come and save her” (ABC T20 #091)

“Israel-as-addict” imposes a character profile of Israel as drug-dependent, unreliable, and draining of its supporters — a personification that activates immorality and unreliability registers against personified Israel. The “girlfriend-who-walks-into-a-bar” template casts Israel as habitually quarrelsome, hypocritical (picks the fight then demands rescue), and parasitically dependent on the U.S. boyfriend — activating character-essentialization and dependency registers. The mechanism is the same as the Tucker-citation Junior-Partner case above: ostensibly bounded framing recruits an essentializing portrait to do the explanatory work. The threshold question for each individual case is whether the imposed character is a generic political-relationship characterization (bounded) or a categorical claim about what Israel is (coded).

The personification register is itself further bisected by a verbal-framing distinction that produces opposite coding outcomes. The girlfriend and addict instances impose character-by-nature predicates on Israel — Israel is habitually quarrelsome and demanding, is essentially dependent and draining — and are coded. The following CNN T9 instance reads differently and falls on the non-coded side of the boundary:

“Israel is the drunk friend that talks shit & drags you into bar fights repeatedly & then runs out the back door while you take the punches. At a certain point you realize your friend is the problem” (CNN T9 #0006-20)

The structure is moment-of-recognition rather than essentialization. “At a certain point you realize” is the operative verbal frame: the speaker narrates their own change of view about a specific relationship rather than asserting a timeless character predicate. The bar-fight content is concrete and situational, naming a pattern observable in the current U.S.-Israel configuration rather than a metaphysical claim about Israel-as-state. The construction is also non-generalizing — “you realize your friend is the problem” identifies a particular relationship that has gone wrong, not a categorical attribute Israel possesses prior to and apart from the relationship. The contrast with the T20 girlfriend and addict instances is the contrast between specific-pattern recognition and categorical character claim, and T9 #0006-20 is preserved on the non-coded side of that boundary.

Targeted personal criticism of Netanyahu. The third concerns criticism directed at Netanyahu as a specific named political figure — his policy advocacy, his record, his manner — rather than at Israel as a state or Jews as a collective. The corpus contains many comments that criticize Netanyahu sharply without crossing into coded territory; the boundary turns on whether the target of the predicate is the individual political actor or a collective referent his name is being used to reach. Three instances illustrate the non-coded side.

“Netanyahu has been venomous for 30 years. Finally, his only ally has had enough” (Forbes T23 #0037-1)

The framing is transactional. “Venomous” attaches to Netanyahu as a personal-character predicate but stays at the level of the individual politician’s conduct over his political career; “his only ally has had enough” describes the state of one specific alliance after a long pattern of strain. There is no categorical predicate about Israel as a state, no collective claim about Jews, and no orchestration mechanism. The comment is a bounded cut-ties characterization of a particular political relationship between a U.S. administration and an individual Israeli prime minister.

“Netanyahu has been trying to convince American presidents to destroy Iran for him for the last 20 years. He just found one dumb enough to actually do it in Trump” (CBS T11 #0011-16)

The comment describes Netanyahu’s twenty-year advocacy for confrontation with Iran — a documented feature of his political record — and attributes the deal’s outcome to Trump’s particular decision-making. The framing is biographical-political: Netanyahu has, in fact, argued for U.S. military action against Iran across multiple administrations, and the comment names this as the kind of pressure an allied leader exerts through ordinary diplomatic channels. The threshold here is not the twenty-year claim itself, which is historically grounded, but whether the speaker frames Netanyahu’s advocacy as bounded political pressure or as evidence of a covert multi-presidential orchestration. The CODED parallel at NBC T3 #091 (”Ben pitched the same war to bush Obama and Biden”) and the same orchestration register at Forbes T8 #002 perform the second move; CBS T11 #0011-16 does not — the predicate stays attached to Netanyahu’s known foreign-policy position and stops short of the orchestration mechanism that would activate the conspiracy register.

“The Nutty Yahoo has promoted a lot of US aggression over the last few decades” (Forbes T8 #002)

Name-mockery directed at Netanyahu through phonetic distortion. “Nutty Yahoo” targets the individual political figure — registering as personal ridicule rather than as a claim about Jews or Israel collectively. The contrast with adjacent name-distortion variants is informative: “Satanyahu” substitutes a demonological vehicle (satan) for “Netan-” and codes accordingly under Lexicon 1.2 because the substitution does the demonization work directly on the name; “Nutty Yahoo” performs no such substitution and stays at the level of comic distortion attaching to a named political actor. The boundary turns on what the wordplay substitutes in. Where the substitution recruits a categorical register (demonological, vermin-coded, financial-extractor), the name-distortion codes; where it operates as personal ridicule of a specific political figure, it stays bounded.

Across the corpus, the distinction is not between criticism and support for Israel, but between criticism that remains politically bounded and criticism that recruits classical antisemitic frames to explain that political position. Part IV does not treat the political rejection of U.S.–Israel alignment as antisemitic per se; the analytical interest is the share of the rejection-discourse that recruits antisemitism registers to do the explanatory work.

Quantitative Apparatus

The analyses below are descriptive of this 23-thread corpus. They identify structured associations inside the sampled material; they are not presented as population-level estimates for YouTube, U.S. news audiences, or online antisemitism as a whole.

Overview diagrams

The four diagrams below visualize the per-thread density distribution, the framing-position correlation, the within-Power-and-Conspiracy sub-form shift documented in Section 5.B.1, and the register-class composition by phase.

Figure D1 — Antisemitism register density per phase. Strip plot of all 23 threads grouped by phase, each thread labeled, with median line and min–max whiskers per phase. Phase A median 17.8 %; Phase B median 11.0 %; Phase C median 13.2 %. Phase A’s distribution shows two concentrations — a high cluster at 24–26 % (T3 NBC, T6 AP, T8 Forbes) and a lower cluster at 14–18 % (four threads) — with T4 CBS isolated at 7.5 %. Phase B contains the corpus density maximum (T13 AP at 33.5 %) but its median is pulled down by the four Position 4–5 threads. Phase C is the tightest distribution and contains no Position 1 threads

Figure D2 — Density × framing-position correlation. Scatter of all 23 threads with linear fit and phase markers. The rank-order association between framing-position and density is strong and negative in this corpus: Spearman ρ = −0.75 (p < 0.001, n = 23); as descriptive shorthand, the Pearson correlation is r = −0.79 (R² = 0.62) and Kendall τ = −0.60 (p < 0.001). Because framing-position is ordinal, the Pearson correlation is reported descriptively and the rank-order association is the primary statistic. The linear fit suggests density falls by approximately 3.9 percentage points per step on the framing-position axis (1 = direct Netanyahu-focused → 5 = Iran-deal-with-Netanyahu-minimal); the relationship holds across all three phases — Phase A, B, and C dots distribute along the same trend line.

Figure D3 — Cat 1.1 blackmail / hidden-control sub-register by Phase C thread. Horizontal bar showing per-thread counts of the sub-register across all eight Phase C threads. Forbes T23 dominates (13 instances); AP T21 (7) and MS NOW T17 (5) follow; Fox News T22, CNN T16, NBC T18, and ABC T20 sit at the floor (2–4 instances). This is the visual record of the within-Power-and-Conspiracy shift toward the blackmail / hidden-control sub-form documented in Section 5.B.1.

Figure D4 — Register-class composition across Phases A, B, C. Stacked bar showing approximate per-thread averages (n = 8, 7, 8 threads) by the five sections of Section 5.B. Power and Conspiracy is the dominant category across all three phases (≈8.0 → 8.8 → 11.0 instances per thread). Classic Antisemitism (Section 5.B.2: Demonology, Dehumanization, and Character Stereotypes, including 2.7 Instrumentalization of Antisemitism) is stable in Phases A and B (≈2.2 → 2.0) and rises in Phase C (≈3.2), driven by the religious anti-Judaism resurgence, the dual-loyalty / Jews-belong-elsewhere register, and the antisemitism-instrumentalization register at 2.7. Attacks on Israel’s Legitimacy — combining Nazi Analogy, Terrorist State, Genocide, and Legitimacy Denial — peaks in Phase C (≈3.3 per thread) under post-signing Israel-keeps-killing framings. Classic Financial sits roughly flat across phases (≈1.4 → 1.4 → 1.5). Aggressive Speech Acts is a smaller category that shows Phase C concentration.

Two-stage determination of antisemitism density

The framing-position of the video structures the opportunity for register activation, with the outlet-audience composition determining the actual density inside that structural condition. Two natural-experiment pieces of evidence support the model.

Controlled framing. T18 NBC and T20 ABC cover the identical Strait-of-Hormuz blockade event under identical Position-5 framing; densities converge within ~2.5 percentage points (7.5 % vs 10 %). The pair suggests that comparable framing can produce comparable density.

Residual outlet effect. Phase C samples four threads at Position 3 (Trump/VP-vs-Netanyahu confrontation) across four different outlets (MS NOW, AP, Fox News, Forbes); densities range from 5 % at Fox News to 19 % at AP — a ~3.8× spread within one framing-position. Phase A at Position 1 (the most direct Netanyahu-focused framing, sampled across six outlets — T2, T3, T5, T6, T7, T8) shows a similar pattern: density runs from 14 % at MS NOW to 26 % at AP, a ~1.9× spread within one framing-position.

The magnitude of the outlet variation suggests that hate-speech measurement studies aggregating across outlets without holding framing constant — or across framings without holding outlet constant — should be interpreted with caution when used for cross-platform or cross-period comparison.

The Phase C compositional shift

The within-Cat 1.1 composition shifts as follows across the three phases:

The base subordination form (dog/master, puppet, owns-the-party) declines slightly per thread across the three phases. The blackmail / hidden-control sub-form (Mossad-Epstein, Trumpstein, golem-Trump, “Bibi has Trump by the curlies,” Israel-first / Epstein-administration) grows from approximately 1.2 to approximately 5.4 instances per thread between Phase A and Phase C — roughly a 4.5× increase — with Forbes T23 alone containing 13 instances in a single 200-comment sample. The sub-form’s share of total Cat 1.1 volume rises from approximately 18 % in Phase A to approximately 57 % in Phase C: by Phase C, the sub-form is producing more than half of all Cat 1.1 coded content. Persistence and mutation are not exclusive — the base form persists alongside the sub-form’s growth — and the qualification is preserved in the Conclusion.

The Phase C density-peak diagnostic

The headline numbers show Phase C as the lowest-peak phase: density tops out at 19 % (T21 AP) versus 33.5 % (T13 AP) in Phase B and 26 % (T6 AP) in Phase A. Read naively, this would suggest the antisemitism register attenuated under the changed political conditions — but that interpretation is not supported by the data. Phase A and Phase B both contain Position 1 threads (direct Netanyahu-focused framings), which produce the corpus’s highest densities; Phase C contains none, with its dominant framing at Position 3 (Trump/VP-vs-Netanyahu confrontation) and four threads at Position 5 (Iran-deal-with-Netanyahu-minimal). Extrapolating from Figure D2 would place a hypothetical Phase C Position 1 thread in the 20–24 % range, comparable to Phase A and B Position 1 outcomes — but this remains an extrapolation, since Phase C contains no Position 1 cases. The Phase C dataset does not contain the evidence to test the attenuation hypothesis at Position 1, because the June 17–21 news cycle did not produce direct-Netanyahu-focused videos.

Discourse Dynamics

The preceding sections treat what antisemitic content appears in the corpus and at what density. This section treats a different analytical level: how that content circulates — the production-level patterns by which a single comment becomes a multi-comment thread, by which a single speaker reaches an audience, and by which one register migrates across outlets. These are not antisemitic concepts in themselves; they are properties of the discourse-production environment that shape how the registers documented in Section 5.B reach readers. Three recurring patterns are visible in the corpus.

Affirmation Cascade Same-thread audience extension of a coded framing through compressed affirmation replies. The reply does not independently restate the antisemitic register but carries the parent’s framing into compressed content:

“Israel? Absolutely agree” (Forbes T23 #029-2) [reply to a coded Israel-as-evil framing]

“Yeah, America don’t have friends” (CNN T16 #0013-1) [in reply to a coded subordination-frame parent]

The pattern is sequence-dependent and only becomes visible because chronological sampling preserves parent–reply relationships. Affirmation cascades function as low-cost endorsement signals: the reply commits the speaker to the parent’s framing without the rhetorical labor of restating it, and makes the parent-comment framing appear more broadly endorsed to subsequent readers.

Same-Speaker Templates A single speaker deploys an identical antisemitism template across multiple top-level comments in one thread to maximize visibility. T22 Fox News supplies the corpus-strongest instance — one speaker (@keving5711) deploys the Jews-threaten-Christians template three times in the same thread, replying to three different top-level comments:

“Ben Shapiro is a threat to American Christian peaceful values, USA would be better off if he fled to Tel Aviv instead of Tennessee” (Fox T22 #0004-3)

“Ben Shapiro is a threat to American Christian peaceful values, USA would be better off if he fled to Israel instead of Nashville” (Fox T22 #0008-01)

“Zionist Bot. Ben Shapiro is a threat to American Christian values, America would be better off if he fled to Israel instead of Nashville” (Fox T22 #0009-3)

The template-deployment pattern is telling because it indicates the speaker treats the comment-section infrastructure as a broadcast medium: replying to three different top-level comments with the same template maximizes the framing’s reach across the thread.

Cross-Thread Amplifiers A single speaker appears across multiple threads in the corpus carrying the same register — for example, one speaker appears at MS NOW T17 and at ABC T20 carrying the antisemitism-instrumentalization register treated at Section 5.B.2 (2.7) (different threads, different outlets, same coded content). Speaker identity across threads is established via the stable YouTube channel handle attached to each comment, which persists across comment sections regardless of video or outlet. The amplifier pattern is distinct from intra-thread template deployment: the speaker is moving between videos, not within a single comment section. The pattern’s visibility depends on cross-thread author tracking that single-thread studies cannot perform; it is one of the methodological yields of the per-phase per-outlet sampling design used in Part IV.

The three patterns are mutually reinforcing. Affirmation cascades extend a coded framing inside one comment thread; same-speaker templates extend it inside one comment section across multiple parent-comments; cross-thread amplifiers extend it across the outlet boundary. Together they constitute the production-level architecture by which the registers documented in Section 5.B circulate, and they belong to the study of discourse-production rather than to the study of antisemitism content per se.

Conclusion

The three-hypothesis test set out in the Four-Study Arc section produces a clear empirical answer: the mutation hypothesis is supported, with two qualifications.

The mutation finding. Of the three hypotheses set out in the Four-Study Arc section, the Phase C composition is most consistent with mutation over attenuation or stable persistence. Under the changed political conditions — Trump signing a deal Israel publicly opposes, Vance publicly rebuking the Israeli Cabinet, Netanyahu refusing to withdraw from southern Lebanon — the subordination frame documented in Part III does not attenuate. Its internal composition shifts within the Power and Conspiracy register, with the dominant sub-form moving from the base subordination idiom (dog/master, puppet, owns-the-party) toward a blackmail and hidden-control sub-form built around Mossad-Epstein blackmail, Trumpstein, golem-Trump, “Bibi has Trump by the curlies,” and Israel-first / Epstein-administration framings. The blackmail sub-form grows from approximately 1.2 instances per thread in Phase A to approximately 5.4 in Phase C, and its share of total Cat 1.1 volume rises from approximately 18 % to approximately 57 % across the three phases (Section 5.D table). Forbes T23 alone contains 13 instances in a single 200-comment sample. The pattern is consistent with the conspiratorial framework absorbing apparent counter-evidence through internal recomposition rather than collapsing under it; whether this reflects individual speakers adapting their framing, a change in which speakers are most active, or differences in the framing-position mix across the three phases, the dataset cannot directly distinguish. Persistence and mutation are not exclusive — the base form persists alongside the sub-form’s growth — and the central evidentiary support for mutation is therefore the within-Cat 1.1 compositional shift, not a wholesale replacement of one register by another.

First qualification: density does not spike. The raw density numbers do not show Phase C as the densest phase — the corpus maximum sits at T13 AP in Phase B (33.5 %), and Phase C tops out at 19 %. This is explained by the framing-position mix of the Phase C news cycle (no direct Netanyahu-focused videos were available in the June 17–21 window), not attenuation of the antisemitism register itself. The regression line in Figure D2 predicts that a Position 1 thread at Phase C would have produced density in the 20–24 % range — comparable to Phase A and Phase B Position 1 outcomes. The mutation hypothesis is therefore supported on compositional grounds, not volumetric ones.

Second qualification: religious antijudaism resurges. Alongside the mutation finding in the Power and Conspiracy register, Phase C surfaces a concentrated resurgence of the religious antijudaism register (Christ-killer + Jews-as-usurers, anti-Christian framings, Jews-threaten-Christians) and a continued presence of the Nazi analogy at right-leaning outlets. These were latent in Phase A at Forbes and Fox News and re-surface at higher cross-outlet concentration in Phase C, accompanied by new lexical-allusion vehicles entering the selected mainstream-news YouTube comment sections — yarmulke-coded allusions (”tiny hats,” “small hats”), neo-Nazi numeric codes in usernames (1488), and Holocaust-denial book recommendations.

A second finding: the two-stage determination of antisemitism density. One broader methodological contribution of this study is the two-stage determination model. In this corpus, video framing strongly constrains observed antisemitism density (Spearman ρ = −0.75, Pearson r = −0.79 as descriptive shorthand, across the full 23-thread corpus). Outlet-audience composition modulates density inside that constraint by approximately 3.8× at the same framing-position. The methodological implication: any antisemitism density figure reported without outlet specification, or without framing-position specification, is not meaningfully interpretable for cross-platform or cross-period comparison. Hate-speech measurement studies that aggregate across outlets without holding framing constant — or across framings without holding outlet constant — produce numbers that cannot reliably support cross-platform or cross-period claims. The two-stage model is a minimal correction; richer models would add speaker-demographic and platform-affordance terms, and the generalizability of the framing-position–density relationship beyond this 23-thread corpus is itself an empirical question for future studies.

On the cross-ideological symmetry question. Part III’s striking CNN–Fox News convergence (21 % vs 20 %) does not survive into Part IV. Across the 23 Part IV threads, Fox News produces 5–17.5 % density depending on framing-position and phase; CNN produces 5–18.5 %. The two outlets remain in the same density band, but the tight one-percentage-point convergence appears to have been specific to the period when the U.S. was actively fighting alongside Israel. The Phase C divergence — Fox News at 5 % under the Shapiro “DISASTER” framing (T22), CNN at 5 % under the Trump-responds-to-GOP-backlash framing (T16) — is consistent with a right-wing audience navigating competing commitments between America-First and pro-Israel positions: the America-First base welcomes the deal while the pro-Israel base struggles to defend both Trump and Netanyahu. The pattern visible in the data is that the right-wing comment population redirects its register-production toward the blackmail and hidden-control sub-form (Trumpstein, Epstein-administration) and toward the religious antijudaism register (T22 anti-Christian template, T23 Christ-killer + usurer) rather than toward base-form subordination. The outlet distribution of the base form in this corpus skews to CNN, MS NOW, AP, and ABC; the Phase C right-leaning audiences appear to favor the conspiratorial and theological registers over the base subordination idiom, an observation that warrants further testing on a larger right-wing sample.

On the Trumpstein thread under policy reversal. The Trumpstein thread does not attenuate. The Trump-applied form consolidates at Phase B T12 ABC and saturates at Phase C across T16, T19, T22, and T23, paired with golem-Trump and “Donald Judas Trump” variants. The conspiracy thread is more autonomous from immediate political reality than Part III’s argument allowed for, but the autonomy is bounded: the thread deepens by attaching itself to the deal as a new occasion for hidden-control narration (Mossad-Epstein blackmail explains why Trump signed it, Netanyahu-manipulated-Trump explains why he is now publicly criticizing Israel, the Epstein-administration framing explains everything else). The persistence and the mutation hypotheses are not exclusive; the Trumpstein thread evidences both.

The broader pattern: conspiracist discourse under empirical contradiction. The mutation finding has a counterpart in the wider literature on conspiracy belief. Festinger, Riecken, and Schachter (When Prophecy Fails, 1956) documented that the members of a UFO group whose explicit prediction of world-destruction was empirically falsified did not abandon their framework; they reinterpreted the falsification as further evidence of the framework’s truth. Barkun’s A Culture of Conspiracy (2003) supplies the complementary structural account: conspiracism operates as an epistemic system in which appearances are presumptively deceptive, hidden connections are foundational, and accidents are excluded by definition — premises that make the reinterpretation of contradiction not an ad hoc rescue but a native move within the framework. The general property — that conspiratorial explanatory systems can absorb apparent empirical contradiction by reinterpreting the contradictory event within the framework’s own terms — is well-established in the broader conspiracy-belief literature as a structural feature of the genre rather than an idiosyncratic finding.

What the literature has documented less often is the discursive mechanism through which that persistence unfolds inside mainstream public communication. Part IV contributes a documented within-case observation of the mechanism operating at scale in a non-cult, mainstream-news comment-section setting. The empirical material is a 4,598-comment corpus across 23 threads in eight outlets, tracking a single conspiratorial framework — Jewish or Israeli control of U.S. foreign policy — across a political condition that on its face contradicts the framework’s core claim. The observed pattern is consistent with the mechanism the broader literature predicts: the framework does not collapse, it shifts toward a sub-form that absorbs the contradictory evidence. The blackmail and hidden-control sub-form converts Trump’s public break with Netanyahu into evidence of Mossad’s leverage over Trump rather than disconfirmation of Israeli control; the “fake feud” framing converts the visible rupture into staged performance preserving the hidden coordination underneath. Each is an instance of the same operation — apparent counter-evidence reinterpreted as further confirmation of hidden coordination.

The mechanism identified here may extend beyond antisemitic discourse. Comparable compositional mutation could plausibly occur in other conspiratorial frameworks — including anti-vaccine narratives, Great Replacement discourse, anti-Muslim conspiracy narratives, and QAnon — but establishing this requires separate corpora and domain-specific codebooks. Whether compositional mutation is a general property of conspiracist discourse remains an open empirical question.

The contribution to the antisemitism-research field specifically is a documented instance of the mechanism operating in a contemporary mainstream-news venue rather than a fringe one. The 23-thread corpus is not representative of antisemitic discourse on social media as a whole — that limitation is flagged in the Sampling rationale section — but it does establish that the mutation pattern is observable in the same outlets, in the same comment-section environment, in which the prior installments documented the base register. Under conditions that on their face contradicted the register’s core claim, the register did not move to a more extreme venue. It stayed where it was and adapted.

Implications for the broader field. Three implications follow from the corpus. First, hate-speech measurement reports that do not specify outlet and framing should be read with caution; the underlying numbers cannot be compared across studies that do not control for these variables. Second, the antisemitism register documented in this series is event-conditioned in the sense Part III argued, but the event-conditioning is more flexible than a simple political-fact-causes-register model allows: under the changed political conditions, the register’s composition shifts within itself rather than collapsing. Third, the religious antijudaism register — the medieval Christian-anti-Jewish theology that the prior installments tracked as recessive in mainstream U.S. comment sections — re-surfaces under conditions of perceived Israeli moral failure, with new lexical-allusion vehicles entering the corpus. The recession was conditional, not stable.

References

Barkun, M. (2003). A culture of conspiracy: Apocalyptic visions in contemporary America. University of California Press.

Becker, M. J. (2024). Jordan Peterson and conservative antisemitism online: The dethroning of an intellectual icon following his interview with Netanyahu. In M. J. Becker, L. Ascone, K. Placzynta, & C. Vincent (Eds.), Antisemitism in online communication: Transdisciplinary approaches to hate speech in the twenty-first century (pp. 47–74). Open Book Publishers. https://doi.org/10.11647/OBP.0406.02

Becker, M. J. (2025a). Antisemitism in real time, Part I: U.S. YouTube reactions to Israel’s strike on Iran. Decoding Antisemitism. https://decodingantisemitism.substack.com/p/antisemitism-in-real-time-us-youtube

Becker, M. J. (2025b). Antisemitism in real time, Part II: U.S. YouTube reactions to Iran’s retaliatory strike on Israel. Decoding Antisemitism. https://decodingantisemitism.substack.com/p/from-retaliation-to-rhetoric-digital

Becker, M. J. (2026). Antisemitism in real time, Part III: “Israel Made America Do It” — YouTube reactions to the US–Israel campaign against Iran. Decoding Hate, NYU Center for the Study of Antisemitism. https://decodingantisemitism.substack.com/p/antisemitism-in-real-time-part-iii

Becker, M. J., Troschke, H., Bolton, M., & Chapelan, A. (Eds.). (2024). Decoding antisemitism: A guide to identifying antisemitism online. Palgrave Macmillan. https://link.springer.com/book/9783031492372

Festinger, L., Riecken, H. W., & Schachter, S. (1956). When prophecy fails: A social and psychological study of a modern group that predicted the destruction of the world. University of Minnesota Press.

This article is Part IV of the Antisemitism in Real Time series. Parts I and II were conducted under the Decoding Antisemitism project (TU Berlin, HTW Berlin, King’s College London; June 2025). Part III marked the continuation of this work under Decoding Hate, the expanded research hub at NYU’s Center for the Study of Antisemitism. Part IV continues under Decoding Hate. All four parts draw on the Decoding Antisemitism Lexicon.

AddressHate Research Scholar at NYU’s Center for the Study of Antisemitism | PI, Decoding Hate | Research Advisor, AddressHate | Editor-in-Chief, Digital Hate Review

Executive Summary

On June 18, 2026, at Brooklyn’s Kings Theatre, New York City Mayor Zohran Mamdani gave a rally speech in which he described AIPAC — the American Israel Public Affairs Committee — as one of the “monsters” New Yorkers face. He accused the organization of moving “millions in dark money” toward “a single goal” of “turning us against one another,” and said AIPAC fears the end of “genocide and Netanyahu’s wars.” On June 22, at City Hall, he defended and extended that framing.

A substantive political critique of AIPAC’s spending and influence — no other lobby PAC was mentioned in the rally — is as legitimate as criticism of any other lobbying organization. However, this study examines something different: the specific phrases the mayor chose stand in the foreground, and what happened when audiences received them. The corpus comprises three strata of digital reception assembled in the five days following the rally: 15 supportive Instagram posts (N = 1,497 codable comments), 5 critical Instagram posts (N = 499 codable comments), and 4 mainstream-press YouTube threads (Forbes Breaking News, CNN, NBC New York, CBS New York; N = 400 codable comments). Total coded N = 2,396 across all three strata. The full reception corpus is coded against a published catalog of antisemitic discourse patterns developed over five years of empirical research and documented in Decoding Antisemitism (Becker et al., Springer Nature/Palgrave Macmillan, 2024).

Three findings stand out.

One: across 1,497 codable comments in 15 supportive Instagram posts, 75% took up or extended speech patterns that overlap with documented antisemitic discourse. Per-post combined rates ranged from 58% to 96%, with 22% deploying antisemitic patterns directly (compression from “AIPAC” to “Jews” as a group, claims about Jewish control of US politics, dehumanizing imagery with documented Nazi-era genealogies, blood-libel registers, conspiracy-network expansion) and a further 53% substantively endorsing or amplifying the post’s speech patterns without themselves adding directly antisemitic phrasing. The 75% measures the overall audience-response environment, not the share of comments that themselves activate antisemitic tropes; the 22% is that strict count. In threads critical of the mayor’s framing, the comparable combined rate was 1.4% — a 53-fold cross-stratum gap. The same rally language entered very different comment environments, depending on whether the post endorsed or criticized it.

Two: two of the mayor’s signature phrases propagated through pro-Mamdani comment threads on both Instagram and mainstream-press YouTube; the others did not. “Monsters” and “dark money” each generated extensive direct uptake: 23 and 22 direct-extension instances respectively across the five most-engaged supportive Instagram posts (N ≈ 500). Across the four mainstream-press YouTube threads (N = 400), “monsters” was reproduced about 30 times — roughly 22 extending the rally’s framing toward Jewish-associated targets, and about 8 redirected against Mamdani himself — while “dark money” generated 98 activations of the broader Power/influence framing. Audience commenters extended the predicates from AIPAC toward Jews and Israel as a group, including what this study treats as the corpus’s most consequential single activation: the two-word “Jewish playbook” compression that drops AIPAC entirely. Within this corpus, other phrases the mayor used — “single goal of turning us against one another,” “fear of peace” — produced no observed uptake across the supportive Instagram stratum, the critical Instagram stratum, or the mainstream-press YouTube stratum. The phrases that traveled were the ones that named a moral antagonist in absolute terms; the longer attributive phrases — describing actions or motivations — did not. Much of the audience-side reception therefore appears to build on his lexical material, even as commenters also draw on pre-existing framing repertoires.

Three: where mainstream-press accounts engaged the controversy on the platform, the reception was not moderating — it produced a markedly higher antisemitism-relevant activation rate than rates observed in comparable prior project studies. On Instagram, mainstream-press outlets produced almost no rally-related posts with meaningful audience engagement in the seven days following the rally; the Instagram reception ran almost entirely through partisan-advocacy accounts on either side. On YouTube, by contrast, four high-engagement mainstream-press threads (Forbes Breaking News, CNN, NBC New York, CBS New York; combined ~309K views; N = 400 codable comments (the first 100 visible comments per thread)) produced a combined antisemitism-relevant rate of approximately 80% — within the range of the supportive Instagram stratum (58–96%) — and a 42.5% explicit antisemitic-trope activation rate, approximately double the supportive Instagram rate (22%). No substantive anti-antisemitism counter-speech was observed across the 400 coded comments. The mainstream-press YouTube ecology around the Mamdani rally functioned not as a moderating third option but as a polarized middle: the antisemitism-relevant activation rate the supportive Instagram stratum produces in diffuse form appears concentrated and intensified in the coded mainstream-press YouTube threads. The platform-architectural observation about mainstream-press absence holds for Instagram; the cross-platform finding sharpens it: increased mainstream-press platform presence is not, by itself, the missing moderating force, because the mainstream-press comment ecology where it does engage propagates the same patterns.

The study makes no claim about the mayor’s intent — intent is not the question, and the framework does not require an answer to it. What the study claims is that the specific language chosen overlaps with patterns cataloged across a century of historical scholarship on antisemitism, that this language was received in measurable ways the study quantifies, and that the substantive political critique of AIPAC is available in language that does not overlap with these patterns. The specific phrases chosen — not the political criticism itself — were associated with measurable audience-side uptake of historically documented antisemitic discourse patterns. Those audience-side uptake patterns are what this study measures.

Implications for journalists, policymakers, and political actors

The specific phrases Mayor Mamdani used about AIPAC varied dramatically in how audiences took them up. "Monsters" and "dark money" — the two phrases with the longest historical genealogy of overlap with antisemitic tropes — were widely reproduced across all three reception strata, intensified and — in supportive and editorially neutral mainstream environments — frequently redirected from AIPAC to Jews as a group. The two phrases without that historical genealogy ("single goal of turning us against one another," "fear of peace") produced no observed uptake across N = 2,396 coded comments. These specific phrases — not the political critique itself — drove the high audience-side activation the study measured. The same critique of AIPAC could have been made in other words. Political actors choosing words about Jewish-associated institutions, journalists framing controversies of this kind, and platforms shaping reception environments should weigh not only what language is intended to mean but also which historical repertoires that language is likely to activate in the audiences that receive it.

Introduction

Political language does not stay on the stage where it is spoken. It travels into the audiences that receive it, the platforms that propagate it, the communities that contest it, and the broader digital discourse environment in which contemporary publics encounter one another. When that language works through demonization — by attributing to a particular group an essential malign quality, hidden coordinated agency, or destructive purpose — it can alter the discourse environment in ways that reduce democratic deliberation. These effects are structural and largely independent of the speaker’s intent.

Five features of these effects are relevant to the case this study examines. First, sustained demonization of a target community in mainstream political speech can contribute to the withdrawal of members of that community from public conversation through harassment, intimidation, and the normalization of hostile discourse. Second, demonization replaces argument with enemy-identification: the targeted group is no longer a political actor to be argued with but a malevolent collective to be defeated, neutralized, or expelled. Third, demonization paired with conspiracist explanation — visible institutions reframed as surface effects of hidden hostile agency — denies the legitimacy of disagreement and produces a permanent emergency frame in which exceptional measures against the imagined enemy are licensed. Fourth, essentializing enemy constructions corrode the democratic premise that political opponents are legitimate counterparts to be argued with rather than coordinated enemies to be defeated; these effects extend beyond the targeted community to the civic culture in which all political actors participate. Fifth, in some cases sustained online demonization has formed part of the discursive background to offline violence — including attacks in Pittsburgh (2018), Christchurch (2019), Buffalo (2022), and the Capital Jewish Museum in Washington (2025) — an observation that is contextual, not evidentiary: it does not establish a causal claim about the case examined here, but it explains why demonizing public language warrants careful analysis when it enters mainstream political speech.

This essay examines a focused instance: NYC Mayor Zohran Mamdani’s June 18, 2026 Kings Theatre rally and his June 22, 2026 City Hall press conference defending the rally’s framing. The substantive political question — what AIPAC does, how it spends, what positions it advocates, how it should be criticized — is a legitimate subject of democratic contestation. The discourse-analytic question is different: when criticism of AIPAC takes the specific phrasings the mayor used, what historical predicates do those phrasings activate, what tropes do they invite, what readings do they make available to audiences carrying the relevant repertoire, and how does the resulting discourse circulate in the publics that encounter it? These questions are testable, both through close reading of the speech and through empirical analysis of audience-side reception.

The contribution this research attempts is methodological as well as empirical. Read at the broadest level, this article is a case study in how codes and frames travel from authoritative speakers into online cultures of reception. Previous discourse-analytic work on antisemitism has primarily examined speakers' language — the predicates, frames, and tropes that appear in source texts. This study extends that apparatus to the audience side, tracing how historically loaded predicates are received, transformed, amplified, contested, and generalized across digital reception environments. The argument is not only that the rally’s language overlaps with documented antisemitic tropes at the textual level; it is also that this overlap is observable, at measurable rates and with documented bidirectional dynamics, in the reception that actually received the rally. The empirical center of this paper is the audience-side trace.

The case is also analytically distinctive at the level of topical choice: the rally foregrounded AIPAC’s federal-lobbying activity and the Israel–Hamas war — terrain on which a New York City mayor has no direct authority — rather than the housing, transit, and affordability policy questions that municipal office actually controls. Among the super PACs active in the June 2026 New York primaries, AIPAC is the only one named, elaborated, and made the rhetorical center of the speech. The mayor’s later defense — that he meant “super PACs at large” — does not change which organization the speech actually foregrounds. The choice to mobilize around a federal-lobbying organization widely perceived as Jewish-associated is itself a rhetorical move: it shifts the discursive ground from policy-lever questions to questions of moral antagonism on a highly emotionally charged topic, and the framing accomplishes that shift before the substantive critique of AIPAC’s spending or lobbying activity has been made.

What discourse analysis can do is identify what language draws on, what it can evoke, and how audiences read it. What it cannot do is determine what the speaker believes. This essay makes no claim about Mamdani’s intent. It claims that the language of the rally and the press conference overlaps with documented antisemitic tropes, that this overlap makes a group-level reading of the language available to audiences carrying the relevant historical repertoire regardless of speaker intent, and that the empirical reception corpus assembled in the days following the rally — coded across all three strata at total N = 2,396 codable comments — exhibits audience-side patterns consistent with this discourse-analytic reading.

The article proceeds in three parts. Section 1 reads the speech and its defense, identifying the specific phrases and framing moves that warrant attention and locating each within the documented tropes of the Decoding Antisemitism Lexicon. Section 2 turns to the empirical reception: a stratified sample of 24 posts and threads in the 5-day window following the rally (15 supportive Instagram posts, 5 critical Instagram anchor cases, and 4 mainstream-press YouTube threads from Forbes Breaking News, CNN, NBC New York, and CBS New York), with the first 100 visible comments per post or thread coded against the Lexicon’s categories and against counter-discourse escalation categories — 2,396 codable comments in total. Section 3 presents what the reception shows: how supportive audiences extended the framing, how critical audiences identified it, how cross-spectrum convergence emerged across otherwise-opposed political positions, what bidirectional vocabulary propagation across all three strata reveals about the rhetorical reach of the rally’s signature predicates, and what the platform-architectural environment within which the controversy propagated reveals about the contemporary digital discourse environment more broadly. The conclusion returns to the opening stakes: what political language of this kind does to the discourse environment that receives it, and what — narrowly but consequentially — follows from the diagnosis.

1. The Speech

1.1 The framework’s starting point

Decoding Antisemitism catalogs antisemitic concept-clusters across forty chapter-level categories, organized in five parts: Classic Antisemitism (chapters on Other/Foreign, Evil/The Devil, Blood Libel, Repulsiveness and Dehumanization, Immorality, Lie and Deceit, Vengefulness, Disloyalty, Blame for Antisemitism), Concepts of Power (Greed, Power, Conspiracy Theories, Disintegration, Self-victimization), Secondary Antisemitism (Rejection of Guilt, Holocaust Denial, etc.), Further Post-Holocaust Categories, and Israel-Related Antisemitism. When discussing the Power concept, the Lexicon is explicit on one point: lobbying is a lawful political process in modern democracies, and the work of Israeli lobby organizations is a matter of fact that does not in itself constitute antisemitism.

Critique of AIPAC is a legitimate political activity. The concern is that some phrases used in such critique have a long history in antisemitic discourse, where they attributed hidden power, division, or moral corruption to Jews as a group. These phrases developed historically as forms of group attribution, and they have appeared, across different periods and political settings, in traditions that helped justify exclusion, persecution, and violence against Jews.

When those phrases are used together and attached to a Jewish-associated institution, they can evoke those tropes in audiences (consciously or unconsciously) familiar with them. This is one way historical repertoires operate in public discourse. The audience does not perform a logical inference from “AIPAC bad” to “Jews bad.” The phrases already belong to a historical repertoire built around Jews as a group. Referring to an institution rather than a group does not necessarily prevent audiences from hearing those associations.

This is not a hypothetical concern. On the day of Mamdani’s rally, a federal grand jury in the Southern District of Florida indicted Forrest Kendall Pemberton for an alleged December 2024 mass-shooting plot targeting AIPAC employees at the organization’s Plantation, Florida office; two days earlier, federal authorities had announced charges against five men in a separate alleged accelerationist plot whose target list (per the FBI affidavit) included AIPAC-supported lawmakers. Both proceedings are allegations, not convictions, and they do not establish a causal link to any political speech. They establish only that violent threat environments around AIPAC are not hypothetical — a context in which public speech about the organization warrants particular analytical attention. The ADL’s Audit of Antisemitic Incidents 2025 and the corresponding ADL New York / New Jersey regional report recorded 1,160 antisemitic incidents in New York State (the highest of any state), with New York City accounting for 860 of those incidents and 85 of the state’s 90 antisemitic assaults.

Audiences process political references analogically. Historically loaded references bring broader associations and narratives into a discussion, even when those associations are not stated explicitly. What follows is a diagnosis of six recurring phrases and framing moves across the June 18 rally and the June 22 press conference.

1.2 The “monsters” predicate

Mamdani identifies AIPAC as one of the “monsters” New Yorkers face, in a passage framed by Antonio Gramsci’s line that “now is the time of monsters.” At the rally, the mayor’s own framing of the predicate is universalist — “these monsters take many forms today” — not NYC-specific; the closing reinforces it: “In the wealthiest city, in the wealthiest country in the history of the world, we need not live in fear of monsters any longer.” Pressed on the term in a press conference after the rally, the mayor reaffirmed it.

The phrase is dehumanizing in form. “Monster” is not a description of conduct or position; it is a category of being. Problematic activities of a lobby group can be criticized in the strongest terms, but the dehumanization of the people who constitute that lobby serves no intellectual or ethical purpose. The Lexicon’s Repulsiveness and Dehumanization chapter documents the antisemitic stereotype that depicts Jews as monstrous or subhuman, from medieval Christian iconography through twentieth-century propaganda. The Evil/The Devil chapter — whose sub-section on the depiction of Jews as “demonic or Satanic creatures” cross-references the dehumanization chapter — extends the same diagnosis. The concern is not the word alone. It is the combination of that predicate with a Jewish-associated target in a passage that also uses the phrases analyzed below.

1.3 The Gramsci frame

The rally opens its central passage with: “And we need that because, as Gramsci once wrote, ‘The old world is dying, and the new world struggles to be born. Now is the time of monsters.’” The press conference reaffirmed the citation under direct questioning.

The citation is not decorative. Gramsci wrote the Prison Notebooks under confinement by Mussolini’s regime; for audiences who carry that historical context, the citation can make available an interpretive frame in which the speaker occupies the anti-fascist position and the named opponents occupy the role of obstructing forces. The mayor did not say AIPAC is fascist. He did not say Israel is fascist. The diagnostic point is narrower: the framing makes that analogical reading available, and the substantive critique of AIPAC’s spending and lobbying does not require this scaffolding.

1.4 The “dark money” claim

The rally attributes to AIPAC the movement of “millions in dark money to accomplish a single goal.” The press conference elaborates: AIPAC “filters money that would have previously been directly from AIPAC now through other shell organizations whose identities of their contributors are only made clear after an election.”

A distinction is needed. AIPAC’s principal political vehicle, the United Democracy Project, files publicly with the Federal Election Commission and discloses its donors. Per FEC filings reported by CNN on June 22, 2026, UDP made two political contributions in May 2026 totaling over $600,000 to BOLD America, which has funded ads supporting Rep. Adriano Espaillat against Mamdani-backed challenger Darializa Avila Chevalier in New York’s 13th Congressional District. Applied to UDP itself, the term “dark money” — which has a specific meaning in U.S. campaign finance: money from donors whose identities are not disclosed — does not fit.

The press conference’s more elaborated claim is closer to the campaign finance record. Several partner and pop-up super PACs active in the same primary cycle, including groups recently created and reported alongside the broader pro-Israel super PAC ecosystem, are not required to disclose donors until after the June 23 primary. Investigative reporting has documented that pattern in detail. As a factual claim about parts of the broader architecture in which AIPAC’s allied money moves, the “shell organizations” elaboration is defensible.

The diagnostic concern is therefore narrower than a charge of factual error. It is that the phrase “AIPAC moves millions in dark money” collapses two distinguishable things: a disclosed super PAC (UDP), and a network of partner PACs whose disclosure timing is genuinely opaque. Collapsing them under a single label, attached to a Jewish-associated institution, renders the historical resonance of the term available — and risks importing it — beyond what the underlying campaign finance facts about AIPAC’s own spending arm carry. The image of concealed Jewish financial influence operating through opaque structures is a classic antisemitic frame, cataloged in the Lexicon’s Conspiracy Theories chapter as part of the world-domination sub-cluster and central to the Protocols of the Elders of Zion and to conspiracy narratives reproduced across the twentieth century. The political point about post-election disclosure can be made — and made forcefully — by naming the partner PACs whose timing actually fits the description, rather than by collapsing AIPAC’s disclosed spending into the same category.

This diagnostic reading is supported by the speaker’s own June 22 elaboration. Pressed at the press conference on the “dark money” phrasing, Mamdani explained: AIPAC defends the status quo “through dark money, by filtering money that would have previously been directly from AIPAC now through other shell organizations whose identities of their contributors are only made clear after an election” (official transcript, NYC Mayor’s Office, June 22, 2026). The phrasing is analytically revealing. “Shell organizations whose identities of contributors are only made clear after an election” is structurally an accurate description of the partner-PAC disclosure-timing pattern that the diagnostic above identifies as the legitimate object of post-election-disclosure critique. When pressed to defend the original framing, the explanation moves toward the analytically more precise concept — post-election-disclosure partner PACs — but does not perform the disaggregation between UDP and the partner-PAC network: the umbrella label “dark money” remains attached to AIPAC. The speaker’s elaboration is therefore not a contradiction of the diagnostic but a confirmation of it: the legitimate critique exists, can be made precisely, and is being made imprecisely in the original framing.

Rabbi Misha Shulman, who leads the progressive Brooklyn synagogue The New Shul and is a member of the progressive Israeli-American group Israelis For Peace, told the Jewish Telegraphic Agency on June 22, 2026 that the dark-money framing was more troubling to him than the “monsters” line itself. “For me, the question of dark money was the tougher knot,” he said, calling Mamdani’s remark a “tactical mistake.” “For a left-wing leader to use that phrase, and invite traditional antisemitism into this conversation in that way, was not smart” (Andrew Lapin, JTA, June 22, 2026). Shulman is a Mamdani supporter; the criticism comes from within the mayor’s progressive Jewish base, not from his political opposition.

Rep. Josh Gottheimer, a Jewish Democrat and critic of the mayor, named the content the phrases draw on directly: “’Monsters.’ ‘Dark money.’ A hidden hand ‘turning us against one another.’ Swap ‘AIPAC’ for ‘Jews’ and it’s the oldest antisemitic conspiracy theory in the books. That’s not criticizing a lobby. That’s laundering antisemitism from your podium as Mayor of a city with more than a million Jews.” His reaction matters not as proof of intent, but as evidence of how the phrases were heard by a Jewish public figure using a straightforward substitution test.

1.5 The “fear of peace” attribution

AIPAC is characterized in the rally as an organization “for whom the only thing more frightening than democracy being allowed to run its course is an end to the genocide in Netanyahu’s wars.”

The relevant Lexicon categories are Evil/The Devil and Vengefulness, with possible cross-activation of the Warmongering Conspiracy Theories sub-cluster within Conspiracy Theories. The Evil/The Devil chapter catalogs the attribution to Jewish actors of essential malicious nature — “the intent to commit evil for evil’s sake” as “the essential core of the individual or group as Jewish.” The Vengefulness chapter catalogs the supposed “irrational, misanthropic, egoistic drive to punish those perceived as having wronged them,” including “alleging aggressive or bloodthirsty behavior on the part of Jewish or Israeli figures” and “alleging Jews or Israelis take a particular sadistic pleasure in taking violent retaliatory action against opponents.” The Warmongering sub-cluster catalogs “the idea of Jews seeking to convince, coerce or bribe rulers of states into declarations of war.” All three chapters and sub-clusters draw the necessary distinction: such accusations are different from legitimate criticism of Israel’s foreign policy.

Substantive critique of AIPAC’s positions on the Israel-Hamas war is legitimate political activity. The attribution that AIPAC fears the end of war is structurally different. It does not describe a policy position; it ascribes an essential disposition. The predicate “fears peace” is closest in form to the Evil/The Devil cluster: it attributes not what AIPAC does (lobby for particular military policies) but what AIPAC is (an actor whose essence is to fear the end of war). The surrounding framing — “the genocide in Netanyahu’s wars” — can be read as adding a Vengefulness dimension: the wars are characterized as disproportionate retaliatory violence, and the wording invites a reading in which the named actor is positioned as wanting that violence sustained. This is a structural reading of the predicate, not an interpretation of any actual organizational goals; what the analysis identifies is what the wording makes available to audiences carrying the relevant repertoire. Whether the Warmongering sub-cluster also activates depends on whether audiences read the predicate as ascribing essential disposition (Evil) or hidden orchestration (Conspiracy Theories / Warmongering). Both readings are available; both are cataloged in the Lexicon as antisemitic tropes distinct from legitimate political criticism.

1.6 The “single goal” — Disintegration

The rally specifies AIPAC’s “single goal” as “to preserve their power, so that they can turn us against one another instead of turning our leaders toward the moral change we all know to be necessary.”

The relevant Lexicon category is Disintegration — the attribution of a tendency toward separation or destruction of a community, with the key identifying characteristic of provoking divisions, endangering social cohesion, or seeking to abolish desirable structures of life. The chapter traces this trope through European antisemitic discourse from the nineteenth century through Nazi propaganda to contemporary forms, including the “Great Replacement” narrative. The Conspiracy Theories chapter cross-references Disintegration through a “White Genocide / Disintegration Conspiracy Theories” sub-cluster, distinguished by attention to the means by which Jewish groups supposedly undermine the “foundational values” of the nation-state or civilization as such. What links these representations across the Lexicon’s chapters is the structural move of attributing to a Jewish actor an essential purpose of producing division.

The Disintegration concept is also the broader analytical category that grounds this essay’s opening claim. Language that endangers social cohesion, that provokes division, that attributes essential malign purpose to a community — this language works against the conditions of democratic discourse itself, regardless of which group is targeted or which political direction the speaker comes from. The antisemitism diagnosis in this case is the specific instance; the disintegration of democratic discourse is the broader phenomenon.

The Lexicon’s own non-antisemitic example shows the line clearly. “Netanyahu is a political troublemaker” is not antisemitic because the criticism is tied to his political actions, not to his Jewish identity. The difference is between attributing specific divisive actions in a specific context and attributing a single purpose of causing division.

The rally’s phrase is in the second category. It does not say AIPAC engages in divisive tactics in specific races. It says AIPAC’s single goal is “to turn us against one another.” The attribution describes not what AIPAC does, but what AIPAC is supposedly for.

1.7 The June 22 press conference

The City Hall press conference was the mayor’s deliberative response to criticism, not an extemporaneous rally line. It contained three further framing moves that warrant attention.

The scope expansion. AIPAC is said to have “fought any attempt to actually deliver safety to people not just in Palestine, but frankly through much of the region.” A single American lobbying organization is positioned as a causal force in regional insecurity across the Middle East. That is a very broad claim. The Lexicon’s Power chapter catalogs the attribution to Jewish actors of disproportionate, region-wide or world-wide causal power as a feature of the Jewish omnipotence stereotype. Whether that historical pattern applies in any given case depends on the specific attribution being made; the concern here is the structural form of the attribution.

The Gaza-to-New-York causal chain. The press conference opens with: “We’re talking about a status quo where children are being killed on a daily basis.” The argument begins with an emotionally salient image of suffering and ends with an attribution of responsibility to AIPAC — sustaining “a status quo that is quite literally starving people in this city, all in the name of sustaining something that we simply cannot defend any longer” — creating a causal chain that is asserted rather than explained. The Lexicon’s Power chapter notes that claims that Jews cause local hardship have been especially volatile in the historical record, because they translate abstract suspicion into immediate, locally felt grievance.

The complicity-naming move. The press conference closes: “I think that it is important that when we ask ourselves how such death and destruction is happening overseas, we also name those who allow it to take place.” The phrase “those who allow it” can reach beyond the named organization to the people who support it. This is not, in isolation, a distinct Lexicon category — extending political accountability from an institution to its members is a generic move in democratic discourse. But paired with the surrounding framing (the “single goal” attribution from the rally, the scope expansion to “much of the region,” the “starving people in this city” causal chain), the complicity-naming move contributes to the Disintegration attribution and to Power: it separates AIPAC and “those who allow it” from the broader American political community by attributing to them sustained malign purpose. The historical record shows that extending culpability from a named Jewish institution to its supporters has been a recurring step in the broadening and normalization of antisemitism. The key issue is not whether the mayor intended that extension. The key issue is that the wording, in combination with the surrounding predicates, makes it available.

One more note. Asked about the “monsters” term, the mayor said: “I used the term to describe all those who are preventing the birth of a new world, not solely AIPAC, but frankly super PACs at large.” That hedge helps him answer the charge of singling out AIPAC. But as noted in the opening of this section, AIPAC remains the only super PAC the rally and the press conference actually name, elaborate, and tie to regional insecurity and to starvation in New York. The hedge does not substantially alter the discourse-analytic reading developed here.

1.8 The contested Jewish reception

The most important evidence is not the phrases in isolation. It is how they were received.

Rabbi Jill Jacobs of T’ruah called the framing part of a “disturbing trend” of left-wing attacks on AIPAC, writing that the lobby “absolutely deserves to be criticized, sidelined, and rejected for its decades of negative influence on American foreign policy” — but “without dehumanizing language, and without hinting at a grand Jewish conspiracy” (Jacobs, Substack, June 22, 2026; quoted in Lapin/JTA). Rabbi Shulman’s concern about the dark-money framing, quoted above, is part of the same intra-progressive pushback. Neither voice comes from the mayor’s usual political opposition; both share substantial political ground with him on many other questions. Their concern matters because they read the language as going beyond ordinary policy criticism.

Rep. Gottheimer’s substitution test — swap ‘AIPAC’ for ‘Jews’ and it’s the oldest antisemitic conspiracy theory in the books — captures the same issue in plain language. That does not settle the argument by itself. But it shows that the discourse made a group-level reading available to Jewish audiences, including Jewish voices outside the mayor’s usual opposition.

This is the kind of ambiguity in which institutional critique can become available for broader group attribution: not because the speaker intends it, but because the wording allows that reading.

2. The Reception

2.1 From speaker to audience

Discourse analysis identifies what language can evoke in audiences. It does not, by itself, establish what audiences did evoke. The reading offered in Section 1 makes a structural claim — that the speech’s predicates overlap with documented antisemitic tropes and make a group-level reading available to audiences carrying the relevant historical repertoire. That claim is testable, and this section describes the empirical apparatus designed to test it.

Instagram and mainstream-press YouTube comment threads provide tractable test environments. Both are public, time-bounded, stratifiable by post stance and platform-architectural position, and large enough to support quantitative analysis of how predicate-clusters propagate across strata of the reception ecosystem. This study applies the Decoding Antisemitism / Decoding Hate coding methodology to both the rally speech and its reception. This section describes the corpus and the methodology and reports the coded findings across three strata: Instagram supportive (S), Instagram critical (K), and mainstream-press YouTube (M). Codable N = 2,396 across 24 posts and threads (1,497 supportive Instagram + 499 critical Instagram + 400 mainstream-press YouTube).

Notation used in this section

2.2 The Dataset

A qualitative reception pool of 38 Instagram posts plus 5 reference posts was assembled in the 5 days following the rally (17–23 June 2026). From this pool, a stratified sample of 20 posts was selected for formal coding: 15 supportive and 5 critical anchor cases. The critical stratum is structured around analytical anchors rather than engagement-proportional coverage; the resulting cross-stratum N-imbalance (1,497 supportive vs. 499 critical) is a feature of that design. A separate mainstream-press YouTube stratum (M) was assembled in parallel, covering all four mainstream-press YouTube videos covering the Mamdani-AIPAC controversy that generated substantial audience engagement (≥100 codable comments) within the same window. Codable N totals 2,396 (1,497 supportive Instagram + 499 critical Instagram + 400 mainstream-press YouTube).

The 24 posts and threads in the coded stratum are listed below. URLs are abbreviated to the Instagram post short-code (full URLs in Appendix A).

Supportive Instagram posts (N = 15), chronological

Critical Instagram posts (N = 5 coded anchor cases), chronological

Mainstream-press YouTube threads (N = 4 coded anchor cases), chronological

2.3 The mainstream news absence in the Instagram corpus

A feasibility scan of high-engagement mainstream news Instagram accounts (Newsweek, NBC New York, CNN, AP, NYT, Gothamist) found that only Newsweek produced a substantive rally-related Instagram post with meaningful audience engagement in the 17–23 June window. The other mainstream news outlets either did not post on the rally or produced posts with very limited Instagram engagement, even where their broadcast or print coverage was substantial.

Whatever the explanation — Instagram strategy, audience behavior, or algorithmic curation — the observed reception of the Mamdani-AIPAC controversy on Instagram was almost entirely partisan-account-mediated. Audiences encountering the rally on Instagram were structurally exposed only to either supportive amplification of Mamdani’s framing or critical identification of it as antisemitic. The “third option” of measured news framing — which exists in the broadcast and print record — did not propagate to Instagram in significant volume. The polarization the Instagram corpus documents is therefore at least partly a function of platform architecture. The mainstream-press YouTube stratum (M1–M4) addresses the YouTube-side reception of the same controversy and is reported alongside the Instagram findings.

2.4 Methods

This study uses qualitative content analysis grounded in pragmalinguistic discourse analysis. The pragmalinguistic premise: meaning in political comment threads is not exhausted by what individual words denote in isolation. It is constructed at the level of predicates and framings — how speakers and commenters attribute properties, motives, agency, and identity through specific phrasing choices, lexical registers, and cross-linguistic equivalences. The analytical question is therefore not whether a given comment contains a “bad word” but whether the comment’s predicate-attribution operations align with documented tropes in antisemitic discourse.

The analytical instrument is the Decoding Antisemitism Lexicon (Becker et al., Springer Nature/Palgrave Macmillan, 2024) — the same reference framework introduced in Section 1 and used here to code audience comments. The Lexicon catalogs forty concept-clusters distinguishing the predicate-level operations of antisemitic discourse (Power, Repulsiveness and Dehumanization, Conspiracy Theories, Evil/The Devil, Disintegration, Blood Libel, Vengefulness, Relativization and Denial of Antisemitism, Instrumentalization of Antisemitism and the Holocaust, and others) from the surface vocabularies they recruit. The methodology has been applied across 300,000+ expert-annotated comments in the project’s prior research cycles.

For each post or thread, the first 100 visible comments are coded against this Lexicon-anchored codebook. The codebook combines primary codes for the antisemitism-relevant tropes with modifier codes (target-specified vs. target-implied; explicit vs. implicit; predicate-level vs. framing-level) and counter-discourse escalation codes (Holocaust analogy, deportation rhetoric, anti-Muslim ad hominem, identity-based slur, jurisdictional delegitimization). The codebook in full is provided in Appendix B.

The coding convention

Each comment in the coded stratum is assigned one of three antisemitism-relevant flags. The flag determines what the headline numbers mean, so the convention is reported here before the findings rather than as a footnote afterwards.

• Explicit activation (E): the comment activates one or more Decoding Antisemitism Lexicon categories through identifiable predicates, framings, or named-target attributions. The Lexicon codes (Power, Conspiracy Theories, Repulsiveness and Dehumanization, etc.) are the diagnostic instrument; activation is determined by predicate, modifier (explicit/implicit, target-specified/target-implied), and framing operation.

• Contextual confirmation (C-AS): the comment does not itself activate a Lexicon category, but it provides substantive M+ engagement that endorses, amplifies, or builds on the post-level framing of an antisemitism-relevant post. Pure affirmation (a “great job” or emoji response under a post coded as antisemitism-relevant) qualifies as contextual confirmation; substantive endorsement of the post’s specific framing does as well. The convention follows the discourse-analytic reasoning that audience-side reception of a post coded as antisemitism-relevant is itself a form of participation in the post’s discursive operation, even when the comment adds no new predicate.

• Anti-antisemitism counter-speech (NO with AS-COUNTER tag): substantive M+ engagement that distinguishes legitimate political critique from antisemitism, defends Jewish people from conflation, or pushes back against the post’s antisemitism-relevant framing — even when the comment opens with affirmation. Under the refined convention, these are not counted as antisemitism-relevant.

The convention is consequential because it determines the denominator. A narrow E-only rate reports only comments that themselves activate a Lexicon cluster; a combined E+C-AS rate also counts substantive engagement with a post coded as antisemitism-relevant as participation in its discursive operation. Both numbers are defensible; they measure different things, and both are reported in the empirical findings below.

One further convention deserves brief notice. When an antisemitic predicate is deployed against an unnamed target in a comment thread whose post-level framing has positioned a particular group as the targeted out-group, the Lexicon code attaches to the implied target rather than to the surface text — but only when four criteria are jointly met: (a) the post-level target is explicit, (b) the comment appears in a sub-thread responding to that target or to a defender or critic of that target, (c) the predicate is historically and lexicographically mapped to the post’s target group within the Lexicon, and (d) no alternative target is more plausible. The four-criterion test confines the convention to comments where the inference is constrained by the discursive context.

2.5 Empirical findings

2.5.1 Supportive Instagram (N = 1,497 across 15 posts)

At a glance — Supportive Instagram (N = 1,497)

Audience uptake of Mamdani’s rhetoric.

Two of Mamdani’s signature phrases were widely taken up in the supportive Instagram stratum: “monsters” (≈23 observed direct-extension instances across S01–S05, N ≈ 500 codable comments) and “dark money” (≈22 instances). The “monsters” predicate was extended from AIPAC toward Israel, Israelis, Zionists, and Jews as a group, often through religious-evil escalation registers (demons, Satan, shaytan, Satanyahoo). The “dark money” predicate was intensified into “Jewish playbook” (S03 010-004 — the central compression from organization-level critique to Jewish-agency framing), “AIPAC mafia” (S02 031), “blood-stained shekels” (S01 030-001), and broader claims of Jewish/Israeli control of American politics. Two phrases Mamdani also used at the rally — “single goal of turning us against one another” (Disintegration) and “fear of peace” (Evil/Vengefulness) — produced no observed direct uptake across the same N ≈ 500. The pattern is selective: the phrases that named a moral antagonist in absolute terms propagated; the longer attributive phrases describing actions or motivations did not.

Overall rates.

Across the supportive Instagram stratum, the coded comments distribute as follows:

The most important figure for interpreting these findings is the 22% explicit Lexicon activation rate: this is the share of supportive Instagram comments that themselves activate one or more Lexicon categories under strict coding criteria, independent of the C-AS convention. The 53% contextual-confirmation rate measures substantive endorsement of a post coded as antisemitism-relevant, under the C-AS convention. The 75% combined rate measures the share of supportive Instagram comments that participate in the antisemitism-relevant reception ecology under the study’s coding convention. The 75% should not be glossed as “75% antisemitic comments”: it is a reception-level measure that includes substantive endorsement of antisemitism-communicating posts without independent predicate-activation. These figures should therefore be interpreted as follows: the 22% is the strictest measure, the 75% is the broadest reception-level measure, and the 53% is the convention-dependent middle term.

Per-post rates range from 58% (S02, S14) to 96% (S12), a 38-percentage-point spread that correlates with post-framing emphasis and the politically homogeneous composition of the comment audience. The five dominant Lexicon clusters across the stratum are, in order, Power, Repulsiveness and Dehumanization, Relativization and Denial of Antisemitism, Disloyalty/Jewish Loyalty, and Conspiracy Theories. Substantial secondary activations are documented for Genocide, Instrumentalization of Antisemitism and the Holocaust, Blood Libel/Child Murder, and Denial of Israel’s Right to Exist.

Five audience-side mechanisms.

Five distinct audience-side mechanisms account for the bulk of Lexicon activations in the supportive Instagram stratum. Each is illustrated below by representative examples drawn from the coded stratum; the full per-comment catalog is in the coding files. The mechanisms are ordered by analytical centrality rather than by frequency: the “Jewish playbook” compression (Mechanism 2) is the corpus’s most consequential single instance.

Mechanism 1 — Dehumanization through Nazi-era propaganda registers. Audience comments activate the Lexicon’s Repulsiveness and Dehumanization cluster through three lexical registers traceable to Nazi-era propaganda: parasite imagery, organized-crime imagery, and religious-evil escalation. The parasite register recurs across four posts in four lexical forms — parasite (S01), tick (S10), lice (S11), leech (S14) — with each form applied to Israel, to Zionists, or to a Jewish-coded political target. The religious-evil escalation appears most explicitly at S02 (“’monsters’ is too lenient of a term. I’d go with ‘demons’ instead”), with the Arabic-language shaytan and the Satanyahoo portmanteau (Satan + Netanyahu) showing that the same escalation pattern appears across languages. The organized-crime register surfaces as “biggest mob in the world” at S14, and the excrement register reaches its most explicit form at S15 014-005 against a named Jewish-religious target. Taken together, these examples suggest that the supportive Instagram stratum draws on several elements of the Lexicon’s cataloged dehumanization repertoire; Mamdani’s “monsters” baseline is intensified by audience-side registers without additional speaker prompting.

Mechanism 2 — Conspiracy-network expansion: the “Jewish playbook” compression. The Power and Conspiracy Theories clusters together constitute the densest activation register in the supportive Instagram stratum. Mamdani’s organization-level “dark money” framing is extended by audience comments in two structurally distinct directions. The first is broadening: AIPAC is retained as the named actor but the scope is expanded to claims of Jewish-American or Israeli control of US political institutions (“AIPAC controls both parties of Congress,” “93 of 100 senators are owned and funded by Israel”). The second is compression: AIPAC as the mediating organization-level frame is dropped entirely, and Jewish-agency is named directly without scaffolding. The corpus’s most consequential single audience-side activation is the two-word compression at S03 010-004: “Jewish playbook.” This comment performs three discursive operations in two words. AIPAC as the mediating frame is dropped. The Power cluster’s coordinated-strategic-agency predicate (”playbook”) is attached to “Jewish” as a group-level attribute. And the comment requires no contextual scaffolding — the commenter assumes the reader will track the implication. A single empirical instance can nevertheless be analytically central when its structure matches the mechanism predicted by the discourse-analytic framework — and when that structure is observable in two words.

The historical conspiracy compounds operate alongside the broadening and compression patterns. The JFK-AIPAC assassination conspiracy (”Kennedy tried that, months later he died”) is concentrated at S05 with the grammaticalized verb-form “JFK him” and recurs at S15 003-002 and at K3 005-001/005-009 — cross-stratum and bidirectional. The Rothschild-Balfour conspiracy appears at S09. The Jewish-Bolshevism trope (a canonical early-20th-century anti-Jewish conspiracy claim foundational to Nazi propaganda) appears at S15 033 in the formulation “a communist is calling out rich Jews who authored communism.” The synagogue of Satan + Khazar biblical-Christian-AS compound appears at S11 038, and the Chabad-Lubavitch three-branch deep-state framing at S14 002-002 positions a named Jewish religious movement as one of three globally-distributed power centers alongside the Vatican and the British Crown. The cross-stratum cross-direction sub-patterns documented for JFK-AIPAC and Soros confirm that these predicates are not specific to progressive-left politics; they are available across the political spectrum and the speaker’s political direction is not the operative mechanism.

A related single-post concentration at S14 frames Israel and AIPAC as a political-assassination apparatus — 6 explicit activations including “If anything happens to him for speaking the truth, we will know Israel is behind it” (007) and “I’m afraid that they might put a hit on him” (016) — anticipatory victim-blaming that pre-attributes any future violence against Mamdani to Israeli or Jewish agency. The same Conspiracy/Dehumanization apparatus accounts for both the cross-stratum diffusion of “Jewish playbook” and this single-post concentration.

Mechanism 3 — Blood Libel through child-harm essentialization. Comments that report wartime casualty figures are not coded under the Lexicon’s Blood Libel / Child Murder cluster; comments that essentialize, ritualize, depict child harm as a site of Jewish/Israeli sadism, or attribute Jewish/Israeli essence-based agency to child harm are. These cases are coded as Blood Libel only where the comment moves from casualty or atrocity allegation to essentialized, collective, or demonizing attribution. The conservative coding convention yields a small but documented set of activations. The cross-thread libel of famine spam pattern from a coordinated spam-account network deploys identical text across S04 and S06, attributing child-starvation to deliberate Israeli essence. The “Doctors find IDF sniper bullets in children’s heads” framing at S04 essentializes child harm as deliberate agency. The pedophile-predicate extension at S13 037 (“pdf file nation,” a homophone for pedo file) extends the Blood Libel register through cross-activation with Repulsiveness and Dehumanization and Evil/The Devil. The blood-libel cluster is the corpus’s highest-risk coding territory because casualty-reporting and blood-libel coding share surface features; the conservative convention is the codebook’s response.

Mechanism 4 — State-erasure slogans (bidirectional). The Denial of Israel’s Right to Exist cluster is activated by state-erasure slogans, most clearly MAKE ISRAEL PALESTINE AGAIN / MIPA. The codebook treats this cluster symmetrically: pro-Israel reversals in the qualitative pool (“Israel is Palestine, the name was stolen from the Jews”) perform the same formal operation in the opposite direction.

Mechanism 5 — Audience-side correction toward collective attribution. Even when the speaker exercises restraint in the direction of leader-specific framing — Mamdani’s deliberate choice of “Netanyahu’s wars” rather than “Israel’s wars” — audience-side discourse pulls the framing back toward collective-Israel or collective-Jewish attribution. The clearest case is S15 035: “israel’s wars not Netanyahu’s” — six words that replace the political-leader actor with collective-Israel attribution and activate the Holding Jews Collectively Responsible cluster. A related bidirectional case at S15 038 (“He is the monster in office to push his own radical agenda”) redirects Mamdani’s “monsters” predicate at Mamdani himself from an M- commenter, indicating that audience-side vocabulary propagation is not unidirectional. The mechanism provides empirical support for the audience-side availability claim of Section 1: speaker-side restraint is not preserved by audiences carrying the relevant historical repertoire.

Two meta-discursive sub-patterns warrant brief mention. Filter-evasion variants (israHELL, Isnotreal, Zio, zionazi, Hasbara bot, and approximately ten further forms) bypass Instagram moderation while still activating the underlying Repulsiveness and Dehumanization predicate; the pattern is platform-architectural rather than analytically new. More substantive is the post-Holocaust regulatory compound observed at S07: M+ users who raise antisemitism-structural concerns about Mamdani’s framing are met by other M+ users who deny the substance and impute bad-faith weaponization (a Relativization and Denial + Instrumentalization of Antisemitism and the Holocaust compound), and in one case M+ commentary frames AIPAC as “worse than Nazi Germany” while claiming the historical Nazis had a legitimate basis for their anger at German Jews — a Holocaust Distortion + Nazi Analogy + Blame for Antisemitism compound. The S07 006 sub-thread (fifteen reply comments, including a citation to Norman Cohn’s Warrant for Genocide) shows the regulatory function in extended form: scholarly anti-antisemitism counter-discourse exists within supportive Instagram threads, but it is actively contested from within.

2.5.2 Critical Instagram (N = 499 across 5 anchor cases)

At a glance: Critical Instagram (N = 499)

Audience uptake of Mamdani’s rhetoric.

Mamdani’s signature predicates also propagated in the critical Instagram stratum, but in reversed form: they were redirected at Mamdani himself. The “monster” predicate was taken up in counter-framing in approximately 47 instances across the five posts (~9% combined: K1 ~22%, K2 ~15%, K3 ~7%, K4 ~1%, K5 ~0%) in the form “He’s the monster,” “Mamdani is the monster of our times.” Beyond the simple counter-framing reversals, two of the critical-stratum’s most explicit cases identify the dehumanizing function of the “monster” predicate by naming its historical antecedents: K2 044 performs an audience-side substitution test in first-person (“As one of these ‘monsters’, I call out his blatant antisemitism!”), and K3 064 names historical antisemitic iconography directly (“the Judensau,” “Der Stürmer,” “1930s Germany”). The “dark money” predicate was redirected at Mamdani-funders (Qatari government, CAIR-affiliated PACs, DSA) in five M- counter-deployments across K3, K4, and K5. The bidirectional propagation finding is the central cross-stratum result on rhetorical uptake: appropriable predicates are appropriated by audiences on both sides — the supportive Instagram stratum extends them toward AIPAC and Jews as a group, the critical Instagram stratum reverses them at Mamdani. The two phrases that did not propagate in the supportive stratum — “single goal of turning us against one another” (Disintegration) and “fear of peace” (Evil/Vengefulness) — also show little to no M- counter-uptake here (1 observed instance of “single goal” counter-uptake at K2 002; zero for “fear of peace”).

Overall rates.

Across the five critical Instagram posts, the coded comments distribute as follows:

The 1.4% explicit Lexicon activation rate is approximately one-sixteenth of the supportive Instagram baseline (22%) — a 16-fold cross-stratum gap that holds across post-architectural diversity (Jewish-journalistic, Jewish-individual, institutional-Jewish-advocacy, right-coded mobilization, institutional-hybrid). All five posts produce E rates below 5%. The contextual confirmation category (C-AS) does not apply in critical post-ecologies: the post-level framing diagnoses antisemitism rather than activating it, so substantive M+ engagement endorsing the post-level framing is anti-antisemitism counter-speech rather than participation in an antisemitism-relevant discursive operation.

Audience-side dynamics.

The 98.6% NO share is not analytically empty. Critical comment ecologies escalate using historical-discursive vocabularies that do not activate Lexicon antisemitic predicates against Jews as a group: anti-Muslim and anti-immigrant attacks drawing on Mamdani’s Ugandan background (~10% of critical Instagram codable); Hitler-rhetoric and Holocaust-analogy positioning Mamdani as a contemporary perpetrator (~3%); Jewish-internal-Holocaust-slur compounds (Kapo, cattle car) applied to Jewish-progressive Mamdani-supporters like Bernie Sanders and Brad Lander (~2%); deportation and denaturalization rhetoric drawing on Mamdani’s Ugandan birthplace (~2%); religious-eliminationist framing (Amalekite, Yimach Shemo — ~1%). These vocabularies escalate against Mamdani personally rather than against Jews as a group; they are analytically distinct from the antisemitic tropes the supportive Instagram stratum activates.

Two cross-stratum cross-direction sub-patterns are worth noting alongside the main finding. The JFK-AIPAC assassination conspiracy and the Soros conspiracy both appear from M- direction in the critical Instagram stratum (K1, K2, K3) as well as from M+ direction in the supportive stratum (S05, S13, S15). The pattern documents that some Lexicon categories operate independently of post-stance direction: they activate where the historical-discursive material matches an audience-side repertoire, regardless of the political stance of the comment ecology.

In this corpus, the critical Instagram stratum is not where antisemitism-relevant activation is concentrated. Its 1.4% E rate is the cross-stratum benchmark against which the supportive Instagram stratum’s 22% and the mainstream-press YouTube stratum’s 42.5% acquire their analytical significance.

2.5.3 Mainstream-press YouTube (N = 400 across 4 threads)

The mainstream-press YouTube stratum shows the highest antisemitism-relevant activation rate documented in the present study. Two methodological observations frame the empirical findings below. First, the mainstream-press YouTube comment ecology around the Mamdani rally is not a moderating third reception environment but one characterized by comparatively high levels of antisemitism-relevant activation. Second, audience commenters reproduce Mamdani’s specific word choices — “monsters,” “dark money,” “genocide” — at high rates throughout the four threads.

At a glance — Mainstream-press YouTube (N = 400)

Audience uptake of Mamdani’s rhetoric.

Three of Mamdani’s signature phrases were widely taken up in the mainstream-press YouTube stratum. The “monsters” predicate produced approximately 22 M+ direct-extension instances and approximately 8 M- counter-uptake instances, extending from AIPAC toward Israelis, Jewish billionaires, and Jews as a group, often with escalation: “Anyone who supports Zionism-infected genocidal, ethnic cleansing, expansionist Israel is a monster” (M3 014 — extends the predicate to Israel-as-state with a DEH “infected” + GEN + COL compound); “Stop questioning the people who condemn a genocide and start questioning and attacking those who support a genocide like THE DEVIL MONSTOR AIPAC” (M1 0035 — adds EVIL escalation); “AIPAC is a monster. Unfortunately, it’s not the fun kind that breathes fire and fights robots. It’s the kind that shows up in a suit and funds genocide with American tax dollars” (M4 069 — adds POW + GEN); “Apologize? These demons don’t get phase” (M3 069 — religious-evil escalation). The “dark money” / Power-cluster framework propagated as 98 POW activations across N = 400 (~25% of comments), extending to Jewish billionaires, Jewish media, and Israeli foreign-influence framings: “Foreign lobbies don’t belong in politics, and AIPAC is probably the worst offender” (M1 0014 / M2 007 — same commenter cross-posting); “AIPAC has overthrown the American government, anyone supportive of the organization is treasonous” (M3 004 — CON-WD + DSL); “a lot of the billionaires who buy the politicians who fund Israel’s atrocities happen to be Jewish” (M3 063 — Jewish-billionaires TGS); “FINALLY, someone has the balls to call AIPAC the genocidal bankroll murders they are” (M1 0049, 1,655 likes — DEH + GEN + POW compound). The “genocide” / “children being killed” framework produced approximately 36 GEN + 10 BLO activations across the stratum; “21,000 children bomb to death. That’s antisemitism? Isn’t that the truth?” (M4 048 — GEN + BLO + RDA compound); “AIPAC is genocide” (M4 015 — direct predicate-attribution); “We have a media that is accepting the genocide so naturally they’re accepting of child molestation and sex trafficking” (M3 086 — GEN + BLO compound). The two phrases Mamdani also used at the rally — “single goal of turning us against one another” (Disintegration) and “fear of peace” (Evil/Vengefulness) — produced no observed uptake across N = 400. The selective-uptake pattern is now supported across N = 2,396 total codable comments combining all three reception strata: Mamdani’s signature phrases that named a moral antagonist in absolute terms (”monsters”) or alleged corrupt political power (”dark money”) were widely reproduced across every reception environment, while his longer attributive phrases describing actions or motivations produced no observed direct uptake anywhere in the coded corpus.

Overall rates.

Across the four mainstream-press YouTube threads (Forbes Breaking News, CNN, NBC New York, CBS New York; 309K total views; 3,331 total comments; first 100 coded per thread), the coded comments distribute as follows:

Across 400 mainstream-press YouTube comments, 321 (80.25%) participated in the antisemitism-relevant reception ecology. Of these, 170 (42.5%) deployed Lexicon predicates explicitly; 151 (37.75%) endorsed or amplified the post’s framing without independently activating predicates. Per-thread distribution:

The thread with the highest antisemitism-confirming rate (M2 CNN, 91%) and the thread with the lowest (M3 NBC, 70%) both fall within the supportive Instagram stratum’s range (58–96%). Mainstream-press YouTube does not function as a moderating reception environment for this controversy. Substantive AS-COUNTER comments across N = 400: zero. No comment substantively rejected the antisemitism-relevant framework; comments that questioned individual predicates (e.g., dehumanizing language) while confirming the broader conspiracy framing were coded as confirmation (YES C-AS), not as substantive AS-COUNTER.

The cluster-activation density across N = 400 follows the same hierarchy observed in the supportive Instagram stratum: POW (98 activations, ~25% of comments), RDA (46), OF (~38), GEN (~36), DEH (~33), LIE (~23). Because comments may activate multiple tropes simultaneously, cluster frequencies exceed the number of coded comments. The “AIPAC = corrupt political power” framing is the structural anchor of the mainstream-press YouTube reception.

Audience-side dynamics.

The mainstream-press YouTube stratum exhibits a polarization spectrum from M2 CNN (4% M-) to M3 NBC (29% M-) — but the polarization does not produce moderation. Smaller-engagement threads concentrate M+ antisemitism-relevant activation; larger-engagement and Jewish-institutional-foregrounded captions attract more substantive M- voices alongside the dominant M+ reception. The M- pushback does not move the reception toward measured news framing: it produces direct M- antisemitism allegation against Mamdani alongside the dominant M+ antisemitism-confirming reception. The mainstream-press YouTube stratum is, in this sense, the polarized middle, not a moderating one.

Three further sub-patterns warrant brief mention. Cross-direction M- antisemitism activations — antisemitic predicates deployed against Mamdani and Jews from M- direction — appear at M3 074 (Soros + Nazi-analogy compound) and M3 085 (anti-Jewish economic-flight framing); Lexicon predicates activate from both M+ and M- direction across all three reception strata. The Khazar-identity-denial framework propagates at 5 activations (M1 0060, M1 0071, M3 021, M3 099, M4 011) — the typical form: “Jews aren’t really Semites, they’re European converts — so anti-Zionism cannot be antisemitism” — and notably does not appear in the supportive Instagram stratum. Named-Jewish-individual target-specification escalates by thread, correlating with caption framing: M3 NBC (Jewish-institutional caption) activates ADL, Gottheimer, Schumer, Jeffries, “Jewish billionaires,” “Jewish media,” and “Golem newscasters” as named targets; M4 CBS activates Bari Weiss, the Ellison family (Paramount-CBS ownership), and Marcia Kramer; M2 CNN and M1 Forbes show minimal named-target activation by comparison.

2.6 Audience uptake of Mamdani’s rhetoric across all three strata

A coded measurement of speaker-language uptake — how often the specific phrases Mamdani used at the rally and the press conference reoccur in audience comment threads, and how those phrases propagate — provides a way of empirically grounding the discourse-analytic reading from Section 1. The reading there identifies six recurring phrases and framing moves and maps each to one or more Lexicon categories. The empirical question is whether those phrases are picked up and extended by audiences in the reception, or whether they remain speaker-bound.

Coded data across all 24 posts and threads in the corpus (N = 2,396 codable comments combining 1,497 supportive Instagram + 499 critical Instagram + 400 mainstream-press YouTube) supports the selective-uptake hypothesis across all three strata. The two phrases Mamdani emphasized most heavily at the rally — “monsters” and “dark money” — propagate in all three reception environments, in both M+ and M- direction. The two phrases this essay identifies as antisemitism-relevant in concept but not in vocabulary-matching form — the “single goal” / Disintegration framing and the “fear of peace” / Evil/Vengefulness framing — show no observed direct uptake in any of the three strata.

The per-stratum speaker-language uptake findings are reported in the three subsections above. The cross-stratum summary below captures the bidirectional pattern across all three reception strata:

Two cross-direction sub-patterns are documented across the Instagram strata: the JFK-AIPAC conspiracy framework (S05 concentrated cluster, S15 003-002; K3 005 M+ within critical thread) and the Soros conspiracy framework (S13 within politically-diverse M- ecosystem; K1 022-001, K2 030, K3 006-001 from M- direction). The mainstream-press YouTube stratum adds the Khazar-identity-denial framework (5 activations across M1, M3, M4) as a further cross-stratum sub-pattern absent from the supportive Instagram stratum.

2.7 The main tropes in the comment sections

This subsection synthesizes the audience-side activations across all three strata by predicate-cluster, distinguishing dominant clusters, clusters with near-zero uptake, and counter-discourse registers.

Power and Conspiracy Theories together constitute the densest activation register across all three strata: 50+ explicit activations in the supportive Instagram stratum and 98 Power-cluster activations in the mainstream-press YouTube stratum (~25% of comments). Audience extensions take three documented forms: broadening from organization-level critique to claims of Jewish-American or Israeli control of US institutions; compression from organization-level framing to direct Jewish-agency framing (the “Jewish playbook” anchor at S03 010-004); and named-Jewish-individual target-specification (Schneerson, Chabad, Tisch, Rothschild, Soros, Brad Lander). Repulsiveness and Dehumanization is the corpus’s second-most-activated cluster, with parasite imagery (parasite S01, tick S10, lice S11, leech S14), organized-crime imagery, and religious-evil escalation (demons / Satan / shaytan / Satanyahoo) operating across languages. Two conspiracy frameworks — Soros and JFK-AIPAC — appear in both supportive and critical strata, indicating that some antisemitic predicates activate independently of the audience’s political stance toward Mamdani.

Two of the speaker’s predicates that Section 1 identifies as antisemitism-relevant produce near-zero audience-side direct echo. The Disintegration cluster activated by Mamdani’s “single goal of turning us against one another” receives one observed counter-uptake (K2 002 reverses it at Mamdani) and otherwise no propagation across all three strata. The Evil/Vengefulness “fear of peace” predicate receives zero observed uptake anywhere in the coded corpus. The Evil/Devil cluster does activate strongly through the religious-evil escalation of “monsters,” but not through “fear of peace” itself — the predicate that would have carried the cluster propagates only through a different speaker-side phrase. These are among the corpus’s clearest cases of selective uptake.

The critical Instagram stratum documents counter-discourse escalation registers that are analytically distinct from Lexicon antisemitic trope activation. The anti-Muslim register appears in approximately 52 instances across the five critical posts (~10% of critical-stratum codable), spanning Mamdani-as-foreign-Muslim-infiltrator framings, CAIR / Muslim Brotherhood conspiracy framings, and anti-Arab racial slurs, with deportation and denaturalization rhetoric in approximately 10 further instances drawing on Mamdani’s Ugandan birthplace. The Holocaust analogy register, prominent in the critical stratum, deploys Nazi-equation against Mamdani (K4 009’s “Sounded better in the original German” at 313 likes), Jewish-internal-Holocaust-slur compounds against Jewish-Mamdani-supporters (11 cases across K1 and K2), and religious-eliminationist designations (Amalekite, Yimach Shemo). A jurisdictional delegitimization frame (”the mayor of NYC shouldn’t be talking about Israel”) appears across both Instagram strata, and the Knicks-jersey appropriation reading appears at K4 010 and K3 059 as a recurring critical-stratum sub-pattern. The coding convention treats these registers as analytically distinct from antisemitic-trope activation: the Jewish-safety concern about Mamdani’s language and the anti-immigrant rhetoric mobilized in defense of it are different objects, and they can co-occur without being the same.

A move that recurs across strata without forming its own Lexicon cluster is complicity-naming: in the supportive Instagram stratum, Mamdani’s “those who allow it” framing extends into calls for investigation, prosecution, and removal of “those who support AIPAC,” “the Jewish lobby,” “the Zionist infiltration”; in the critical Instagram stratum, the inverse move extends culpability from Mamdani to his supporters (K3 065) and to his immediate family (K2 003 sub-thread; K5 008-007, 010-011). The move contributes to Disintegration and Power attributions in combination with surrounding predicates but is not a standalone cluster.

3. What the evidence shows

The reception analysis across all three strata (N = 2,396 codable comments across 24 posts and threads: 1,497 supportive Instagram + 499 critical Instagram + 400 mainstream-press YouTube) supports the patterns anticipated by the discourse-analytic reading in Section 1 and surfaces several cross-stratum sub-patterns that warrant separate documentation.

First, the audience-side availability claim from Section 1 receives empirical support across all three strata, with an unevenness that is itself analytically important. Explicit Lexicon activation rates run from 1.4% (critical Instagram) to 22% (supportive Instagram) to 42.5% (mainstream-press YouTube). The 16-fold cross-stratum gap between critical and supportive Instagram holds across post-architectural diversity — Jewish-journalistic, Jewish-individual, institutional-Jewish-advocacy, right-coded mobilization, institutional-hybrid critical posts — indicating that these high activation rates are post-stance-dependent. The mainstream-press YouTube rate of 42.5% (combined 80.25%) falls within the supportive Instagram per-post range but with a doubled explicit activation rate. The speaker's vocabulary appears to have supplied some of the lexical material through which antisemitism-relevant framings entered each reception ecology, even as commenters also drew on pre-existing framing repertoires.

Second, the diagnostic operations that this essay performs in Section 1 are independently performed by audience-side commenters in the critical Instagram stratum. At least seven specific instances are documented across the critical stratum, covering the substitution test, the trope identification, the historical-genealogy mapping (with one commenter at K3 022 invoking Mein Kampf and The Protocols of the Elders of Zion as predicate sources), the Weimar / 1930s-Germany historical analogy, the linguistic-manipulation diagnostic, the downstream-effects argument, and the predicates-to-offline-violence framing (referencing the Montreal incident at K5 016). The framework’s diagnostic operations are not unique academic constructions; they are also how critical audiences themselves read political language of this kind.

Third, the bidirectional propagation of Mamdani’s signature predicates — and the selective-uptake pattern that accompanies it — is the corpus’s strongest single empirical finding on rhetorical uptake. The “monsters” predicate was widely reproduced in M+ direct-extension form in the supportive Instagram stratum (~9% per larger post), as M- counter-framing in the critical Instagram stratum (~9% combined, ~47 instances), and in both directions in the mainstream-press YouTube stratum (~22 M+ extensions + ~8 M- counter-uptakes). The “dark money” predicate was reproduced as M+ intensified extension in the supportive Instagram stratum (with the analytically central two-word “Jewish playbook” compression at S03 010-004), as M- counter-deployment in the critical Instagram stratum (five Qatari/CAIR-PAC/DSA counter-deployments across three posts), and as M+ Power-cluster intensification in the mainstream-press YouTube stratum (98 activations, ~25% of comments). The rally’s signature predicates were taken up by audiences on every reception side: audiences are not passive recipients of speaker framing — they appropriate the framing and redirect it. Both phrases Mamdani also used at the rally — “single goal of turning us against one another” and “fear of peace” — produced no observed uptake across N = 2,396 codable comments anywhere in the corpus. The observed reception patterns are consistent with the selective-uptake hypothesis across all three strata: appropriable predicates are appropriated by audiences on every side; non-appropriable predicates show no observed uptake in either direction.

Fourth, the predicate-clusters at issue are not specific to progressive-left politics but are available across the political spectrum. The cross-stratum cross-direction sub-patterns documented for the JFK-AIPAC and Soros conspiracy frameworks — appearing in both supportive and critical strata — make this visible empirically. S13 (a right-coded supportive account) and K3 (a right-coded critical account) both attract cross-spectrum conspiracy activations. The implications are analytically important: the predicates themselves, not the speaker’s political direction, are the operative mechanism. The same predicate-availability cuts across the Jewish-internal contestation visible in the corpus, in which Jewish accounts on both sides of the controversy mobilize Jewish identity as the authority-anchor for opposite analytical positions — a feature of the discourse environment that the framework navigates by anchoring the diagnosis in specific predicate-clusters with documented historical genealogies, rather than in claims about who counts as a legitimate Jewish voice.

Fifth, the platform-architectural observation about mainstream news framing is double-sided. Mamdani’s rally generated extensive broadcast and print coverage, much of it analytically careful and much of it engaging the substantive predicate-level questions this article addresses. That coverage did not propagate to Instagram in meaningful volume — audiences receiving the rally on Instagram encountered it almost entirely through partisan-advocacy accounts on either side, without the moderating presence of mainstream news framing. Where mainstream-press coverage did propagate on YouTube, however, the comment ecology did not function as a moderating reception environment: no substantive anti-antisemitism counter-speech was observed across the 400 coded comments, and the combined antisemitism-relevant rate of 80.25% sat within the supportive Instagram range while the explicit activation rate doubled it. What the platform-architectural finding collectively shows is that the digital reception environment around an authoritative political speech is not organized as a moderator-and-partisans triad. The moderator role does not exist in the comment ecologies observed; the partisan modes operate at higher intensities precisely where audience volume is highest; and where mainstream-press coverage attracts the largest audiences, the comment ecology amplifies rather than tempers the antisemitism-relevant framings the rally activated.

Conclusion

This study began with a question about political language and digital discourse. Using the Mamdani rally as a case study, it asked how political language employing demonization and hidden-agency framings is received in contemporary digital discourse environments. Bringing together the discourse-analytic reading in Section 1 and the reception analysis across three coded strata (N = 2,396), the study offers an empirical answer to that question.

The discourse-analytic reading establishes that six recurring phrases and framing moves across the rally and the press conference overlap with documented antisemitic tropes cataloged in the Decoding Antisemitism Lexicon. The reading does not assess speaker intent. It identifies what the language draws on, what it makes available to audiences carrying the relevant historical repertoire, and which Lexicon chapters the predicates engage.

The reception analysis across all three strata shows that the audience-side dynamics this reading predicts are present at measurable rates, with a methodologically robust cross-stratum gap: 22% explicit Lexicon activation in the supportive Instagram stratum (N = 1,497), 1.4% in the critical Instagram stratum (N = 499), 42.5% in the mainstream-press YouTube stratum (N = 400), and 16-fold cross-stratum gaps holding across post-architectural diversity. Section 2’s empirical findings report the cluster-by-cluster results; this conclusion will not repeat them.

Several audience-side mechanisms illustrate these dynamics. Organization-level criticism was sometimes compressed into direct attributions to Jews as a group; leader-specific references were broadened toward collective attributions; and comments raising concerns about antisemitism were themselves contested through denial and instrumentalization framings. These mechanisms illustrate how audience reception can transform speaker-level formulations into broader historical repertoires.

The corpus’s central empirical finding on rhetorical uptake is the bidirectional propagation of Mamdani’s signature predicates across all three strata: “monsters” and “dark money” propagate as M+ direct-extension in the supportive Instagram stratum (toward AIPAC, Jews, and Israel), as M- counter-uptake in the critical Instagram stratum (toward Mamdani himself), and in both M+ and M- directions across the mainstream-press YouTube stratum. The audience-side reception is not passive; audiences appropriate the framing and redirect it. A second cross-stratum finding concerns audience-side performance of the framework’s diagnostic operations: at least seven observed instances across the critical Instagram stratum show audience commenters performing the substitution test, the historical-genealogy mapping, the Weimar / 1930s-Germany analogy, and the downstream-effects argument in their own vocabulary. The reception patterns are consistent with the framework’s expectations not only at the predicate level but also at the diagnostic-operation level.

More broadly, the findings demonstrate that political rhetoric does not remain at the level of the original speaker. It becomes raw material for audiences, who selectively intensify the predicates that resonate with pre-existing historical repertoires. The historical resonance of political language therefore cannot be assessed solely at the level of speaker intention or textual interpretation. It must also be evaluated through the patterns of audience reception that the language generates.

What the analysis cannot do is determine the mayor’s intent. What it can do is identify the structural features of the language and the discursive effect of its reception. The claim is narrow: the substantive political critique of AIPAC’s spending and lobbying activity is available without phrases that overlap with antisemitic tropes. The choice to use those phrases rather than alternatives that do not carry the historical resonance is a rhetorical choice with documented consequences for the reception environment.

The broader frame returns. Political language that works through demonization and hidden-agency framings raises the risk of fragmenting the discourse public along the dividing line the language draws. The Mamdani case is one focused instance. The political critique that prompted these phrases is available without them.

The central finding of this study is not simply that some audience comments were antisemitism-relevant; it is that particular rhetorical predicates proved highly portable across discourse environments. The phrases “monsters” and “dark money” were repeatedly extracted from the original speech, intensified, generalized beyond AIPAC to Jews as a group in many supportive comments, and reproduced across both partisan and mainstream-platform reception environments. Other phrases from the same speech showed virtually no uptake. Political rhetoric therefore does not diffuse uniformly: audiences selectively amplify the predicates that best fit pre-existing historical repertoires, and reception analysis makes those dynamics empirically observable.

The broader methodological claim the case is meant to demonstrate is transferable. What this study describes — the path from speaker to platform to audience appropriation to cross-platform propagation — is a model of rhetorical diffusion, applicable beyond the specific case examined here. At its broadest, the article is a case study in how codes and frames travel from authoritative speakers into online cultures of reception. Political rhetoric can be analyzed at the level of audience reception by tracing how audiences actually receive, transform, amplify, contest, and generalize it across the discourse environments that propagate it. Speakers do not control the predicate-clusters their language may activate. Audiences bring historically conditioned repertoires that can engage with those predicates independently of speaker intent, while platforms shape which framings circulate and which do not. These audience-side traces are observable, codable, and methodologically tractable.

Extending discourse analysis from source texts to reception ecologies therefore produces a different kind of empirical evidence. Rather than relying on close reading alone, it becomes possible to observe how historically resonant predicates are propagated, transformed, generalized, and contested across large-scale digital reception environments — as the bidirectional propagation of “monsters” and “dark money” across this corpus, the 16-fold cross-stratum gap in explicit Lexicon activation (supportive Instagram 22% vs. critical Instagram 1.4%), and the audience-side performance of the framework’s diagnostic operations each illustrate. The case is local. The method is not.

References

Anti-Defamation League. (2026). Audit of Antisemitic Incidents 2025. Released 6 May 2026. https://www.adl.org/resources/report/audit-antisemitic-incidents-2025.

Anti-Defamation League, New York / New Jersey Regional Office. (2026). New York Leads the Nation in Antisemitism, Again, Even as Incidents Fall Nearly 20% in 2025. Released 6 May 2026. https://nynj.adl.org/resources/report/new-york-leads-nation-antisemitism-again-even-incidents-fall-nearly-20-2025.

Becker, M. J., Troschke, H., Bolton, M., & Chapelan, A. (eds.) (2024). Decoding Antisemitism: A Guide to Identifying Antisemitism Online. Palgrave Macmillan / Springer Nature. https://link.springer.com/book/10.1007/978-3-031-49238-9.

CNN. (2026, June 22). Mamdani defends criticism of AIPAC after being accused of antisemitism. CNN Politics. Published 4:15 PM ET. https://www.cnn.com/2026/06/22/politics/mamdani-aipac-new-york-primaries.

Cohn, N. (1967). Warrant for Genocide: The Myth of the Jewish World Conspiracy and the Protocols of the Elders of Zion. Harper & Row.

Federal Election Commission. (2026, May–June). United Democracy Project (UDP) committee filings. FEC Committee ID C00799031. https://www.fec.gov/data/committee/C00799031/ — reported in CNN coverage (cited above).

Jacobs, J. (2026, June 22). On AIPAC and Bogeymen. Substack.

Lapin, A. (2026, June 22). Some of Mamdani’s Jewish allies criticize his use of ‘monsters’ to describe AIPAC. Jewish Telegraphic Agency / New York Jewish Week. https://www.jta.org/2026/06/22/ny/some-of-mamdanis-jewish-allies-criticize-his-use-of-monsters-to-describe-aipac.

Times of Israel staff. (2026, June 22). Some Jewish allies of Mamdani denounce his use of 'monsters' to describe AIPAC. The Times of Israel. https://www.timesofisrael.com/some-jewish-allies-of-mamdani-denounce-his-use-of-monsters-to-describe-aipac/.

U.S. Department of Justice / FBI. (2026, June 16). Criminal complaint unsealed in connection with alleged plot against UFC Freedom 250 event at the White House, charging Tycen Proper, Bryan Omar Roa, Michael Alan Thomas, Daniel Eskridge, and Abraham Hermosillo Alvarez. FBI Director Kash Patel announcement.

U.S. Department of Justice, Civil Rights Division. (2026, June 18). Federal grand jury indictment in United States v. Forrest Kendall Pemberton, Southern District of Florida. Criminal complaint and indictment documents; DOJ Civil Rights Division X announcement, 18 June 2026.

AddressHate Research Scholar at NYU’s Center for the Study of Antisemitism | Lead, Decoding Antisemitism — PI, Decoding Hate | Research Advisor, AddressHate | Editor-in-Chief, Digital Hate Review

Executive Summary

On May 11, 2026, the New York Times published Nicholas Kristof’s opinion column “The Silence That Meets the Rape of Palestinians,” reporting a pattern of sexual violence by Israeli prison guards, Shin Bet interrogators, soldiers, and settlers against Palestinian detainees and civilians. The column appeared one day before the Times covered the release of the Civil Commission on October 7 Crimes by Hamas Against Women and Children — a development whose timing has become publicly contested between the Israeli Foreign Ministry, which says the Commission “approached the Times months ago,” and the Times, which denies advance knowledge.

What follows is not a fact-check of individual allegations. The target is how a prestigious opinion platform constructs moral and institutional conclusions from heterogeneous forms of evidence, and how those conclusions are subsequently amplified and extended by the comment-section readership the platform reaches. The piece proceeds in two main parts.

The deconstruction. Kristof’s column advances three substantive claims with substantively equivalent confidence — that individual acts of sexual violence occurred (Claim 1), that such abuse is widespread, repeated across institutional settings, and unpunished (Claim 2), and that it constitutes “an organized state policy” and “standard operating procedure” of the Israeli state (Claim 3) — plus a framing claim of moral symmetry between the column’s subject and the October 7, 2023 Hamas attack (Claim 4). I argue that Claims 1 and 2 are substantially supported by the column’s sources and by independent corroboration including the Sde Teiman indictments, B’Tselem’s case files, Public Committee Against Torture in Israel complaints, and reform-oriented Israeli analysts (Sari Bashi inside the column; Haviv Rettig Gur and Matti Friedman outside it). Claim 3 is not warranted by the same sources. The UN report Kristof cites characterizes the leadership’s relationship to the abuse as “implicit encouragement”; the column drops the qualifier and renders it as “standard operating procedure” and “organized state policy.” Sari Bashi, the column’s named Israeli human rights source, tells Kristof on the record: “I don’t see evidence that it has been ordered.” After publication, former Prime Minister Ehud Olmert — whose quote the column placed near the end as confirmation — issued a Free Press statement disclaiming the policy framing the column built around his words: “I have no knowledge supporting these claims as I said to Mr. Kristof. Therefore, the positioning of my quote after pages of such allegations misrepresents my views.” Claim 4 (the symmetry frame) collapses substantive differences in scale, coordination, and operational character between the two phenomena the column treats as parallel — differences the column’s own author concedes elsewhere in the piece. Six argumentative moves perform the transformation from Claim 2 to Claim 3: qualifier-dropping, equivalence-by-presentation, selective hearing of named sources, symmetry as ontological claim, structural conditions framed as evidence of policy rather than as the alternative explanation, and institutional placement as authorization. Each is individually defensible as journalistic technique; their combination on a high-salience charge against a state is what produces the inferential architecture under examination here.

The reception. Analysis of 696 unique reader comments — drawn from the publicly visible portion of a moderated thread whose closed-thread footer reads 809 — published directly below the column on nytimes.com shows that the readership overwhelmingly accepted the column’s core argument: 92.5% of comments support the column’s claims, 2.0% reject them, 5.5% are non-evaluative. The substantive disagreement visible in the thread is not whether the abuse Kristof documents is real but how far the conclusions drawn from it should extend. Sixty-five unique comments (9.3%) escalate beyond what the column itself argues — into genocide framing (30 instances), rogue-state characterization (14), settler-colonial framing (7), and in six cases explicit denial of Israel’s right to exist. Concept-tagging using the 40-concept Decoding Antisemitism Lexicon (Becker et al., 2024, Palgrave Macmillan/Springer Nature) suggests a structural pattern worth registering: while state-rejection concepts cluster entirely in the escalation tail as expected, inversion and comparison concepts — Nazi Analogy (93%), Holocaust Inversion (89%), Conspiracy Theories (83%) — cluster in the modal confirmation register of the thread, embedded in comments whose explicit policy demand matches the column’s. The discursive registers most active in the comment thread are operating not at the extreme tail but in the ordinary register of Times-reader confirmation.

The two halves stand in a relation worth registering as a finding in its own right. The framing operations identified in the deconstruction (qualifier-dropping, equivalence-by-presentation, symmetry as ontological claim, and the others) reappear in the discursive afterlife of this column, documented here through its comment-section reception. Historically-loaded antisemitic registers (Nazi Analogy, Holocaust Inversion, Conspiracy Theories) cluster not at the extreme tail of the discourse but at the modal-reader layer. The relationship between deconstruction findings and reception findings is not coincidental. The inferential architecture identified in the first main part is aligned with the comment-section pattern documented in the second. On the evidence available here, the framing and the uptake are aligned in ways consistent with the framing having shaped or facilitated the uptake. The data show correlation, not causation.

The broader analytical point can be stated narrowly. Reporting on individual abuses, even when the abuses are well-documented and the reporting is otherwise solid, does not automatically support the allegation that such abuses are systemic, organized, or constitutive of the institution under scrutiny. The argumentative move from singular event to systemic crime is a distinct inferential step. It is at that framing layer, not at the layer of the underlying reporting, that distortion and bias enter most easily, and that historically-loaded essentializing registers find their opening in the downstream uptake. I am not claiming that this is what individual columnists or editors intend. The claim is that the inferential architecture, once in the public record, operates regardless of intent.

The Kristof column belongs to a small set of prior New York Times coverage episodes — the October 17, 2023 Al-Ahli Hospital headline and the July 25, 2025 Gaza-starvation front page — in which prominently-placed accusations against Israel were followed by Times-issued corrections whose reach was substantially smaller than the original framing. I do not extrapolate beyond these three cases; what they illustrate is how, under contemporary editorial conditions, high-salience accusations can in particular instances outpace verification standards, with correction mechanisms operating more slowly than the mechanisms that produced the original framing. The Kristof column can reasonably be read as continuous with that small set of prior cases. The claim goes no further than that.

Introduction

On May 11, 2026, the New York Times published Nicholas Kristof’s opinion column “The Silence That Meets the Rape of Palestinians.” The piece reports a pattern of sexual violence by Israeli prison guards, Shin Bet interrogators, soldiers, and settlers against Palestinian detainees and civilians. It assembles fourteen first-person testimonies, three NGO and UN reports, two surveys, named indictments from the 2024 Sde Teiman case, and an on-the-record interview with former Israeli prime minister Ehud Olmert. Its central rhetorical structure is an appeal to symmetry: supporters of Israel rightly condemned the sexual violence of October 7, 2023, and consistency requires applying the same standard now.

The column appeared one day before the release of the Civil Commission on October 7 Crimes by Hamas Against Women and Children. This is a 300-page report drawing on over 10,000 photographs, thousands of hours of video, and more than 400 testimonies, much of it sourced from Hamas’s own GoPro footage. The Commission found October 7 sexual violence to be “systemic, widespread, and deliberate,” “carried out with exceptional cruelty,” and introduced the legal concept of kinoide for forcing family members to witness or perform sexual acts against one another. Former Canadian Justice Minister Irwin Cotler served as a principal contributor to the report. Named endorsers include Sheryl Sandberg, Hillary Clinton, Alice Wairimu Nderitu (former UN Special Adviser on the Prevention of Genocide), David Crane (former chief prosecutor of the UN Special Court for Sierra Leone), and Aharon Barak (former President of the Israeli Supreme Court). The Commission’s report is titled “Silenced No More.” Kristof’s column, published the day before it, is titled “The Silence That Meets the Rape of Palestinians.” The titular relationship between the two pieces, published twenty-four hours apart in the same week, is a discursively significant fact worth recording without adjudicating intent.

Whether the New York Times had advance knowledge of the Commission’s release timing is disputed on the record. The Israeli Foreign Ministry has stated that the Commission “approached the Times months ago regarding its then-forthcoming report” and that the publication “said it was not interested.” Michal Cotler-Wunsh, former Israeli Special Envoy for Combating Antisemitism and now CEO of the International Legal Forum, has characterized the sequencing in a JNS interview as a “complete burial of the commission report” and argued that the Kristof column “took up all the airspace.” Times spokesman Charlie Stadtlander has stated that the Times “never passed on the Civil Commission report and wasn’t told about its completion or the timing of its release” and that the Commission’s work “had no bearing on Nicholas Kristof’s opinion column or its publication timing.” The two positions are in substantive tension: “approached about a forthcoming report” and “told about its completion or timing of release” describe different events, but the institutional question of whether the Times engaged with the Commission’s existence before publishing Kristof’s column is now publicly contested. What is not disputed is the temporal fact: the column was published first, and the Commission’s findings entered the news cycle nearly twenty-four hours later, during the period in which the UN Secretary General was weighing which state actors to place on a sexual-violence blacklist. I record the dispute as part of the case file without adjudicating it; what is analytically relevant for what follows is the temporal sequencing of two publications and the differences in reach between them, which obtain regardless of intent.

Within forty-eight hours the column had drawn detailed response from multiple Israeli and Israeli-American voices: Haviv Rettig Gur, Matti Friedman, Dan Senor, as well as a clarifying statement from Olmert himself distancing his on-the-record words from the column’s strongest claims. The reception in the New York Times’s own comment thread, the public-discourse trace the column left in the venue it was published in, is the subject of the second main part below.

What follows proceeds in two main parts. The first deconstructs the column itself: what it argues, what its sources support, and where the inferential structure separates from the evidentiary structure. The second documents how that column was received in the comment thread published directly below it on nytimes.com: which claims the readers carried forward, which they let drop, and what registers of antisemitism-adjacent discourse saturated the uptake.

I distinguish three substantive claims and one framing claim, enumerated in the dedicated claims section below. I accept that the abuses Kristof reports occurred and that the case for them being widespread and unpunished is substantially supported. Claim 2 is corroborated independently through the Sde Teiman indictments, B’Tselem’s case files, the Public Committee Against Torture in Israel’s hundreds of complaints, Breaking the Silence testimony, and on-the-record acknowledgment from former and current Israeli officials. Reform-oriented Israeli analysts (Rettig Gur, Friedman) arrive at this assessment without needing the symmetry frame Kristof imposes.

My argument is that Kristof’s piece does something beyond establishing the existence of the problem. This additional work, performed by editorial and framing choices rather than by the underlying evidence, slides the column from claim 2 to claim 3 in ways the evidence does not support and the column’s own sources explicitly contradict. Sari Bashi, executive director of the Public Committee Against Torture in Israel, tells Kristof on the record: “I don’t see evidence that it has been ordered. But there’s persistent evidence that the authorities know it’s happening and are not stopping it.” That is a precise statement of claim 2. The column prints it. It then proceeds, in Kristof’s own voice, to assert claim 3. Olmert, after publication, issued a Free Press statement explicitly denying that his quoted words supported the column’s stronger conclusions.

The distinction between condoned and organized is not a quibble. It determines what reform would address and how the Israeli state itself is to be characterized. It also determines what the discourse downstream of the column, in the comment thread, the topic of the second main part below, will repeat, escalate, and inscribe into the public record.

A note on scope. What follows does three things and no more. It accepts that individual acts of sexual violence by Israeli personnel against Palestinians have occurred, and that there is substantial evidence of repeated, unpunished abuse across institutional settings. It examines, narrowly, whether the further step taken in Kristof’s column — that these abuses constitute an “organized state policy” and “standard operating procedure” — is warranted by the sources he cites. And it places that examination within the discourse context in which the column was written, edited, and read. I do not claim that Kristof or his editors acted in bad faith, that the New York Times intentionally amplifies anti-Israel narratives, or that all advocacy-sourced reporting is structurally compromised. Those would be different arguments requiring different evidence. My argument here is narrower and, for that reason, harder to dismiss: a specific inferential move in a specific column, examined against the column’s own sources.

The analytical question raised through this case is general even if the evidence is local. The move from reporting individual abuses to alleging systemic, organized state crime is an inferential step that the underlying reporting does not automatically license. It is at that step, not at the level of the documented misconduct, that framing, generalization, and downstream essentialization gain their structural opening.

What follows proceeds in five steps. It sketches Kristof’s positioning as a columnist and the authority readers project onto the byline; specifies the substantive and framing claims advanced here; tests those claims against the sources the column itself cites; situates the column within a small set of recent New York Times coverage episodes; and turns finally to the comment thread published directly below the column on nytimes.com, asking which of the column’s claims its readership carried forward, which it let drop, and what discursive registers saturated the uptake.

Nicholas Kristof’s positioning, and the authority readers project onto the byline

The column did not arrive as anonymous testimony. It arrived under one of the most recognizable bylines in American opinion journalism, attached to a reporting tradition with specific epistemic features. Three observations about that positioning are relevant for what follows, and they are about positioning only — not about motive, sincerity, or character.

Kristof has been a New York Times opinion columnist since 2001 and a Times journalist since 1984. He is a two-time Pulitzer Prize winner — for international reporting on Tiananmen Square (1990, shared with Sheryl WuDunn) and for commentary on Darfur (2006). His reporting and column work over four decades has centered on humanitarian crises, mass atrocities, sex trafficking, and wartime sexual violence in contexts including Cambodia, Sudan, the Democratic Republic of the Congo, Myanmar, Tigray, and most recently Sudan and Gaza. He has worked across this terrain longer and more publicly than any sitting American op-ed columnist.

That reporting tradition has shaped how a piece under his byline is received. The convention his readers have come to expect is what could be called moral-witness journalism: the reporter travels to a site of atrocity, conducts in-person interviews with survivors whose accounts mainstream Western audiences have not heard, and constructs the column as a vehicle for testimony that institutional power has failed to register. The trust readers extend to such a column rests not on the column’s internal evidentiary apparatus — readers cannot evaluate Sde Teiman indictments or NGO methodology in real time — but on the byline’s accumulated record of having gotten this kind of reporting substantially right in prior cases. The authority is transferred from the prior cases to the present one, before the present case has been adjudicated on its own evidence.

This is the epistemic configuration the column operates within. A claim under this byline does not enter the discourse with the evidentiary weight intrinsic to its sourcing; it enters with the additional weight of the moral-witness tradition. That additional weight is what permits an opinion-section piece to function, for many readers, as documentation. It is also what makes inferential slides under this byline particularly consequential. A claim that would be hedged or qualified in a news-side report — published as opinion by Kristof — circulates as documented atrocity. The deconstruction that follows is, in part, an attempt to read the column as it would be read in a strict evidentiary register, with the moral-witness premium temporarily suspended, in order to see what the underlying source-base actually supports.

The article does not contest the moral-witness tradition as such; it is a legitimate genre with a long and serious record. What I contest is the assumption that a specific column’s specific inferential moves can ride on the genre’s accumulated authority without being independently examined.

The claims under examination

Before the deconstruction proper, it is worth fixing the target. The column advances three substantive claims about Israeli conduct, treated in the column with substantively equivalent confidence, plus one framing claim that organizes the column’s rhetoric:

• Claim 1. Individual acts of sexual violence occurred against Palestinian detainees and civilians, as reported in the column’s 14 first-person testimonies and supporting documentation.

• Claim 2. Such abuse is widespread, repeated across institutional settings (prison guards, Shin Bet interrogators, soldiers, settlers), and effectively unpunished. A pattern of tolerated misconduct under conditions of impunity.

• Claim 3. Such abuse constitutes “an organized state policy” and “standard operating procedure” of the Israeli state. A claim about design, not failure.

• Claim 4 (framing claim). The sexual violence committed by Hamas on October 7, 2023, and the sexual violence committed by Israeli actors against Palestinians are morally and structurally commensurate, such that “consistency requires” applying the same moral standard to both.

I accept Claim 1 as well supported and argue that Claim 2 is substantially supported by the column’s sources and by independent corroboration. The narrower question is whether Claim 3 and Claim 4 are warranted by the same sources. My argument is that they are not. Claim 3 represents an inferential step beyond what the underlying evidence supports, contradicted in places by the column’s own quoted sources. Claim 4 collapses substantive differences in scale, coordination, and operational character between two phenomena the column treats as parallel.

Deconstruction: testing the claims against the evidence

The sections that follow read the column as if the byline’s accumulated authority were temporarily suspended, in order to see what the underlying source base, on its own terms, supports and what it does not. They begin with what the column draws on and where its reporting is strong. They then identify where the inferential structure separates from the evidentiary structure. They close with the column’s diachronic context within the Times itself.

1. What Kristof’s column is built on

Kristof draws on: 14 first-person interviews with men, women, and children reporting sexual assault by Israeli detention staff or settlers; supplementary interviews with family members, lawyers, social workers, and investigators; a 2025 UN Independent International Commission of Inquiry on the OPT report describing sexual violence as part of Israel’s “standard operating procedures” with “implicit encouragement by the top civilian and military leadership”; an April 2026 Euro-Med Human Rights Monitor report describing “systematic sexual violence” as “an organized state policy”; a Save the Children survey (July 2023, “Stripped, beaten and blindfolded”) of detained Palestinian minors; a Committee to Protect Journalists survey of 59 released Palestinian journalists; documentation by B’Tselem, and a Norwegian Refugee Council–led West Bank Protection Consortium report; testimony collected by Breaking the Silence; and on-the-record statements from the Israeli Prison Service, Itamar Ben-Gvir, Benjamin Netanyahu, and Ehud Olmert.

2. Where the reporting is strong

The sourcing is triangulated: testimony, NGO reports, surveys, named indicted cases, and acknowledgment from inside Israeli legal and political circles.

The Sde Teiman case is a hard empirical anchor: a Palestinian prisoner was hospitalized in July 2024 with a torn rectum, cracked ribs, and a punctured lung; when military police detained reservists for questioning, a mob stormed the base; five reservists were indicted in February 2025; the charges were dropped in March 2026. The settler attack on the farmer Suhaib Abualkebash in the Jordan Valley is similarly well-documented. Matti Friedman, discussing the column on Dan Senor’s Call Me Back podcast, classified it as “as far as I know accurate, much to our shame.”

Bashi’s account that her organization has filed “hundreds of complaints” with “not a single case” leading to charges is significant. It is reinforced by the prison service spokesman’s response to Kristof, which declined to say whether any prison staff member had ever been fired or prosecuted for sexual assault. That declination, combined with Bashi’s numbers, is the column’s strongest single piece of structural evidence. Complaints exist in volume. Prosecution does not. This is claim 2 documented at the institutional level.

The piece is methodologically transparent about limits. Kristof writes that it is “impossible to know how common sexual assaults against Palestinians are” and acknowledges that not all cases could be corroborated. Rettig Gur, from inside the Israeli debate, lands in the same place. Many indictments. Case numbers in the dozens and probably the low hundreds across three years. “The problem is real. It’s there, it’s real, and it doesn’t seem to be stopping.” Friedman: putting a politician of Itamar Ben-Gvir’s stated views in charge of Israeli law enforcement creates structural conditions in which “many reasons for concern” become predictable. Two reform-oriented Israeli analysts arrive at this assessment without needing the symmetry frame Kristof imposes. The question for what follows is what Kristof’s column does beyond establishing the existence of the problem.

3. Where the reporting is weaker

Corroboration confirms disclosure, not occurrence. When confidants confirm that a victim told them about an assault, this confirms the disclosure happened, not that the underlying event occurred as described. This is standard trauma-reporting practice and not a defect in itself. But the piece treats corroborated and uncorroborated cases as substantively interchangeable in its cumulative argument. The unnamed-source pattern is heavy. Several journalists who reviewed the piece claim-by-claim described it as built around “an unnamed woman, unnamed soldiers, an unnamed location, dogs trained to rape humans, story after story with no date, no place, no name.”

Kristof’s appended editor’s note in the comments section, “in fact-checking we went through it all over again,” functions as a defensive appeal to verification rigor at precisely the points where the column’s verification is weakest. The note inoculates against the charge it cannot answer.

The Euro-Med Human Rights Monitor source is undercharacterized. Kristof describes it as “often critical of Israel.” Euro-Med is a highly contested source. Critics, including Israeli analysts and at least one Palestinian commentator (the writer Ahmed Fouad Alkhatib), have raised concerns about its sourcing methods, the institutional affiliations of its leadership, and its prior circulation of claims that did not survive verification (including the assertions that Israel uses weapons that "vaporize" Palestinians and that Israel harvests organs of Palestinian dead). These criticisms are not uniformly accepted. Euro-Med disputes them. I do not adjudicate the dispute here. What is worth registering is that a reader weighing the column's central systemic claim, "an organized state policy," should be in a position to know that the source from which that formulation derives is the subject of substantive and ongoing methodological dispute. The phrase "often critical of Israel" does not convey this.

The Norwegian Refugee Council/West Bank Protection Consortium report does not establish what Kristof uses it to establish. As Rettig Gur identifies, the report claims systemic sexual violence in the West Bank on the basis of approximately 16 incidents across three years in a region of roughly 3 million Palestinians and half a million Israelis, with some incidents barely meeting the threshold for harassment. That base does not support “systemic” or “one of the mechanisms driving people from their land.”

The Save the Children and CPJ figures float. “More than half” of detained minors witnessed or experienced sexual violence. But the piece gives no sample size, recruitment method, or operational definition. The CPJ figure (3% rape, 29% other sexual violence) comes from 59 journalists. The numbers reflect a serious problem. They do not, on their own, support a systemic-policy claim.

The trained-dogs claim has a contested source lineage. Both Rettig Gur and Friedman trace it to identifiable origins outside the mainstream press. Friedman describes the claim as sitting “within the world of the darkest conspiracies that you hear while riding in cabs in Middle Eastern countries.” Whether one accepts that characterization or not, the claim had been ignored by mainstream press outlets prior to its appearance in the Kristof column. One specific link in the column’s source chain is worth flagging. Kristof’s piece referenced an April 2026 viral tweet by Shaiel Ben-Ephraim, which had compiled secondhand testimony purporting to confirm the dog allegations. Ben-Ephraim left UCLA in 2020 amid accusations of sexual harassment, which he subsequently acknowledged as “inappropriate behavior.” The column does not flag this provenance for readers. I am not claiming the allegation is false. The claim made here is narrower: the allegation rests on a source chain that overlaps with known propaganda narratives, lacks independently verifiable corroboration, and is presented in the column without any indication to the reader that its evidentiary status differs from that of the named, indicted Sde Teiman case. The equivalence in presentation is the methodological problem.

A further word on the trained-dogs allegation is warranted. My restraint on the truth-value of the claim is principled, not endorsing. The methodological commitment is to examine how claims of varying evidentiary status are presented under common institutional authority and what readers do with them, not to verify the claims themselves. That said, the substantive plausibility of the specific allegation — that dogs were trained to sexually assault human prisoners — is highly improbable on its face, and the column offers no corroboration that would override that prior. The trope of trained-animal sexual assault has also appeared in earlier atrocity-propaganda contexts and merits editorial skepticism it did not receive here. I do not adjudicate the underlying factual question. What I do register is that the column’s editorial decision to present this allegation in the same prose register as the indicted Sde Teiman case is the methodological problem, regardless of whether the allegation is later substantiated. Readers seeking adjudication of the underlying claim should consult Friedman’s discussion on the Call Me Back podcast and any subsequent verification work undertaken by news-side reporters.

4. The central inferential gap: condoned versus organized

The evidence supports Claim 1 and substantially supports Claim 2. Claim 3 is different in kind. It asserts design, not failure.

The most consequential textual evidence here is internal to the column. The UN report Kristof cites characterizes the leadership’s relationship to the abuse as “implicit encouragement by the top civilian and military leadership.” That word implicit is doing significant work. It is the qualifier that distinguishes “leadership tolerates and does not stop” from “leadership orders and organizes.” Kristof prints the UN’s qualifier accurately. Then, in his own framing voice, he drops it. The UN’s implicit encouragement becomes Kristof’s standard operating procedure and organized state policy. The slide is visible inside the column’s own citation chain.

The column’s named Israeli human rights sources reinforce the gap. Bashi: “I don’t see evidence that it has been ordered.” Ben Marmarelli, the other Israeli lawyer Kristof cites, describes prevalence: rape with objects “is going on across the board.” That is a description of frequency, not of policy. Both Bashi and Marmarelli are claim-2 sources. Neither offers a claim-3 source. The column’s own self-undermining sentences sit alongside these: “There is no evidence that Israeli leaders order rapes,” and “It’s impossible to know how common sexual assaults against Palestinians are.” These are claim-2 hedges. They appear in the same column that calls sexual violence Israel’s “standard operating procedure.”

Rettig Gur’s assessment from inside the Israeli debate lands on claim 2. There is a breakdown of discipline. Ben-Gvir and Netanyahu are not interested in fixing it. The problem will fester if not stopped. Serious, named, internal critique. None of it warrants the phrase “organized state policy.”

5. The inferential mechanism: how the move from incident to policy is performed

The previous section named the destination: Kristof’s claim that the abuses constitute “an organized state policy” and “standard operating procedure,” a claim the column’s own sources do not warrant. The mechanism is what remains to be shown. By what specific argumentative moves does a competent columnist transform an evidentiary base that supports claim 2 (widespread, unpunished abuse) into a published statement of claim 3 (organized state policy)? The sections that follow examine particular instances. This section names the moves in sequence, so the pattern is visible.

• Move 1: Qualifier-dropping. A cited source contains an epistemic hedge — implicit, appears to be, allegedly, consistent with — that distinguishes one claim from a stronger neighbor. The columnist quotes the source accurately. Then, in his own framing voice, he paraphrases it without the hedge. The UN report’s “implicit encouragement by the top civilian and military leadership” becomes “standard operating procedure” and “organized state policy” in Kristof’s framing voice a few paragraphs later. The qualifier did precise work in the original. The distinction it preserved is between a leadership that knows misconduct is happening and does not stop it, and a leadership that has organized the misconduct as policy. Both are serious; they are not the same. Its removal performs the inferential slide. A reader encountering Kristof’s stronger formulation has no internal cue that the original wording was softer.

• Move 2: Equivalence-by-presentation. Claims of widely different evidentiary status appear in identical textual registers: same paragraph weight, same source-citation form, same authorial tone. The reader has no internal cue distinguishing a documented case from an uncorroborated allegation. The Sde Teiman case, in which five reservists have been indicted by Israeli military authorities, sits in the same prose pattern as the trained-dogs allegation, whose source chain Friedman and Rettig Gur trace to known propaganda channels. Both are rendered as elements of "the same pattern" of conduct. The equivalence in presentation encourages readers to process the claims as evidentially comparable.

• Move 3: Selective hearing of named sources. When a source quoted in the piece offers both a strong claim and a weak claim, the columnist's framing carries forward the strong claim and lets the weak one drop. Sari Bashi tells Kristof on the record: "I don't see evidence that it has been ordered." The sentence is printed. It does not appear in the column's framing. The same sentence that disqualifies the policy claim is published and then ignored. After publication, Ehud Olmert issues a statement clarifying that his quoted words do not support the column's stronger conclusions. The statement appears in a different outlet.

• Move 4: Symmetry as ontological claim. A rhetorical frame presents two phenomena as morally and structurally parallel: "Whatever our views of the Middle East conflict, we should be able to unite in condemning rape"; "The horrific abuse inflicted on Israeli women on Oct. 7 now happens to Palestinians day after day." The frame operates as if the comparison were definitional rather than empirical. The column's own scope acknowledgment contradicts the parallelism: Tigray and Sudan reflect sexual-violence scales "far greater than anything committed by either Hamas militants or Israeli guards or settlers." That acknowledgment does not displace the symmetry frame from its position in the column's framing.

• Move 5: Structural conditions as evidence of policy rather than as the alternative explanation. The column documents prolonged detention without external oversight, overcrowding, exposure of reservist guards to October 7 attack videos, and an ideological climate set by Ben-Gvir. Bashi inside the column, and Rettig Gur, Friedman, and the Israeli Public Defender's Office outside it, read these same conditions as the cause of widespread misconduct under impunity. That reading supports claim 2. Kristof presents the same conditions as confirmation of claim 3. The observations are identical. The conclusions licensed by them are different, depending on the framing. The stakes of the difference are not interpretive only. A reading that locates the cause of the misconduct in degraded oversight, undertrained reservist staff, and the absence of external accountability mechanisms identifies a set of structural conditions whose remedy is institutional reform. A reading that locates the cause in organized policy identifies the state itself as the object requiring characterization. The factual base supports the first reading. The column's framing licenses the second.

• Move 6: Institutional placement as authorization. The piece appears on the opinion page of the New York Times. Material that, by reported newsroom assessment, would not have passed news-side verification reaches readers under the paper's institutional authority. The byline does the authority-conferring work. A reader has no internal cue distinguishing opinion-page material from news-side reporting. The column's framing carries that authority into the public record, and, as the second main part documents empirically, into the discourse downstream.

Each of these six moves is individually defensible as journalistic technique. Any of them can be deployed in a column where the underlying argument is correct, and the result reads as competent reporting. The methodological problem named here concerns what happens when these techniques are deployed in combination on a high-salience charge against a state, with downstream readership numbering in the hundreds of thousands. The inferential architecture transforms a documented pattern of impunity into a published claim of designed policy. No single move is identifiable as the one that performed the transformation. The sections that follow — on the symmetry frame, the Olmert quote, the “American tax dollars” frame, the Abu Ghraib analogy, the NGO source pipeline, and the opinion-page mechanism — examine these moves at the level of individual instances in the column.

6. The symmetry frame

The symmetry frame is what the claims section identified as Claim 4. The piece opens and closes with structural equivalence between October 7 sexual violence and the abuses Kristof documents:

“It’s a simple proposition: Whatever our views of the Middle East conflict, we should be able to unite in condemning rape.”

“Think of it this way: The horrific abuse inflicted on Israeli women on Oct. 7 now happens to Palestinians day after day.”

The two phenomena being compared are structurally different in ways the column does not address. October 7 was a coordinated mass attack by an armed group, executed within several hours, in which sexual violence functioned as a tactic of war against civilians. The Civil Commission characterized it as “systemic, widespread, and deliberate,” documented across more than 400 testimonies and over 10,000 photographs. What Kristof documents is misconduct by individual state actors distributed across institutional settings, over years, under conditions of impunity. I am not claiming that comparison between the two is impermissible. Moral consistency is a legitimate frame, and the column’s appeal to it is not in itself the problem. The problem is that the comparison the column draws rhetorically compresses major differences in scale, coordination, evidentiary certainty, and operational character. Both are condemnable. They are not, on the evidence presented, parallel expressions of the same phenomenon.

Kristof’s own scope acknowledgment cuts against his closing. He writes that he has covered places “where the scale of sexual violence is far greater than anything committed by either Hamas militants or Israeli guards or settlers”: Tigray, Sudan. The man writing the “day after day after day” closing concedes, in passing, that both phenomena together are scaled-down relative to mass-rape events he has covered elsewhere. The symmetry frame is performed against that admission.

The publication timing reinforces the frame’s effect. Whatever the Times’s advance knowledge of the Civil Commission’s release schedule (now publicly disputed; see the introduction), the temporal fact is that Kristof’s column reached Times readers first. The Commission’s findings reached them the following day. Independent of intent at the byline level, the institutional consequence is that the Times’s first contribution to its readership’s understanding of the comparative scale of wartime sexual violence in this conflict was the symmetry frame. The underlying documentation arrived afterward, into a discourse the column had already shaped.

Dan Senor, on his Call Me Back podcast, offered one reading of what the symmetry frame produces in practice. The column allows readers to argue, in effect, that “Hamas is a sick, depraved society. And now we know, as Kristof albeit says, Israel is a sick, depraved society. Isn’t it a shame that both societies are so broken.” Kristof does not use that language. I am not imputing it to him. My claim is narrower: the structure of the column’s framing channels readers toward something close to Senor’s formulation, regardless of the framing’s intent at the level of authorial psychology.

7. The Olmert quote and the post-publication clarification

“Do I believe it happens?” he asked. “Definitely.”

“There are war crimes committed every day in the territories,” he added.

Olmert’s printed acknowledgment lends weight to claim 2. It does not lend weight to claim 3. The placement of the quote near the end of the piece, after the “organized state policy” framing has been introduced, encourages the inferential slide. The quote itself does not perform it.

After publication, Olmert addressed the column directly in a statement to the Free Press:

“Mr. Kristof’s article includes claims of extraordinary gravity: that Israeli authorities have directed the rape of children, that dogs have been used as instruments of sexual assault, that systematic sexual torture is state policy. I have no knowledge supporting these claims as I said to Mr. Kristof. Therefore, the positioning of my quote after pages of such allegations misrepresents my views.”

Olmert’s clarification separates two propositions the column rhetorically fuses. He confirms that abuses occur (“Do I believe it happens? Definitely”). He explicitly disclaims having authority for the three escalated claims he names: directed rape of children, dogs as instruments of sexual assault, and systematic sexual torture as state policy. He also names the rhetorical mechanism by which the misrepresentation operates. It is the positioning of his quote after pages of such allegations.

The column’s most authoritative on-the-record Israeli source has, in his own words, separated claim 2 from claim 3. He has stated that the column’s structure performs the inferential slide in his name without his authority. A column that drew an “organized state policy” conclusion from sources who explicitly say only that abuse occurs and is not being stopped did not, even at the level of its own quoted material, ever have the authority for the conclusion its framing implies.

8. The “American tax dollars” frame

“Yet our American tax dollars subsidize the Israeli security establishment, so this is sexual violence in which the United States is complicit.”

This restructures the column from “documentation of abuses by a foreign government” to “documentation of abuses in which the reader is implicated.” The reader’s response is no longer assent or disagreement. It is complicity or repudiation. The column closes by addressing this implication directly to the Trump administration and naming Marco Rubio, now Secretary of State, in the opening as one of those who condemned October 7 sexual violence. The column positions itself as a moral challenge to figures with current policy authority.

The argument from US aid → US complicity → US obligation is defensible foreign-policy commentary. It is also an argument, not a premise. The piece treats it as a premise. The rhetorical effect is to remove the option of analytical distance.

9. The Abu Ghraib analogy and the breakdown-of-discipline reading

Kristof’s interpretive framework is explicit:

“Decades of covering conflict have taught me that a combination of dehumanization and impunity can propel people into a Hobbesian state of nature.”

He invokes Abu Ghraib and the Abner Louima case as comparative anchors. His exact phrasing, that dehumanization-plus-impunity “roughly explains how American soldiers came to sexually abuse prisoners at Abu Ghraib,” is hedged. He does not claim Abu Ghraib was US policy. The same hedge, applied consistently, would foreclose the “organized state policy” framing for Israel.

The structural conditions Kristof himself documents support the breakdown-of-discipline reading. By his own count, Israel has detained 20,000 people in the West Bank since the October 7 attacks. More than 9,000 are still being held. Most have been denied visits from the Red Cross and lawyers since 2023. Closed institutions without external oversight, overwhelmed in capacity, staffed in part by under-trained reservists drafted after October 7 and exposed to Hamas’s own attack videos: these are the conditions Rettig Gur identifies as the proximate cause. They produce predictable patterns of misconduct under impunity. They do not require organized policy. Kristof benefits from the moral force of the stronger framing while his interpretive scaffolding sits with the weaker one.

10. The NGO source pipeline

A piece like this is not assembled from scratch. Matti Friedman, who covered the Israeli–Palestinian conflict for the Associated Press’s Jerusalem bureau (2006–2011) and whose 2014 analyses in The Atlantic and Tablet mapped the structural mechanics of Western press coverage of this conflict, identifies the assembly pattern:

“Nicholas Kristof is not based here. He doesn’t have a handle on events here. What’s happening is that he’s being handed a package by NGOs.”

A note on positionality. Friedman and Rettig Gur, who are central to the analysis here, are not neutral observers. Both write from within a broadly Zionist frame and have their own substantive disagreements with Western press coverage of Israel. I draw on them not as final authorities but because their factual descriptions of newsroom workflows and Israeli institutional processes are independently checkable and converge with evidence from other sources, including Israeli reform-oriented institutions and at least one Palestinian commentator (Alkhatib) who criticizes the same column from a different starting point. Friedman’s more polemical formulations — for instance his characterization of the Kristof column as “essentially an NGO hit job” — are his analytic verdict, not mine. What I take from Friedman is the structural description. I do not adopt his rhetorical conclusion.

The structural pattern Friedman describes is well-documented in press-criticism literature and is not specific to coverage of Israel. As Western press resources have shrunk and production demands have grown, well-funded advocacy NGOs have moved into the gap, producing reports, source packages, and interviewee introductions. NGO documentation is often indispensable in access-restricted environments where journalists cannot independently reach sources. Israeli detention facilities are one such environment. The issue raised by Friedman is not the use of NGO sourcing as such. It is reliance without differentiation, and the migration of advocacy framing into journalistic framing without visible differentiation at the level of presentation.

Kristof tells the reader directly how the reporting was seeded. He writes that he “became interested in reporting on sexual assaults against Palestinian prisoners after Issa Amro” (a public activist) “told me when I previously visited that he had been sexually assaulted by Israeli soldiers and that he believed this was common but underreported because of shame.” Amro’s framing, common but underreported because of shame, became the column’s working hypothesis before the reporting began. The reporting then confirmed the hypothesis. This is what the structural diagnosis names: framing precedes evidence collection, and evidence is gathered in ways consistent with the framing.

This pattern is consistent with how a column can simultaneously include real cases (Sde Teiman, Abualkebash) and contested or unverifiable claims (trained dogs, the “organized state policy” formulation) without an editorial check distinguishing between them. In the NGO-handoff workflow, the reporter is not the primary source-verifier. The NGO is. When the NGO is a contested source, the verification layer is weakened accordingly.

Friedman extends the analysis to Israeli human rights organizations: B’Tselem, Breaking the Silence. He characterizes them as having become “effectively franchises of the international left or the international anti-Israel campaign,” funded “almost completely from abroad” and addressing audiences “in English to audiences abroad.” That characterization is contested. B’Tselem and Breaking the Silence reject it, and their domestic Israeli readership is not negligible. What is harder to contest is that their specific case documentation is in many instances rigorous (Israeli reform-oriented debate draws on it), and that the institutional question of whether they function as a trusted internal mechanism for identifying Israeli moral failures, or as an input to a foreign-audience discourse aimed at delegitimization, is the subject of substantive disagreement inside Israel and outside it. What is worth noting is the disagreement itself, and the tangle of documentation-and-campaign at the level of the source apparatus. The Kristof column reflects this tangle and does not name it.

11. The opinion-page mechanism, and the broader pattern of unfact-checked claims

One defense of the column has been that it appeared on the opinion page, where standards differ from those of the news section. Friedman’s response is twofold. First, the distinction is largely meaningless in current practice: “much of what is presented as news coverage is kind of passive aggressive opinion.” Second, the distinction was nonetheless operative in this case in the opposite direction. Several Times newsroom reporters reportedly observed that the column’s central claims, particularly the trained-dogs allegation, “would have never made it through the news process. It only slipped through the opinion process.” That detail is analytically important. The opinion-page placement was not a label disclaiming reporting. It was the route by which material that could not pass news-side verification reached Times readers under the paper’s institutional authority.

Rettig Gur documents the broader pattern. A striking claim about Israeli atrocity is advanced by an NGO, a UN office, or an open letter. It spreads through advocacy networks and into headlines without being checked. When it later proves unsupportable, it fades quietly. The early-2024 mass-starvation predictions for Gaza, repeated throughout 2024 and 2025 in escalating formulations, never materialized at the predicted scale. A Lancet letter projected hundreds of thousands of deaths, not seriously read by most who repeated it. A UN rapporteur claimed 380,000 infant deaths under five, exceeding the actual population of Gazan children under five. Friedman’s stronger characterization of the broader press environment is that the operative question in many newsrooms is “not, is this information accurate? The question is, who does this help?” That is Friedman’s own diagnosis, made in the context of a sympathetic discussion of reform-oriented Israeli critique. I record it as part of the discursive context the column was edited within, not as an institutional finding about specific Times editors.

The “trained dogs” claim and the Euro-Med “organized state policy” formulation enter the column from this ecosystem. The column’s editorial choice to treat them as equivalent in evidentiary weight to the indicted Sde Teiman case reflects the pattern Rettig Gur and Friedman identify.

12. What this adds up to

The piece is well-reported in the conventional sense. It gathers testimony, triangulates sources, anchors claims in a documented case, names its limits, and includes on-the-record acknowledgment from inside the institution it criticizes.

It is analytically uneven in five respects:

1. It allows contested sources (Euro-Med, a Norwegian Refugee Council report whose case base does not support its headline) to carry argumentative weight that its own strongest internal source explicitly declines to carry.

2. It builds an “organized state policy” claim atop on-the-record sources who, when given the chance, deny that their words support the claim. Bashi inside the piece. Olmert in his post-publication clarification.

3. It drops qualifiers from its own cited material. The UN report’s “implicit encouragement by the top civilian and military leadership” becomes Kristof’s “standard operating procedure” and “organized state policy.”

4. It frames its findings through a symmetry with October 7 that is rhetorically powerful, analytically unargued, and contradicted by the column’s own scope acknowledgment. The publication timing relative to the Civil Commission report amplifies the effect.

5. It enters the public record through an editorial pipeline (NGO source handoff, opinion-page placement on material that did not meet news-side standards) that consistently favors advocacy framings over verification.

None of this means the abuses Kristof documents did not occur. The case for claim 2, widespread, repeated, unpunished, is strong on the evidence assembled. The case for claim 3, that this is policy, not breakdown, is not made in the piece. The column’s own sources push back against the policy claim in print and after publication.

Senor’s methodological point is worth registering. The instinct to concede the existence of real problems before contesting the column’s framing (”yes, there are problems in the Israeli prison system”) appears to be the principled response. But this concession, made on the column’s terms, accepts the frame the column establishes. The structure of the column channels the discussion not toward whether problems exist in Israeli detention, but toward whether Hamas’s coordinated sexual violence on October 7 and Israeli prison conduct are commensurable expressions of a shared moral failure. Conceding the first as a way of contesting the second draws the conceder into the second discussion on the column’s structural terms. This is an observation about the framing’s effect on subsequent discourse, not a claim about authorial intent.

The reform-oriented Israeli analyst’s framework (Rettig Gur, Friedman, Bashi inside the column) offers a way out of the false binary the Kristof column sets up. Real abuse exists and must be addressed with the seriousness Israeli society has so far failed to bring to it. This is fully compatible with declining to carry forward, into the public record, claims of systematic dog-rape and the unverifiable “organized state policy” formulation on the same evidentiary footing as the documented misconduct. Acknowledging the first does not require accepting the second. The Kristof piece elides the distinction. A reader who wants to take Palestinian suffering seriously needs to hold the distinction more carefully than the column does.

The convergence of perspectives on this analytical point is worth noting. The same critique has been articulated, in the days following publication, by analysts approaching the column from very different starting points. The Israeli policy analyst Eli Kowaz argued that Kristof had “foregrounded the most sensational allegations” while neglecting better-documented material (Ben-Gvir’s rhetoric, a recent Israeli Public Defender’s Office report documenting systematic violence in Israeli prisons), with the predictable result that “by Thursday, the conversation will be about Euro-Med’s credibility and whether unverified accounts can be trusted. The documented case, the one that required no advocacy org, no anonymous source, no unverifiable claim, will be largely beside the point.” The Palestinian writer Ahmed Fouad Alkhatib, a public critic of Hamas, made the same evidentiary point from the other side. While “incidents of sexual abuse have occurred in Israeli prisons” without doubt, Kristof’s reliance on Euro-Med and similar sources, which have “troubling records on accuracy, conduct, and associations,” undermines rather than supports the case for serious investigation. A column intended to draw attention to documented abuse instead produces, by its method of assembly, a discourse in which the documented case is overshadowed by the unverifiable one. That outcome is not what reform-oriented critique of Israeli detention conditions requires. It is the opposite.

It is worth stating plainly, at the close of this analysis, what the column does and does not put into the historical record under the Times byline. Some of what the column reports is supported by named cases and indictments. The most sensational specific allegations, including the trained-dogs claim, rest on source chains that overlap with known propaganda narratives and lack independently verifiable corroboration. Readers of the column cannot tell, from the column itself, that this is so. Olmert, the column’s most authoritative on-the-record source, has explicitly disclaimed the policy framing the column built around his words.

The column gathers a range of allegations of widely varying evidentiary quality, treats them as broadly comparable, and presents them as expressions of organized institutional depravity. The structure shares features with earlier media episodes in which highly emotive accusations circulated faster than verification standards could stabilize them. What the institutional setting of the New York Times affects is the column’s reach, not the mechanism of its assembly.

13. The diachronic pattern: this column in the context of two and a half years of NYT Israel coverage

A note on the claim being made in this section. The Kristof column did not arrive in a vacuum. There are at least two prior instances in roughly thirty-one months in which the New York Times prominently published serious accusations against Israel that subsequently required institutional correction. I am not claiming that three episodes establish a stable pattern, prove institutional bias, or demonstrate editorial intent. Three cases do not, on their own, settle questions of that magnitude. What three cases can do is illustrate how, under contemporary editorial conditions and access conditions, high-salience accusations can outpace verification, and how the institutional reach of an initial framing systematically exceeds the reach of any subsequent correction. The Times during this same period has also published reporting damaging to Hamas — the news-side coverage of the Civil Commission’s October 7 report on May 12, 2026, is one example — and I do not claim otherwise. The argument here is narrower: these three episodes raise structural questions about editorial process and correction-asymmetry that the institution itself has not yet addressed. What follows does not attempt a systematic content analysis of New York Times coverage overall; the three-case comparison is illustrative, not statistically representative. The cases selected here are not presented as exhaustive, but as illustrative instances in which subsequent qualification materially altered the evidentiary status of the original framing.

October 17, 2023 — Al-Ahli Hospital. Hours after an explosion at the Al-Ahli Baptist Hospital in Gaza City, the Times published the top-of-homepage headline "Israeli Strike Kills Hundreds in Hospital, Palestinians Say." The number — 500 dead — came from the Hamas-run Gaza Health Ministry and was not independently verified. Subsequent investigation by Israeli, US, French, British, and Canadian intelligence services concluded the explosion was caused by a failed Palestinian Islamic Jihad rocket. US intelligence put the death toll at 100–300; a European official at 50 or fewer. On October 23, the Times published an editor's note acknowledging that "the early versions of the coverage — and the prominence it received in a headline, news alert and social media channels — relied too heavily on claims by Hamas, and did not make clear that those claims could not immediately be verified. The report left readers with an incorrect impression about what was known and how credible the account was." By the time the correction appeared, the original framing had circulated globally, contributed to violent protest, and shaped the early international discourse of the war.

July 25–29, 2025 — The Gaza-starvation front page. The Times ran a front-page image of an eighteen-month-old Palestinian child, Mohammed Zakaria al-Mutawaq, held emaciated in his mother's arms wearing a diaper made from a black garbage bag, under the headline "Gazans Are Dying of Starvation." The caption stated that the child had been "born healthy." Within forty-eight hours, independent investigation revealed that the child had been diagnosed with cerebral palsy and a serious genetic disorder; the photographer's wider image set showed the mother and the child's older brother both appearing well-nourished. The Israeli Consul General in New York, Ofir Akunis, who had requested the correction, said in a statement: “It’s unfortunate that the international media repeatedly falls for Hamas propaganda. First they publish, then they verify, if at all.” The Times issued an Editors’ Note on July 30 acknowledging that the original article had “lacked information” about al-Mutawaq, and that after publication the paper had learned from his doctor that the child had pre-existing health problems that, had they been known, would have been included in the original story. The note was posted on a Times public-relations X account with fewer than 90,000 followers, not on the main Times account, which has over 55 million. The original image, by then, had been reproduced by CNN, BBC, the Daily Mail, The Guardian, The Times of London, and the Daily Express, and had become the global visual emblem of the Gaza famine narrative.

May 11, 2026 — The Kristof column. Trained dogs raping Palestinian prisoners. Israeli sexual violence as "organized state policy." Olmert quoted as confirmation; Olmert subsequently disclaiming the confirmation. The same shape: a high-salience claim, weak source verification on parts of the claim, prominent placement, and a clarification that arrives, when it arrives, after the original framing's circulation.

Two observations are consistent with these three cases.

• First, each case involves at least one claim that arrived through a source whose institutional positioning would, on conventional verification standards, have warranted significant editorial skepticism. The Hamas-run Gaza Health Ministry. A freelance photograph distributed via a Turkish state agency partnered with Getty. Advocacy NGOs including Euro-Med and the Norwegian Refugee Council. In each case, the Times led with the source’s framing and corrected or qualified afterward.

• Second, the corrections did not catch up to the original framings. The Al-Ahli editor’s note appeared six days after the homepage headline. The Gaza front-page correction appeared on an auxiliary X account three days after the original image ran. The Kristof column has not been corrected; Olmert’s clarification was issued not by the Times but by the Free Press. In each case, the asymmetry between the reach of the original claim and the reach of the correction is structural rather than accidental. The original framing reaches readers at scale. The correction documents that the framing did so.

A note on the days following publication. The institutional reaction unfolded along familiar lines and is worth recording as part of the case file. On May 14, Prime Minister Netanyahu announced that Israel was preparing to file a defamation lawsuit against the Times and Kristof. He had previously threatened the step over the Times’s 2025 starvation coverage but had not taken it. The American Jewish Committee characterized the dog-rape allegation as a “modern-day blood libel.” Michal Cotler-Wunsh, in the JNS interview already cited in the introduction, characterized the column itself as “blood libelous” and argued that “blood libel, once released into the air, has this way of spreading and becoming accepted ‘truth,’ regardless of the failure to provide evidence.” She warned that what such framings normalize is “the ever-mutating lethal hate that’s literally targeting, enabling, fueling the vandalism, violence and even the murder of Jews.” Approximately 200 demonstrators protested at the Times’s Manhattan headquarters on May 14, with signs reading “Antizionism gets Jews killed” and “J’accuse” (Dreyfus reference). One protest organizer drew a rhetorical comparison between the column’s discursive climate and Der Stürmer in remarks to assembled press. The use of that analogy in this context is itself worth noting as a marker of how the column landed in segments of the affected community. The Times defended the piece in spokesperson statements characterizing Netanyahu’s lawsuit threat as part of “a well-worn political playbook that aims to undermine independent reporting” (Danielle Rhoades Ha). That defense takes the form of changing the subject from the column’s substantive sourcing problems to the political motivations of one critic. Olmert’s Free Press clarification, published in the same window, did not change the subject.

I take no position on the lawsuit, which raises separate press-freedom considerations that exceed the scope of this piece. The reaction is recorded here as documentation of the discourse environment within which the column was received.

Reactions from Times readers: an analysis of the comment-section uptake

The first main part examined what the Kristof column does on the page. This second part examines what its readers did with it.

The relationship between Part 1 and Part 2 is not merely sequential. Part 1 named six argumentative moves by which the column transforms documented misconduct into a published claim of organized state policy. Part 2 examines what happens in the discourse downstream when those moves are deployed in combination under mainstream-press institutional authority. The analytical question is whether and how the framing operations identified in Part 1 are registered in the readership’s response — and, if so, whether they are registered at the extreme tail of the thread, where one would expect them by the standard reading of moderated press comment-spaces, or in the modal layer, which would represent a different kind of finding about how high-salience framings move through moderated discourse.

Dataset

The corpus comprises 731 reader comments extracted from the publicly visible comment thread published directly below the column on nytimes.com, captured before the comments section was closed. After deduplication by content hash (35 duplicates removed, paste-capture artifacts from the extraction process), 696 unique comments form the analytical corpus. All counts that follow are exact counts on this set. The page footer for the closed thread reads 809 comments; the difference between 809 (footer) and the 731 captured at the time of extraction reflects a combination of (a) comments held back from the public sort views into which the capture was scoped — the page defaults to “Newest” and “Reader Picks,” which together do not display every comment in the thread — and (b) post-publication moderation actions taken on a subset of comments between the time of the original capture and the time the thread was closed. The analytical corpus is therefore the publicly retrievable subset under capture-time conditions, not the full moderated thread.

Research Design

The coding framework applied here draws on methodology developed in Decoding Antisemitism (2020–2024) — which produced the 40-concept Decoding Antisemitism Lexicon (Becker et al., 2024, Palgrave Macmillan/Springer Nature) and the operational distinction between explicit, coded, and context-dependent layers of antisemitic discourse — and extended in Decoding Hate, its successor project at NYU’s Center for the Study of Antisemitism. The present case study applies the antisemitism-track coding apparatus to a single empirical event, coded manually by the author.

Each unique comment was coded along two dimensions.

• (i) Position re Kristof’s three claims, using the same 1/2/3 ladder the first main part introduced (claim 1: individual acts occurred; claim 2: widespread, unpunished abuse; claim 3: “organized state policy”). Six position categories, defined and operationalized in §14 below.

• (ii) Antisemitism-related concept-tagging, applying the DA Lexicon framework. Coding followed the Lexicon’s published definitions, with each comment coded independently for presence or absence of each of the 40 concepts via pattern-matching against the corpus followed by sample validation. Concept categories were not mutually exclusive; a single comment could receive multiple concept tags when distinct rhetorical operations co-occurred in the same comment. The unit of counting in concept-frequency tables is the concept instance (a single comment carrying three concept-instances is counted three times), because the analysis tracks the distributional weight of each concept in the discourse, not the number of commenters. As with all qualitative discourse coding, category assignment involves interpretive judgment shaped by contextual reading and the operational definitions of the DA Lexicon; the counts reported below should be read as one disciplined application of those definitions rather than as the output of a fully mechanical procedure.

A note on what this corpus is and is not. The New York Times moderates its comment threads. Explicit hate registers are largely filtered out before posting; what passes moderation is a coded layer that carries the relevant concepts through historically-loaded analogy, structural framing, and group-attributed agency rather than through surface markers. The frequencies reported below are accordingly the moderation-filtered floor. What the data show is the patterns the moderation process does not remove — and that finding is informative on its own terms.

13. Pro Kristof, against Kristof

Before turning to the concept-level analysis, the simplest question: how many comments support the column, and how many push back?

The unique comments distribute as follows:

Pro + beyond combined: 92.5% — i.e., the column’s claims, including the strongest of them, are sustained or extended in roughly nine of every ten substantive comments. Against: 2.0%. The thread is not balanced. The Kristof column landed in a readership that overwhelmingly accepted its core argument. The substantive disagreement visible in the thread is not whether the abuse Kristof documents is real, but how far the conclusions to be drawn from it should extend.

A pure pro/against split — collapsing the beyond-Kristof category into the pro side — gives:

• For the column’s claims: 644 comments (92.5%)

• Against the column’s claims: 14 comments (2.0%)

• Non-evaluative: 38 comments (5.5%)

The asymmetry is large enough that it should be named directly. The mainstream-press opinion-section comment thread on a column making the claims this one makes is not, in this case, a contested space. It is a space of broad confirmation. The analytical disagreements that exist visibly inside the column itself — Sari Bashi’s on-the-record disclaimer of the policy claim, Olmert later disclaiming the policy framing the column built around his words — do not enter the comment-section uptake. Bashi’s distinction is invisible at this layer. The “organized state policy” formulation is what readers carry forward.

Position categories — the breakdown

The six-category breakdown that produces the headline numbers above:

Category A — Full confirmation (n = 570). The modal comment in this thread. Accepts the column in its entirety, including claim 3, and typically calls for policy responses Kristof endorses: cut military aid, condition arms transfers, end administrative detention, prosecute the perpetrators, vote out politicians who support unconditional aid to Israel.

Category B — Confirmation with reservations (n = 9). Accepts that the abuses occurred but flags evidentiary, sourcing, or framing concerns. Linguistically marked by hedging operators: “if true,” “we need evidence and investigations,” “should be independently evaluated,” “demand unfettered access,” “extraordinary rigor required.” Clear exemplars include: “Not saying this didn’t happen — but we need evidence and investigations and proceedings”; “If true, the reported systematic sexual torture of Palestinians … is so far beyond the pale of acceptable activity by an ally, that, at minimum, we should demand of Israel unfettered access, a thorough, independent, international investigation”; “Can medical records for the surgical interventions that Kristof refers to be acquired and independently evaluated…?” The position is the methodological-restraint position, the one closest to the deconstruction’s analytical line. It has 9 comments. 1.3%.

Category C — Escalation beyond column (n = 65). Goes further than the column itself: characterizes Israel as a “vile country,” “rogue state,” “terrorist state,” “abomination,” genocidal, settler-colonial, apartheid, or in six cases as having no right to exist. The column established claim 3 (”organized state policy”) and stopped. Category C readers carry the argument forward into discursive territory the column does not enter. Operationalized rule: any comment carrying at least one of eight escalation-coded concepts (genocide, apartheid, settler-colonial, denial of right to exist, rogue/pariah state, Jewish supremacy framing, anti-Zionist state-rejection register, BDS-adjacent calls) is automatically Category C.

Category D — Rejection on factual grounds (n = 10). Denies, challenges, or characterizes as fabricated specific factual claims in the column. The most analytically forceful instance targets the trained-dogs allegation: “When I read about the supposed training of dogs to rape men… this is completely impossible. Dogs cannot be forced or taught to rape people. … But to write that systemic rape is part of the Israeli prison system for jailed Palestinians is a modern day form of blood libel.” Other instances in this category include: “anecdotal claims and unverified accounts rather than independently confirmed facts”; “How could this be true? This is not the Israel I know”; “Inclined not place much faith in the words of the those that have sworn the destruction of Israel.”

Category E — Rejection on framing grounds (n = 4). Accepts the column’s substance but rejects its framing as one-sided or lacking context. One comment challenges the symmetry frame directly: “Where were you then, Mr. Kristof?” (in reference to the October 7 coverage). Another adds: “the article lacks important historical context. Sexual abuse of prisoners is tragically common across wartime detention systems and prisons worldwide.” A third asks simply: “My only question is why is this published in the Opinion section.”

Category X — Non-evaluative (n = 38). Short comments of thanks, grief, expressions of difficulty in reading, journalist appreciation, requests for organizations to donate to. No substantive evaluation of the column’s claims.

The methodological-restraint cluster (B) and the rejection cluster (D + E) together account for 23 comments — 3.3%. The remaining 96.7% of the corpus either fully accepts the column, extends its argument beyond what it actually says, or does not engage its substance.

Concept-tagging results

Of the 40 concepts, 14 had non-trivial presence in this corpus. The remaining 26 were checked for and not observed at the discursive surface — consistent with the comment thread being a moderated mainstream-press space.

Concepts present, in descending order of frequency:

584 comments carry no concept. 91 carry one. 19 carry two. 2 carry three.

The top five concepts merit individual treatment.

1. Genocide claim (30 instances)

The most frequent concept in the corpus. The register is consistent: Israeli action against Palestinians characterized as genocide, genocidal, or as ethnic cleansing. Exemplars from the corpus: “It’s time for my fellow Jews to stand up to this genocide. … What Israel has done and is doing is a genocide. As Americans, we are funding this genocide”; “The sickness of genocide has dehumanized many Israelis”; “Modern Israel has never truly valued human life, the Nakba was planned from the 1890s … now through the Gazicide of the 2020s”; “I don’t know how humanity can move on with a reckoning for all involved in the genocide of the Palestinians”; “Bombing schools and hospitals along with sexual assaults is a Putin tactic he unashamedly admits as he says, ‘to break the will of the people.’” All 30 instances fall in Category C; the genocide framing is, in this thread, the diagnostic marker of escalation beyond the column.

2. Conspiracy Theories (23 instances)

A note on the operational threshold for this category. The DA Lexicon’s formal label for this concept is “Power/Conspiracy Narratives” — covering narratives of small-group power, lobby capture, and media-control attributions; the article uses “Conspiracy Theories” as the more widely-recognized term for the same operational category. Not all criticism of AIPAC, of campaign finance, or of organized political lobbying counts as Conspiracy Theories in the DA Lexicon sense. Generic criticism of lobby influence on US policy can be made, and routinely is made, in entirely non-antisemitic registers. The threshold for coding a comment under this concept is that the comment ascribes to a Jewish, Israeli, or Israel-aligned actor a degree of disproportionate, ethnicized, or covertly coordinated agency over institutions, governments, or media — the structural marker that distinguishes conspiracist antisemitism from ordinary institutional critique. The 23 instances counted here meet that threshold; comments criticizing AIPAC, lobbying, or campaign finance in registers that do not meet that threshold were not coded.

Within that filtered set, the structure of the 23 instances is consistent: a small but powerful Jewish or Israel-aligned lobby (AIPAC, “Zionist lobby,” “wealthy Zionists,” “the lobbies that have bought their vote,” “American Jewish organizational leadership”) is described as the mechanism by which US foreign policy is captured against the will of the American public. “There is silence from the US government because pro-Israel lobbies in the US have basically purchased that silence and support for Israel’s policies … with political donations.” “Wealthy Zionists, both Israeli and American, have had the political, economic, financial, and media power to condition Western governments and many of their citizens to believe that the Zionist way is crucially important.” “It is the no-criticism allowed supporters of Israel — Jewish AND gentile — who need to be held to account.” “[Ritchie] Torres who is more a spokesperson of Bibi than a US Representative.” 19 of the 23 instances appear in Category A — that is, in confirmation-of-Kristof comments, not in escalation comments. This is worth noting: the Conspiracy Theories register operates at the ordinary confirmation layer of this corpus, not as a marker of escalated discourse.

3. Holocaust Inversion (18 instances)

The trope: the historically persecuted Jewish people have, in the form of the Israeli state, become the persecutors, and now resemble or have become “the very thing they most despise.” “Israel is becoming the type of government that their ancestors left Europe to avoid.” “Israel, under Bibi, is the abused child who has grown up to become the abuser.” “Just as an abused child will often grow up to mistreat and torment children, so Israel, a country born out of abuse and cruelty, has emerged to be cruel and abusive itself.” “Their ‘Holocaust dividend’ has been depleted” — a particularly stark formulation in which Jewish historical persecution is treated as a moral credit-line Israel is described as having drawn down. 16 of the 18 instances appear in Category A — confirmation of the column without state-rejection escalation.

4. Nazi Analogy (14 instances)

The Israeli state, Israeli soldiers, or Israeli society compared explicitly to Nazi Germany. “The State of Israel dehumanizes Palestinians. The very same thing the State of Germany did to Jews in 1930s and 40s.” “Goering denied the gassing of Jews and Eichman wrote it off as accounting ledger entries.” “Hannah Arendt … concluded that evil does not swoop down in some vast, horrific, identifiable form. … Israel now embodies what she saw in Eichmann’s rationalizations.” “There were reports leaking out of Europe in the 1940’s of atrocities … All denied by the Nazi regime. Until it couldn’t be denied.” The DA Lexicon distinguishes Nazi Analogy from Fascism Analogies; in this corpus the former outnumbers the latter by 14 to 1, meaning that when commenters reach for an analogy to characterize Israeli state behavior, they reach for the specific historical case of the Holocaust-perpetrating regime, not for a generic political category. 13 of the 14 instances appear in Category A — operating, like Holocaust Inversion, at the confirmation layer of the thread rather than the escalation tail.

5. Israel as ‘rogue state’/pariah (14 instances)

“Israel is a rogue state.” “Time the US and the rest of the western world stopped supporting and funding this vile criminal regime.” “It is the country that is the problem. Its people widely participate in the horrors it inflicts. … Israel is an abomination.” All 14 instances are in Category C — the rogue-state framing is, in this corpus, definitionally escalation beyond the column.

6. Concepts at lower frequency

Colonialism Analogies (7 instances) characterize Israel as a settler-colonial project: “the Nakba was planned from the 1890s, initially implemented in the late 1940s, and has progressively continued on, now through the Gazicide of the 2020s”; “Israel that conquered and dispossessed the indigenous population of Palestine”; “it’s a pure shame that western countries keep protecting this colonial project”; “reminiscent of the attitudes of white populations toward the native peoples of the Western Hemisphere.” All 7 in Category C.

Denial of Israel’s right to exist (6 instances) is distinguishable from criticism of any particular Israeli government or policy. Six commenters cross the line explicitly. These six span an analytical range that is worth marking. At one end are bare anti-Zionist legitimacy-rejection formulations that reject Israel’s right to exist on principled grounds (whether religious-myth-based, settler-colonial, or generic-state-skeptical); at the other end is the explicit call for the removal of Israeli Jews from the territory. Both ends sit in Category C of the position coding, but they describe different end-states. I note the range; the count (6) holds. “Present-day Jewish Israelis have no intrinsic right to a homeland at the expense of Palestinians… And no amount of torture and dehumanization will give them that. The U.S. should immediately cease any and all support for Netanyahu’s and Ben-Gvir’s government.” “No country has a ‘right’ to exist and never has. Why do Americans think Israel has a right that nobody else has or ever had other than due to Israeli propaganda that nobody calls them out on.” “The existence of Israel is based on a biblical claim — myth — not reality, and does not have some inherent right to exist.” “How much longer can we shy from the conclusion that Israel must be defeated, and Israeli Jews removed and dispersed? We’re plainly bombing the wrong country.” “Moderate, admirable, secular Israels [sic] should emigrate and the US should take them in as refugees.” “Apartheid Israel is beyond the pale and should be dismantled.” The “removed and dispersed” formulation — the explicit call for Israeli Jews to be removed from the territory — is the most explicit eliminationist statement in the corpus; it appears, was published by NYT moderation, and drew no reply.

Apartheid Analogy (5), Anti-Zionism state-rejection register (4), Jewish supremacy/religious-essentialist framing (4), Boycott/BDS-adjacent calls (3) round out the Category C–clustering set.

Whataboutism/Hamas-deflection (5 instances) is the only concept-cluster appearing on the rejection side. “Where were you then, Mr. Kristof?” “Hamas did not allow the Red Cross to visit the hostages. How come no one seemed to care much about that?” “Hamas terrorists raping Gazan widows … Where’s Kristof on that one?” The whataboutism comments distribute across positions: 3 in A, 1 in C, 1 in E.

Blood libel (1 instance) appears once in the corpus — and only as a critique of the column’s dog-rape claim: “to write that systemic rape is part of the Israeli prison system for jailed Palestinians is a modern day form of blood libel.” The blood libel register is typically a marker of the discourse being analyzed; in this corpus, the only invocation of the term is as a critique of one of Kristof’s specific allegations.

Cross-tabulation: where do the concepts cluster on the position-coding?

This is the analytically load-bearing finding of the second main part. The distributional pattern of the fourteen observed concepts across the six position categories (defined in §14) shows two distinct clusters.

Cluster 1 — state-rejection concepts that cluster entirely in Category C.

The distributional pattern shows two distinct clusters.

• Cluster 1 — state-rejection concepts that cluster entirely in Category C. Genocide claim, Israel-as-rogue-state, Colonialism Analogies, Denial of Israel’s right to exist, Apartheid Analogy, Anti-Zionism, Jewish supremacy framing, Boycott/BDS: all 100% Category C. These concepts function by definition as escalation markers. Their presence in a comment denotes that the commenter has moved past the column’s claim 3 into a register of state-delegitimization or replacement.

• Cluster 2 — inversion/comparison concepts that cluster heavily in Category A. Nazi Analogy: 93% in Category A. Holocaust Inversion: 89% in A. Conspiracy Theories: 83% in A. These concepts function as discursive registers that can be invoked without state-rejection. A commenter can compare Israeli prison guards to the Nazi regime, or characterize Israel as having “become what it sought to escape,” while staying within Kristof’s claim 3 and demanding policy reform rather than state dissolution.

This second cluster is the analytically important finding. The standard reading of moderated mainstream-press comment threads would predict that the antisemitism-related registers cluster only on the extreme escalation tail. The data from this corpus disconfirm that prediction. The inversive registers (Nazi Analogy, Holocaust Inversion, Conspiracy Theories) are operating in the ordinary confirmation register of the thread, embedded in comments whose explicit policy demand is the same as Kristof’s (cut aid, condition arms transfers, prosecute the perpetrators). These commenters are not the outliers in the thread. They are the modal reader.

Category B, the methodological-restraint position, is empty of these concepts. The pattern that holds: the antisemitic register is not a property of criticism of Israel. It is a property of how that criticism is structured and extended. Readers who confirm the abuse but flag the column’s inferential structure do not, in this corpus, reach for the registers that saturate Category A.

Synthesis: column, inferential architecture, afterlife

What the column documents

The May 11 column reports a pattern of sexual violence by Israeli prison guards, Shin Bet interrogators, soldiers, and settlers against Palestinian detainees and civilians. The reporting is well-sourced on the central question of whether such abuses occurred. The piece draws on 14 first-person interviews, supplementary interviews with family, lawyers, social workers, and investigators, the Sde Teiman indictments, B’Tselem’s case files, the Public Committee Against Torture in Israel’s complaint volume, Breaking the Silence testimony, and on-the-record acknowledgment from former and current Israeli officials. Reform-oriented Israeli analysts (Rettig Gur, Friedman, and Bashi inside the column itself) converge on the assessment that there is a serious pattern of unpunished misconduct under impunity. I do not contest that assessment. Claim 1 (individual acts occurred) and Claim 2 (widespread, unpunished abuse) are substantially supported by the column’s sources and by independent corroboration.

The inferential architecture

Claim 3 (that the abuse constitutes “an organized state policy” and “standard operating procedure” of the Israeli state) is different in kind. It asserts design, not failure. The column’s sources do not support it. The UN report Kristof cites speaks of “implicit encouragement by the top civilian and military leadership.” Kristof drops the qualifier in his own framing voice. Sari Bashi tells him on the record: “I don’t see evidence that it has been ordered.” Olmert, after publication, issues a Free Press statement explicitly disclaiming the policy framing the column built around his words. §4.5 named six argumentative moves: qualifier-dropping, equivalence-by-presentation, selective hearing of named sources, symmetry as ontological claim, structural conditions framed as evidence of policy rather than as the alternative explanation, and institutional placement as authorization. These are the moves by which a competent columnist transforms a claim-2 evidentiary base into a published statement of claim 3. Each move is defensible in isolation. The methodological problem is what happens when they are deployed in combination on a high-salience charge against a state.

The comment-section afterlife

Three patterns in the dataset turn out to be the most substantive empirical findings here.

First, the column’s claim 3 was taken up by some readers as authorization for further escalation. Once claim 3 is in the discursive record under the Times byline, 65 commenters (9.3%) appear to draw on it as authorization for moves Kristof himself does not make: the genocide characterization (30 instances), the rogue-state framing (14), the colonial-project framing (7), and in six cases the explicit denial of Israel’s right to exist. The column does not say these things. It does not need to. Its inferential structure permits readers who carry those framings into the thread to claim Times-byline support for them, and the post-publication discursive trace of the column is shaped accordingly. Bashi’s claim-2 disclaimer, the column’s most directly claim-2 line, is not visible in the comment-section uptake. The “organized state policy” formulation is.

Second, moderation filters the explicit registers but does not filter the coded ones, and the coded ones operate at the modal-reader layer. Explicit antisemitic registers (slur-level dehumanization, eliminationist call-and-response, explicit Holocaust denial) are largely absent from the NYT thread. The coded layer is present at meaningful frequency and largely uncontested in the comment thread itself. What is new in this corpus, and worth registering as a finding, is that several of these concepts (Nazi Analogy, Holocaust Inversion, Conspiracy Theories) operate in Category A. That is, in the modal confirmation register of the thread, not in the escalation tail. The discursive labor of moderation is in filtering registers that mark themselves as hateful at the surface. The registers that operate through historically-loaded analogy, structural framing, and group-attributed agency pass through and saturate the ordinary register of confirmation. This is a structural feature of moderated comment spaces, not a NYT-specific deficiency.

Third, the corpus does not match the framing that emerged in some post-publication reaction. The American Jewish Committee characterized the column as a “modern-day blood libel.” Cotler-Wunsh used the same register. Both framings carry an implicit claim about reception: that the column produces, at scale, antisemitic uptake of the kind the term blood libel historically denotes. The corpus does not bear that prediction out at the level of the Times-reader population. Explicit blood-libel-register material is absent from the comment thread. The term itself appears once, used by a commenter as a critique of one of Kristof’s specific allegations rather than as part of the discourse the column produces. What the corpus does show is something different and arguably more analytically interesting. The antisemitic registers most active in the thread (Nazi Analogy, Holocaust Inversion, Conspiracy Theories) saturate the modal confirmation layer of the readership, not an extremist tail. That is not the pattern an AJC-style framing would predict. It is the pattern a discourse-analytic framework would predict. Under conditions where explicit registers are filtered by moderation, the analytically significant uptake migrates into coded registers carried through historically-loaded analogy, structural framing, and group-attributed agency. The AJC framing locates the antisemitism in the column’s most sensational specific claims. The discourse-analytic framing locates it in the registers that pass moderation and saturate the ordinary reader’s response. The corpus, on the evidence available, supports the second account.

The joint finding

Taken together, the textual analysis and the comment-section data converge on a single structural point. Once the “organized state policy” framing and the October 7 symmetry frame enter the public record under the Times byline, those are the formulations the readership overwhelmingly retains and extends. The hedges and internal dissenting voices present inside the column (Bashi’s claim-2 disclaimer, Olmert’s Free Press clarification, Kristof’s own “no evidence that Israeli leaders order rapes” and “impossible to know how common sexual assaults against Palestinians are”) are not visible in the uptake. The discursive trace the column produces is shaped by its strongest formulations, not by its qualifications.

The diachronic frame from §12 adds a further layer to this finding. The discursive consequence of correction-asymmetry is not just that the original framing remains in circulation while the correction does not. It is that the uptake of the original framing, the comment-section register documented here, runs ahead of both the framing and the correction. The Kristof column has not been corrected. Olmert’s Free Press clarification appeared elsewhere. Bashi’s disclaimer is visible in the column for any reader who reaches it. The comment-section trace nonetheless shows that the column’s claim 3 is being metabolized at scale. 9.3% of unique commenters carry the argument further than the column itself goes. And the registers most active in the thread (Genocide claim, Conspiracy Theories, Holocaust Inversion, Nazi Analogy) saturate not just the escalation tail but the modal confirmation layer of the discourse the column has authorized.

On the evidence available in this case, the structural finding holds. High-impact framings about Israel and antisemitism, when published under mainstream-press institutional authority, are taken up not at the level of their hedges and qualifications but at the level of their strongest formulation. The discursive uptake reaches for the historically-loaded registers the framing makes available. The Times comment thread on this column is one such trace.

Conclusion

The preceding pages have examined a single opinion column and the comment thread that gathered beneath it. The conclusions worth carrying away from that single case are general only in the disciplined sense that the mechanisms identified here can in principle operate in other cases. Whether and how they do so is a question for further work.

The first general implication concerns the inferential structure of opinion journalism on high-salience topics. Reporting that documents individual abuses, however well-sourced, does not automatically license a published claim that those abuses are systemic, organized, or constitutive of the institution under scrutiny. The move from individual incident to systemic-crime allegation is a distinct inferential step. The six argumentative moves named in §4.5 (qualifier-dropping, equivalence-by-presentation, selective hearing of named sources, symmetry as ontological claim, structural conditions framed as evidence of policy rather than as the alternative explanation, and institutional placement as authorization) are not innovations of this particular column. They are recognizable techniques of opinion-journalistic argumentation, available across many topics and many columnists. What is worth examining is the conditions under which their deployment in combination, on a high-salience charge against a state, transforms documented misconduct into a claim about state design. And what editorial practices would distinguish a column that performs this transformation from one that does not.

A second implication concerns the empirical character of moderated mainstream-press comment spaces. The conventional reading of such spaces treats them as buffered against historically-loaded registers by editorial moderation. The corpus examined here is consistent with a different account. Moderation in such spaces effectively filters surface-explicit hate registers (slurs, eliminationist exhortations, explicit Holocaust denial). It is not designed to filter coded registers that operate through historically-loaded analogy, structural framing, and group-attributed agency. Under conditions where the explicit registers are filtered out, the analytically significant uptake of high-salience framings migrates into the coded registers that pass moderation. This feature is not particular to one publication or one moderation policy. It is a feature of moderated comment spaces as a class. Discourse-analytic work on such spaces would benefit from distinguishing what moderation removes from what moderation does not remove, and from treating the latter not as a residue but as the analytically meaningful layer.

The third implication is the one the two halves jointly support, and it is the central contribution. The framing operations identified in the deconstruction, the six moves catalogued in §4.5, do not operate only at the level of the printed page. The corpus examined here shows that under the institutional authority of a mainstream-press byline, those moves are followed downstream by a readership pattern with two features. First, the column’s strongest formulation (”organized state policy”) is retained and extended while the column’s qualifications and internal dissenting voices are not. Second, historically-loaded antisemitic registers (Nazi Analogy, Holocaust Inversion, Conspiracy Theories) cluster not in the extreme tail of the discourse but in its modal confirmation layer. On the evidence available here, the framing and the uptake are aligned in ways consistent with the framing having shaped or facilitated the uptake. The data show correlation, not causation. The alignment is itself the finding. It distinguishes this analysis from purely textual analyses of opinion journalism on one hand, and from purely empirical analyses of online discourse on the other. The combination is what the discourse-analytic method makes available, and what would be lost if either half were undertaken alone.

A fourth implication concerns the asymmetry between framing and correction in mainstream press institutions. An original framing reaches readers under prominent institutional authority; a correction reaches readers through a smaller, slower, or differently-routed channel. The original framing reaches readers at scale, and the correction documents that the framing did so. This asymmetry is not unique to the cases examined here. It is a feature of contemporary mainstream press operations under conditions of high audience-attention competition and rapid news cycles. What the data above add is that the uptake of the original framing, the readership trace it generates, can run ahead of both the framing and any subsequent correction. The discursive register the framing authorizes continues to operate in the public record after the framing itself has been qualified or walked back.

A fifth implication, more methodological than substantive, concerns evidentiary differentiation. I have worked throughout to distinguish: (i) whether abuses occurred, (ii) whether they are widespread and unpunished, (iii) whether they constitute organized state policy, and (iv) whether comparing them to other phenomena collapses or preserves analytically relevant differences. These distinctions are not pedantic. They are the distinctions that the inferential moves catalogued in §4.5 work to elide. The capacity to keep them separate, in journalism, in policy discussion, and in the public record, is a precondition for serious engagement with the underlying issues. My hope is that the framework offered here, narrow as it is, can be put to use in other cases where the same distinctions are at stake.

I do not claim to have settled questions of editorial intent, institutional pattern, or downstream political consequence. The claim is, more modestly, that a discourse-analytic examination of one column and its reception can produce a named set of argumentative moves, an empirically-grounded account of what the readership did with the column’s framing, and a structural observation about the relationship between the two. Whether the methodology travels to other cases (other columns, other topics, other publications) is the question further work would address.

AddressHate Research Scholar at NYU’s Center for the Study of Antisemitism | Lead, Decoding Antisemitism | Research Advisor, AddressHate | Editor-in-Chief, Digital Hate Review

Executive Summary

In April 2026, Dan Bilzerian — a social media personality with 29.6 million Instagram followers — filed to run for the US Congress against incumbent Randy Fine. Within ten days, he had been interviewed across right-wing, left-anti-imperialist, manosphere, and tabloid outlets, calling his Jewish opponent a “fat Jew” on TMZ and naming “Jewish supremacy” as the greatest threat to America. We analyzed 3,000 comments under six of these video appearances. The findings show a pattern that current digital antisemitism monitoring is not designed to detect: when a high-reach speaker continuously produces antisemitic content, the comment-section response is not a spike that rises and falls around an external event but a baseline that holds steady at saturation levels — averaging 40.8% across our corpus, with no thread below 23.6%, even where journalists pushed back on camera. This study examines what that pattern looks like, why existing analytical approaches do not capture it, and what it means that the institutional vehicle now hosting it is a Republican congressional primary.

Most digital antisemitism research, including the work of this research line, has focused on a familiar pattern: an external event happens, online discourse responds, and the response is what we analyze. This approach rests on three working assumptions — that antisemitic discourse spikes in response to events, that adversarial host framing on camera can reduce audience saturation, and that different forms of hate (antisemitism, anti-Black racism, misogyny, anti-Muslim rhetoric) operate as separable analytical categories. Each captures something real. The AiRT series, the Capital Jewish Museum analysis, the Charlie Kirk and WHCD studies, and the Temple Israel analysis all rest on the first of these — the event-response premise — and have produced rigorous findings under it.

The Bilzerian corpus documents a different configuration. Here, all three assumptions fail at once. They fail not as an exception but as the steady operating condition of a different kind of online hate ecosystem — one driven not by external events but by a sustained primary speaker who keeps the conspiracy material flowing continuously. The argument is not that the event-response approach is wrong. It is that there is a second mode of digital antisemitism that the event-response approach is not designed to detect, and this mode requires its own analytical apparatus.

The framework we propose for this second mode is the Conspiracy Architecture Model. It specifies four inputs and three governing principles that together explain when and how antisemitic saturation activates under speaker-driven conditions. The model treats antisemitic discourse as continuously produced by speaker supply rather than reactively triggered by events. It treats host framing as conditional on whether the audience grants the host legitimacy in the first place — not as an automatic suppressor. And it treats antisemitic conspiracy as a meta-frame that organizes other hate registers underneath it, rather than as one category among many. The model complements rather than replaces the event-conditioned analytical apparatus this research line has developed since 2025. The two modes capture different operating conditions of the same discourse ecology.

The Bilzerian congressional candidacy provides the empirical demonstration: one well-documented case of the speaker-driven mode that current event-conditioned monitoring is not designed to detect. The corpus is structured rather than illustrative. Three threads (Shroyer, Galloway, Jimmy Dore) vary the host context — sympathetic-right, sympathetic-left anti-imperialist, and sympathetic-left commentary — while holding the speaker constant. This tests whether saturation persists across radically different audiences. Two threads (the Sneako pair) form a within-host title-framing test: same host, same speaker, but different framing in the video title. This isolates the effect of framing from the effect of audience composition. The sixth thread (TMZ) is the only configuration in the corpus where adversarial host framing meets a sympathetic audience — a stress test the design did not have to construct, because the media cycle supplied it directly. Across 3,000 YouTube comments from these six clips, three findings establish the speaker-conditioned mode as analytically distinct.

• First, antisemitic saturation operates independently of external triggers. The corpus-average antisemitism density reaches 40.8%. The lowest single-thread rate (23.6%) still exceeds the highest event-conditioned rates documented in prior studies (12–20%). The pattern holds across right-wing, left-anti-imperialist, and left-commentary host contexts, which means this is not a host effect or an audience effect. It is baseline production sustained by a single speaker’s continuous ideological supply.

• Second, adversarial host framing fails when the audience has already decided the host is part of the problem. The TMZ thread is the boundary-condition test. The hosts deliver sustained pushback on camera — they name “fat Jew” as antisemitic and refuse Bilzerian’s redirections. Antisemitism density nonetheless reaches 52%, the highest in the corpus. The mechanism is the finding. Audiences absorb host challenges as further conspiracy evidence rather than registering them as suppression. Harvey Levin’s documented appearance in the Epstein files becomes an activation trigger rather than a journalistic credential. The within-host Sneako pair (post-candidacy vs. pre-candidacy) provides the controlled framing test in the opposite direction: same host, same speaker, only the title varied. Explicit antisemitic framing produces a 29.4-percentage-point increase (53.0% vs. 23.6%). Together, the two tests specify when framing operates as a suppressor and when it does not.

• Third, antisemitic conspiracy operates as the organizing logic for other hate registers. Anti-Black racism, misogyny, anti-Muslim rhetoric, and homophobia in this corpus do not appear alongside antisemitism as separate categories. They are subordinated to the conspiracy meta-frame. Antisemitic conspiracy supplies the causal explanation that links them — formulations like “the media is run by Jews and pushes X” appear repeatedly. Monitoring systems that track hate categories separately measure each layer in isolation and miss the architecture connecting them.

The implications go beyond academic analysis. The Bilzerian candidacy is not just a case of antisemitism. It is the convergence of four streams under a single campaign: manosphere-cultivated audience reach (29.6 million Instagram followers built up over a decade), horseshoe-positioned anti-establishment populism (anti-Trump, anti-AIPAC, pro-Palestinian, anti-Iraq War — coherent as anti-Jewish-power rather than as left or right), openly eliminationist discursive content (Hadrian endorsements, Nova-massacre death wishes, “FINISH THE JOB THIS TIME”), and a federal electoral vehicle (the Republican congressional primary). Each stream has its own research literature. Their convergence does not. This is what mainstreaming looks like at its operational stage: not an abstract worry about gradual normalization, but a documented case in which the streams have already merged and the institutional vehicle is a federal primary campaign. The methodological infrastructure to track this exists. The analytical assumptions under which it currently operates are calibrated for a different mode of the discourse.

Introduction

On April 2, 2026, Dan Bilzerian filed paperwork to run for the US Congress in Florida’s 6th district, declaring his intent to challenge Republican incumbent Randy Fine in the 2026 Republican primary. Within forty-eight hours the announcement had been carried across the alternative-media circuit — Owen Shroyer (former Infowars), George Galloway, Sneako, the Hodgetwins, OANN, Jimmy Dore — and within ten days had crossed into mainstream coverage with a TMZ live segment in which Bilzerian used a racial slur on camera. His organizing framing, deployed under the heading of “Jewish supremacy” or “Jewish supremacist mindset,” delivers a comprehensive conspiracy architecture: claims of Jewish institutional control of US foreign and domestic policy, conspiracy claims about media and finance, and disloyalty attributions directed at Jewish American elected officials including Randy Fine specifically. The phrase “fat Jew” surfaces once in his own voice, on TMZ, when host Harvey Levin reads it back to him as evidence of antisemitism and Bilzerian declines to retract it. Across the other five clips it does not function as Bilzerian’s own framing of his opponent — Sneako and the Jimmy Dore Show recirculate the TMZ moment, and Shroyer riffs on it as commentary, but Bilzerian’s preferred framing of Fine is “Jewish supremacist.” The “fat Jew” line is the campaign’s most quotable moment in downstream coverage; it is not Bilzerian’s organizing framing of his opponent.

What this signifies in the institutional register matters. The speaker is a 45-year-old Navy veteran, a former SEAL trainee, the holder of a National Defense Service Medal, the candidate of one of the two major parties in a US congressional district. He is calling a sitting Jewish member of Congress a “fat Jew” on national television and arguing that “Jewish supremacy” is the greatest threat to America — not in a private conversation, not in a deniable joke, but as material delivered across multiple long-form interviews and doubled down on when challenged. The institutional infrastructure that hosts a candidacy of this kind — the FEC filing system, the primary ballot, the journalistic interview booking, the social media platform that monetizes the resulting clips — was not designed for speech of this register and has not adapted to it. What has changed is not that an antisemitic celebrity exists in American public life. He has existed for years. What has changed is that the institutional vehicle of the federal congressional campaign now hosts him without apparent friction.

The Structural Problem

Across this research line — the Capital Jewish Museum analysis (May 2025), the Antisemitism in Real Time series (June 2025–March 2026), the Charlie Kirk assassination study (September 2025), the Bondi Beach analysis (December 2025), the Temple Israel analysis (March 2026), and most recently the White House Correspondents’ Dinner study (April 2026) — political-violence events have served as the structural anchor: a discrete event triggers a wave of social media response, and the response is the analytical object. Bilzerian breaks that structure. He is a sustained content producer whose output is the antisemitic discourse the rest of the research line tracks as response to other events. The congressional run supplies the trigger event in name only; the corpus design treats it as an institutional crossing point rather than as a stimulus.

A Unified Analytical Framework

Current monitoring approaches treat digital antisemitism analysis as three independent problems — when it spikes, who frames it, what it targets. This division of analytical labor has been productive for event-conditioned discourse and remains so. The Bilzerian corpus, however, demonstrates that under speaker-driven conditions these are not separable problems: they are three views of a single integrated system. The framework below extends rather than replaces the prevailing paradigm — it adds the analytical apparatus needed for the configuration this corpus exemplifies, while leaving event-conditioned analysis intact for the configurations it captures well.

The Conspiracy Architecture Model

Four inputs:

1. Event-context supply — external triggers (wars, attacks, crises) that activate the architecture

2. Audience-ideological supply — pre-existing audience receptivity to antisemitic conspiracy frameworks

3. Speaker-supplied scaffolding — direct provision of conspiracy frameworks by the primary speaker (this study’s contribution)

Intermediary framing strategy — host/outlet amplification or challenge of the content

Three governing principles:

• Principle A — Speaker substitution: When input #3 is high, input #1 becomes unnecessary. Sustained speaker-supplied scaffolding produces baseline saturation that exceeds event-triggered spikes, decoupling the architecture from external triggers.

• Principle B — Framing dependence on audience legitimacy: Input #4 only suppresses saturation when input #2 grants the framer legitimacy. When audiences arrive pre-coded against the framer, adversarial framing is absorbed into the architecture as additional evidence rather than functioning as suppression.

• Principle C — Hierarchical multi-target organization: The architecture is not narrowly antisemitic. It operates as a meta-frame that subordinates other hate registers (anti-Black racism, misogyny, homophobia, anti-Muslim rhetoric) through formulations like “the media is run by Jews and pushes X.” Multi-target hate is the architecture’s organizational logic, not a parallel finding.

The terminology in this article — speaker-conditioned saturation, scaffolding, audience-ideological supply, conspiracy-organized multi-target embedding — describes a single mechanism viewed from different angles, not multiple competing mechanisms. Saturation names the steady-state output. Scaffolding names the input by which a speaker produces it. Audience supply names the receptivity condition that determines whether scaffolding takes hold. Multi-target embedding names the hierarchical structure under which adjacent hate registers are organized. Together they describe one architecture; each term picks out a feature of the same system.

The three principles are interlocking rather than parallel. Each falsifies one of the three foundational assumptions named at the start of this article:

• Principle A (Speaker substitution) → falsifies the assumption that antisemitic discourse spikes in response to events. The trigger condition is internal to the speaker rather than external to the discourse.

• Principle B (Framing dependence on legitimacy) → falsifies the assumption that adversarial host framing suppresses audience saturation. The suppression mechanism is conditional on legitimacy granted by audience pre-coding.

• Principle C (Hierarchical organization) → falsifies the assumption that hate categories operate as separable analytical categories. The targeting structure is hierarchical.

The scope claim is precise: monitoring frameworks built on these three assumptions describe one mode of digital antisemitism well, but cannot detect, measure, or intervene against the speaker-driven mode documented here. They are not under-tuned for this configuration; they are calibrated for a different one. The Bilzerian corpus operates outside all three founding assumptions of the event-conditioned mode. It is not an outlier within the speaker-driven mode; it is what that mode produces.

The Candidate: Dan Bilzerian as Empirical Instance

Dan Brandon Bilzerian, 45, is analytically significant for this study not because of personal biography but because his candidacy provides a clear empirical instance of the structural condition the Conspiracy Architecture Model describes. The model is not about Bilzerian; Bilzerian is one well-documented case of a speaker who supplies the conspiracy framework sustainedly enough to produce speaker-conditioned saturation in his audience. Other cases — different speakers, different platforms, different political vehicles — would activate the same architecture under the same governing principles.

What makes Bilzerian empirically useful for the model is the volume and consistency of his pre-candidacy output. Across more than a decade of social media production — 29.6 million Instagram followers, an established YouTube presence, sustained podcast circulation, and brand affiliations including a poker venture and merchandise lines — he has operated as a high-reach producer of explicitly antisemitic content. The scale matters for the analysis: this is not a fringe figure crossing into electoral speech. The audience-ideological supply that the Phase 3 analysis traces was already in place years before the candidacy filing, cultivated by the speaker himself across platforms.

The substantive output spans the standard inventory of antisemitic conspiracy claims. He has attributed responsibility for the Holodomor, the Kennedy assassination, September 11, and the Iraq War to Jewish actors. He argues that Nazi Germany “flourished” after “Hitler removed Jews from the banking system” and that Spain’s golden age followed the 1492 expulsion. The “109 countries” trope appears repeatedly across his commentary; the financial-control register is sustained; characterizations of Jews as parasitic on host societies recur. The November 2024 Piers Morgan interview is the most-cited public escalation: Bilzerian told Morgan that Judaism “promotes supremacy and rape,” that Holocaust death tolls were inflated, and that Jews “basically invented genocide.” When challenged on any of these claims — by Morgan, by Levin at TMZ, by other interviewers — the consistent pattern is doubling down rather than retreat. The doubling-down pattern is itself part of what makes the case useful for the model: the framework is not deniable, accidental, or context-dependent.

Randy Fine, the opponent the candidacy targets, is a sitting Republican congressman from Florida’s 6th district and one of the most publicly Jewish members of the current Congress, having centered his Jewish identity in legislative work on antisemitism, Israel, and Holocaust education. Bilzerian’s framing of Fine as “Jewish supremacist” — paired consistently across the corpus with the disloyalty/treason register — is therefore not incidental targeting. It is the application of the conspiracy framework to a specific Jewish elected official, with the candidacy serving as the institutional vehicle through which it is delivered into electoral discourse.

A note on Bilzerian’s political positioning bears directly on the corpus design. He is running as a Republican but his actual issue positions cut across the conventional left/right axis: sustained anti-Trump (a Republican primary candidate attacking the party’s leader as a “narcissistic pathological liar”), anti-AIPAC and anti-Israel, pro-Palestinian, anti-Iraq War, anti-interventionist, paired with right-coded positions on immigration and LGBTQ visibility. The ideological coherence is anti-establishment / anti-Jewish-power rather than partisan. This is what makes the cross-spectrum sympathetic platforming visible in the corpus possible: Owen Shroyer (former Infowars right-populist), George Galloway (heterodox left-anti-imperialist, expelled from Labour over the Iraq War), Sneako (manosphere influencer), and Jimmy Dore (left-populist commentator) all host Bilzerian sympathetically because his anti-Israel / anti-AIPAC framework converges with their respective anti-establishment positions across the horseshoe. The corpus design treats this convergence as analytical opportunity: if speaker-conditioned saturation holds across host audiences this politically heterogeneous, the speaker substitution principle is empirically supported by the variation rather than weakened by it.

The congressional candidacy represents an institutional crossing point: antisemitic content that operated for years under platform content policies now functions as electoral speech with different regulatory frameworks. The campaign’s organizing frame — “Jewish supremacy” — supplies a sustained set of conspiracy claims about Jewish institutional control directed at a sitting Jewish member of Congress. This is what makes the case empirically useful for the model: the framework is non-ambiguous, well-documented across a decade, delivered through a federal electoral vehicle, and reinforced by a pre-existing audience-ideological supply that the speaker himself cultivated.

The Convergence

What this case represents is not antisemitism alone, not manosphere reach alone, not horseshoe-positioned anti-establishment populism alone, and not open endorsement of violence against Jews alone. It is the convergence of all four into a single federal congressional candidacy. The manosphere stream supplies the audience: 29.6 million followers cultivated over a decade through hyper-masculine spectacle (poker, weapons, wealth display, women), now converted into pre-loaded political audience-ideological supply. The horseshoe positioning supplies the cross-spectrum platforming: sympathetic interviews from Infowars-alumni right-populists (Shroyer), anti-imperialist left hosts (Galloway), manosphere influencers (Sneako), and left-populist commentators (Jimmy Dore) alike. The eliminationist register supplies the discursive content visible across the corpus: “wiped off the map,” “Hadrian was successful,” “FINISH THE JOB THIS TIME,” Nova-massacre death wishes directed at named Jewish commenters, Leo Frank lynching threats, yellow-star endorsements, Goebbels quotations laundered as authority. Each stream has its own research literature. Their convergence under a single federal campaign vehicle does not.

This is what mainstreaming looks like at its operational stage — not abstract concern about gradual normalization, but a documented case in which the streams have already merged and the institutional vehicle is the federal Republican primary. The structural conditions that current event-conditioned monitoring is designed for — discrete violence events, identifiable extremist platforms, category-isolable hate registers — are not the conditions under which this convergence operates. It operates as steady-state production by a high-reach speaker with cross-spectrum platforming and an audience already fluent in the conspiracy framework. Detection requires recognizing the pattern as such; the pattern does not announce itself as extremism, because it has already passed through the institutional filter that would have flagged extremism, and arrived at the other side wearing the credential of a federal congressional candidacy.

Dataset and Method

This study draws on six widely circulated YouTube clips from the March–April 2026 Bilzerian media cycle, spanning the pre-candidacy media cycle and the immediate post-announcement window. Five of the six threads operate within the sympathetic-amplifier ecosystem. The sixth thread (TMZ) occupies a hybrid Phase 2 condition: hosts Harvey Levin and Charles Latibeaudiere push back on Bilzerian’s framings directly on camera, while the audience composition operates in sympathetic-amplifier mode toward Bilzerian. The TMZ thread functions as the corpus’s stress test for whether audience-supply conditions or host-framing conditions dominate the Phase 3 outcome.

Selection prioritized structural coverage of the host-context spectrum (sympathetic-right, sympathetic-left anti-imperialist, sympathetic-influencer, sympathetic-left commentary, partial-adversarial tabloid) and within-host title-framing variation (the two Sneako threads).

Comments were collected during April 10–16, 2026. From each video, the first 500 comments were drawn, with reply chains preserved for sequence and relational analysis where relevant; spam and duplicates were excluded. The 3,000-comment sample (500 per thread × 6 threads) is the analytical unit.

The corpus comprises (organized by ID, with threads grouped by host-context type rather than chronologically):

The corpus IDs reflect host-context grouping (the two Sneako threads are paired as DB-03/DB-04 to support within-host comparison) rather than chronological order. Chronologically ordered results appear in the Phase 3 quantitative table. Two videos that appeared in earlier corpus versions (a Hodgetwins reaction stream and an OANN segment) were dropped during validation due to redundant Phase 2 framing relative to other sympathetic-amplifier threads in the corpus.

All six transcripts were read in full for the Phase 1 coding presented below. Phase 3 quantitative results cover all six threads (N = 3,000 comments), providing complete coverage of the corpus.

Three Strata of Analysis

The analysis maps onto the three-phase circulation model developed in the Mamdani study (Becker, Beacken, Sabra, January 2026):

The Bilzerian study reverses the Mamdani design: Mamdani held Phase 1 ambiguous and varied Phase 2 framing; Bilzerian holds Phase 1 non-ambiguous (the speaker explicitly supplies the conspiracy framework) and varies Phase 2 along the sympathetic-to-partial-adversarial spectrum. Because Bilzerian’s Phase 1 discourse is itself the object of analysis (not just framing context for Phase 3), this study codes Phase 1 explicitly as a stratum, applying the same DA Lexicon framework used for Phase 3.

Coding

Analysis followed the principles of qualitative content analysis, with native-speaker coders applying linguistic and multimodal close reading supplemented by sequence and relational analysis where comments form part of reply chains. Coding combined deductive categories from the Decoding Antisemitism Lexicon (Becker et al., Springer Nature, 2024), which encompasses 41 antisemitic concepts organized across categories including Otherness, Moral/Supernatural Defamation, Dehumanization/Blood Libel, Deceit/Blame, Power/Conspiracy/Economic Domination, Victimhood Inversion, Holocaust & Memory Politics, Israel-Related antisemitism, and Aggressive Speech Acts — with inductive coding of formations specific to the Bilzerian case. For other-target hate registers (misogyny, anti-Muslim racism, anti-Black racism, homophobia/anti-trans rhetoric), coding followed parallel discourse-analytic principles without imposing the DA Lexicon framework, which is calibrated for antisemitism specifically. Inter-coder reliability was addressed through iterative calibration sessions on borderline cases.

Three case-specific coding decisions bear noting. First, the candidacy-specific “vote for him on that alone” register is coded as power/conspiracy confirmation since the electoral endorsement operates as audience confirmation of the scaffold rather than as a distinct discursive move. Second, the Randy Fine characterization cluster (”fat Jew” plus commenter-supplied variants like “tinyhatism”) is coded under dehumanization with secondary conspiracy elements, since the framing operates simultaneously as bodily mockery and as instantiation of the Jewish-power frame at the level of an individual political opponent. Third, short endorsement comments on threads saturated with explicit conspiracy content are coded as co-textual antisemitism with a confirmation sub-code; comments that read as politically neutral in isolation participate in confirming the scaffold when read in the surrounding conspiracy environment.

A range of GPT and Claude models were used to support the processing of the raw corpus — comment extraction, structuring, deduplication, and pre-identification of overt conspiracy and explicit-violence content for human coder triage. The models were not used to make coding decisions and were not used to draft the analytical text. All category assignments, qualitative readings, and analytical claims in this article are the work of human coders.

Phase 1: Speaker-Supplied Scaffolding

This section documents Bilzerian’s primary speech as input layer for the Conspiracy Architecture Model — the speaker-supplied scaffolding that Principle A identifies as the system’s trigger condition. The analysis is not exhaustive discourse analysis of Bilzerian as such; it is targeted documentation of the components that feed directly into the Phase 3 audience response. Three categories matter for the model: (1) the conspiracy architecture that operates as the corpus-wide organizing frame; (2) the antisemitic sub-registers (Israel-related, Holocaust, otherness/deceit, dehumanization) that elaborate the architecture and pre-load specific Phase 3 uptake; (3) the other-target hate registers embedded under the conspiracy meta-frame per Principle C, treated in the subsection that follows.

Antisemitism in Bilzerian’s Discourse

Power/Conspiracy — dominant register across all six clips

Bilzerian’s organizing concept is “Jewish supremacy,” delivered with explicit attempts to displace “Zionism” as the alternative framing. This framework recurs across every clip in the corpus and provides the meta-frame under which his other claims about Jewish institutional control are organized:

“I don’t frame it as a Zionist problem because I believe that it’s a Jewish supremacy problem... it’s a religious issue. I think it’s because they think that they’re better than everybody, that they’re chosen from God.” (DB-05)

The conspiracy architecture extends across foreign policy, finance, and media:

“Netanyahu came and he testified in front of Congress that they had weapons of mass destruction... we fought the Iraq war... for them really.” (DB-05)

“Homosexuality and transgenderism... are being pushed on us by the Jewish media, by you know, Hollywood, which is predominantly Jewish.” (DB-05)

The dual-loyalty register targets Jewish elected officials:

“It’s Jew first, Israel second, America last. And it’s like that with all of them.” (DB-04)

Bridge to the Black-Hebrew-Israelite Register

Bilzerian deploys the antisemitic slave-trade trope at DB-01:

“Jews were disproportionately responsible for the slave trade and they shut the slave trade down on all Jewish holidays because not enough people would show up... to the auction.” (DB-01)

The supply is structurally significant for Phase 3 analysis because it pre-loads the audience for the Black-Hebrew-Israelite register that surfaces at DB-06 — the framing in which contemporary Jews are recoded as imposters and “real” Jewish identity is relocated onto Black populations (”the real jews are us so called black people,” DB-06 11-112). The slave-trade supply establishes a direct causal narrative linking Jewish actors to Black suffering, which the BHI audience then completes by claiming the displaced identity. Without Bilzerian’s Phase 1 supply, the BHI uptake at DB-06 would read as audience importation from outside the corpus. With it, the register operates as continuation of speaker scaffolding.

Israel-Related antisemitism and eliminationist register

Bilzerian’s Israel-Related discourse fuses delegitimization (Israel as inherently terrorist) with explicit eliminationist endorsement. The two operate as a single move: if Israel is a terrorist state by foundation rather than by policy, then its elimination is justified as moral correction rather than as state-on-state violence. The clearest instances:

“[Israel is] a terrorist nation. They were founded by terrorists and they’ve been committing terrorism since their inception.” (DB-06)

“I would sign up tomorrow and go put boots on the ground and [eliminate Israel]... they need to be wiped off the map.” (DB-04)

The “wiped off the map” formulation, delivered in the explicit antisemitism-framed Sneako appearance, takes the founding-as-terrorist claim to its operational conclusion. The Phase 3 audience picks up both registers — delegitimization claims circulate widely; eliminationist endorsements appear as bounded but recurring features of the comment sections.

A linked etymological-redirection register surfaces at TMZ, where Bilzerian rejects the term “antisemitic” itself when Levin applies it:

“I’m not anti-Semitic. I think that’s kind of a made up term. I think the Palestinians are the real Semites... If you look at what a Semite is, was Hitler anti-Semitic?” (DB-06)

The “real Semites are Palestinians” move performs a delegitimization of the term “antisemitism” itself: by rerouting the descriptor onto Arab populations, Bilzerian frames antisemitism not as Jew-hatred but as a category error that contemporary Jews have appropriated. The “was Hitler anti-Semitic?” rhetorical question makes the redirection explicit — if “Semite” properly denotes Arabs rather than Jews, then even Hitler’s hatred of Jews could not technically be antisemitic. The move is an etymological argument deployed to delegitimize the existence of a recognized form of bigotry. This audience picks it up at scale (the Phase 3 “fake Semites” / “real Semites are Arabs” cluster across multiple threads), and it functions as a bridge between Bilzerian’s Israel-Related discourse and his disavowal of the antisemitism category.

Holocaust relativization and denial bridge

Bilzerian provides systematic relativization that primes audience denial content:

“All these instances where Jews are mass murdering Christians, which they never talk about in the textbooks... the Bolshevik genocide, the Holodomor.” (DB-05)

“Almost all those places are places where it’s illegal to deny the Holocaust, and I don’t think that’s a coincidence.” (DB-04)

The DB-05 audience produces explicit denial content — “If we had smart phones 80 years ago there would be no holly narrative” (113 likes) — that operates as direct audience uptake of Bilzerian’s relativization.

Otherness / Deceit-of-allegiance

The “secretly hates” framing operates as the relational equivalent of the conspiracy framework: instead of “Jews control institutions,” the claim is “Jews pretend friendship while harboring hostility.” This register matters because it sets up the broader argument that Jewish disloyalty is constitutive rather than situational — Jews cannot be relied upon as Americans because the affective relationship is itself dissimulation:

“Don’t pretend to be my friend. Don’t pretend to be my ally and then secretly hate me. That’s my issue... cuz Muslims don’t secretly hate Christians.” (DB-05)

The comparative move — Muslims as straightforwardly hostile, Jews as deceptively so — reproduces the standard otherness/deceit pairing in which Muslims are positioned as honest enemies and Jews as dishonest pseudo-friends. This sets up the disloyalty/treason register Bilzerian applies to Jewish elected officials including Fine.

Dehumanization

The clearest dehumanization material in Bilzerian’s primary speech is the Epstein sub-register at DB-04, which produces blood-libel-adjacent framing of Jewish involvement in child sexual abuse as a structural rather than individual phenomenon — generalizing criminal allegations into Jewish-coded pathology rather than restricting them to the individuals named in legal proceedings.

A clarifying note on the “fat Jew” characterization of Randy Fine, since it is the campaign’s most quotable downstream moment: Bilzerian uses the phrase only once in his own voice, at TMZ. Across the other five clips it appears only in hosts’ framings (Sneako replaying TMZ, Jimmy Dore playing the clip, Shroyer in his own voice). Bilzerian’s preferred framing of Fine across all six clips is “Jewish supremacist,” consistently paired with the disloyalty register. The “fat Jew” line is therefore one episode-level deployment rather than the corpus-wide signature it has become in downstream coverage.

Other-Target Hate Registers in Bilzerian’s Discourse

Bilzerian’s primary discourse is not exclusively antisemitic. It contains substantial misogyny, anti-Muslim and anti-Arab racism, homophobia and anti-trans rhetoric, and anti-Black framings — frequently fused with the conspiracy antisemitism framework through what we term conspiracy-organized multi-target embedding: the structural mechanism by which other-target hate registers are subordinated to the dominant conspiracy architecture and given causal explanation through it. The formula “the media is run by Jews and pushes X” — where X is variously LGBTQ visibility, anti-white messaging, rap, or transgender ideology — is the most common instantiation of this mechanism.

Misogyny

The Erica Kirk segment in DB-04 elaborates a sustained character attack on a recent widow (”complete attention whore... she’s jumping up and down on stage with fireworks going off”). The DB-05 anti-feminist register fuses misogyny with white grievance: “every single movie you’ve got some subservient beta male... his wife is usually cheating on him... they’re basically attacking the white Christian man” — both registers routed through the media-control conspiracy framework.

Anti-Muslim and anti-Arab racism

Bilzerian’s anti-Muslim register operates through a structural contradiction worth flagging because it surfaces in audience response. He attacks Fine specifically for anti-Muslim rhetoric — positioning himself as the candidate defending Muslims from a Jewish congressman’s Islamophobia — while simultaneously deploying standard anti-Muslim immigration tropes about the Muslim populations he claims to defend. The DB-02 framing (”military age men from war torn countries that have no allegiance to the land”) deploys the standard anti-Muslim immigration frame; DB-03 intensifies to mass-rape framings without sourcing. The contradiction is not incidental: it allows the conspiracy framework to absorb both pro-Muslim positioning (against Fine) and anti-Muslim claims (against immigrant populations) under the same Jewish-supremacist meta-frame, with Jewish elites positioned as the true threat regardless of which Muslim population is invoked.

Homophobia and anti-trans rhetoric

Three registers, including LGBTQ visibility framed as Jewish-media imposition, “gay” as pejorative (”It was like some stupid gay rah-rah”), and conspiracy-fused anti-trans rhetoric (”The Talmud has eight different genders. The Bible has two. The Quran has two”) — locating gender plurality in Jewish religious text as evidence of Jewish supremacy.

Anti-Black racism

The clearest instance is DB-05’s Candace Owens framing endorsing “black crime in this country” as a legitimate analytical category. A fused register at DB-04 routes anti-Black framings through the conspiracy framework: “Rap should be canceled. Private prisons pay rappers to talk about committing crimes and drugs.”

Anti-Trump positioning as a fused register

Across all six clips, Bilzerian is sustainedly anti-Trump in a register distinct from his conspiracy architecture even while it borrows from it. DB-06 opens with “every decision that Trump makes is based on what is best for him and his ego... this is not America first. It’s been Israel first.” The framing recurs across all clips, fusing with conspiracy antisemitism through the formula “America first turned out to be Israel first.”

This is not a hate-target register parallel to the others above; it is a political-positioning register that the conspiracy framework absorbs. It matters for Phase 3 analysis: a non-trivial share of comments across DB-01, DB-04, and DB-05 endorse Bilzerian electorally on the basis of Trump betrayal rather than exclusively on the conspiracy framework. The two registers fuse, but the entry point for some commenters is Trump disappointment, not antisemitism.

Summary of Phase 1

Bilzerian’s primary discourse is multi-target with antisemitism organized under power/conspiracy claims as the dominant and most-elaborated register — present in all six clips, structurally consistent across them, and developed under the explicit heading of “Jewish supremacy.” The conspiracy framework operates as the unifying meta-frame under which other hate registers (misogyny, anti-Muslim racism, homophobia, anti-Black framings) are organized and given causal explanation. This systematic provision of antisemitic conspiracy frameworks supplies the speaker-conditioned input that the Phase 3 analysis traces through audience response.

The Analytical Role of Each Thread

Before turning to the Phase 3 results, it is worth making explicit what each thread contributes to the model. The corpus is not six instances of the same finding. It is a structured test of different aspects of the Conspiracy Architecture Model, with each thread performing distinct analytical work.

Shroyer, Galloway, Jimmy Dore (DB-01, DB-02, DB-05) — variation across the host-context spectrum (testing Principle A). These three threads vary host-context type (sympathetic-right, sympathetic-left anti-imperialist, sympathetic-left commentary) while holding the speaker constant. The analytical question they address is whether speaker-conditioned saturation holds across radically different audience compositions. Rates do vary across the cluster (46.8%, 27.4%, 42.0%), and that variation is itself analytically informative — see the next paragraph on sympathetic-amplifier vs. sympathetic-restrained hosting. What does not vary is the relationship to the event-conditioned ceiling: all three rates exceed the highest event-conditioned rates documented across prior studies (12–20%). Speaker-conditioned saturation is the floor that holds regardless of host type, not the ceiling — which means the principle is supported even though host amplification modulates how high above the floor any given thread runs.

Galloway (DB-02) — refining the sympathetic-amplifier distinction. DB-02 sits within the cluster above but produces a distinctively lower rate (27.4%) despite clear sympathetic hosting. This is analytically informative rather than anomalous: Galloway provides supportive but measured platforming without actively elaborating Bilzerian’s antisemitic claims, while Shroyer (DB-01) and Jimmy Dore (DB-05) function as sympathetic amplifiers who pick up and extend the speaker’s framing. DB-02 demonstrates that “sympathetic” is not a binary condition — it admits a distinction between sympathetic-restrained and sympathetic-amplifier hosting, and the distinction matters quantitatively. This refines rather than contradicts Principle A: sustained speaker supply produces above-baseline saturation in all three sympathetic threads, but host amplification drives it substantially higher.

Jimmy Dore (DB-05) — Principle C in operation. DB-05 demonstrates conspiracy-organized multi-target embedding at its densest: Hitler rehabilitation, Holocaust denial, anti-trans content, anti-feminist framings, and historical-authority deployment all routed through the same Jewish-media-control meta-frame. The thread is the corpus’s clearest illustration of how adjacent hate registers are subordinated to the conspiracy framework rather than co-occurring with it. The 42% antisemitism density understates the multi-target reach: the thread also produces substantial misogyny, anti-trans content, and anti-Christian theological framings, all organized through the conspiracy frame.

Sneako pair (DB-03 / DB-04) — within-host title-framing test. The two Sneako appearances isolate Phase 2 title-framing effects from cross-host audience-composition effects. This is the only configuration in the corpus that controls for everything except framing direction: same host, same speaker, same baseline audience, varied only by title (explicit antisemitic framing in DB-04 vs. general electoral framing in DB-03). The 29.4-percentage-point difference (53.0% vs. 23.6%) demonstrates that Phase 2 framing retains substantial effect even under sustained speaker supply. This refines Principle A: sustained speaker supply produces saturation, but framing emphasis still modulates the saturation level when audience composition is held constant.

TMZ (DB-06) — Principle B stress test. DB-06 is the only thread in the corpus where host-framing variable and audience-supply variable pull against each other. Adversarial host framing meets sympathetic audience composition, producing the conditions under which the boundary of framing-as-suppressor can be tested directly. The result — 52% antisemitism density, with audiences absorbing host pushback into the conspiracy framework — establishes Principle B: framing functions as suppression only when audience grants framer legitimacy.

Together, the corpus performs three analytical operations. First, it establishes that speaker-conditioned saturation holds across host-context variation (DB-01, DB-02, DB-05). Second, it isolates the framing variable both within-host (DB-03/DB-04 title framing) and cross-host with adversarial framing (DB-06). Third, it demonstrates the multi-target organizing logic of the conspiracy framework (DB-05 most densely, but visible across the corpus). Each thread is doing different analytical work; the redundancy across rates is the finding, not the design.

Phase 3: Audience Response

Quantitative Overview

As of this draft, all six threads have been fully coded (N = 3,000 comments), providing a complete picture of antisemitism saturation across the Bilzerian corpus.

Corpus Average: 40.8%

Three findings emerge.

Finding 1 — speaker-conditioned saturation across host contexts. The lowest figure across the six coded threads (23.6%, Sneako post-candidacy) exceeds the highest rate recorded across the AiRT series (12–20%), the Charlie Kirk assassination study (~3.6%), and the WHCD study (0.5–4%). The corpus average is 40.8%. These rates are not consistent with event-conditioned production; they require speaker-conditioned saturation as the explanatory mechanism — antisemitism levels keyed to a sustained primary speaker rather than to discrete external events.

Finding 2 — Conspiracy-narrative dominance, with structural variation. Antisemitic conspiracy narratives about Jewish power and control are the modal category across all six coded threads. Cluster share runs at approximately 65–75% in the five sympathetic-amplifier threads. The TMZ thread shows a structurally distinctive distribution: conspiracy share at 41% with elevated otherness/deceit registers (21%), driven by the Black-Hebrew-Israelite-adjacent register (the “real Jews are us” framing in which contemporary Jews are recoded as imposters) and adjacent Khazar-theory delegitimization claims that the racial composition of the TMZ comment section surfaces. Conspiracy narratives remain the modal register at TMZ by a wide margin, but the secondary registers are thicker. The slave-trade sub-register Bilzerian supplies at the Shroyer thread provides the bridge: where Phase 1 has him saying “Jews were disproportionately responsible for the slave trade,” the TMZ audience produces “the jdl was founded to lynch blacks as a mossad cut outs... seeking reparations for synagogues running new York slave sales” (TMZ 11-001, 67 likes). The continuity is direct.

Finding 3 — Principle C in operation: hierarchical multi-target organization. The Jimmy Dore thread is the corpus’s clearest illustration of multi-target hate organized through the conspiracy meta-frame. At 42% antisemitism density — third-highest in the corpus, after the explicitly antisemitic-framed Sneako thread (53.0%) and Shroyer (46.8%) — DB-05 produces the corpus’s most prominent Hitler rehabilitation content alongside extensive deployment of historical authorities (Winston Churchill, Rabbi Ovadia Yosef) to support conspiracy claims. The 42% headline figure understates the multi-target reach: misogyny, anti-trans content, anti-Christian theological framings, and Hitler rehabilitation are all routed through the same Jewish-media-control meta-frame. Principle C does not predict that antisemitism rates will be highest in such threads; it predicts that adjacent hate registers will be subordinated to the conspiracy meta-frame rather than co-occurring with it. The Jimmy Dore thread is the cleanest demonstration of that subordination.

The sympathetic-amplifier vs. sympathetic-restrained distinction. Despite Galloway’s clear support for Bilzerian, the Galloway thread shows 27.4% antisemitism density — the second-lowest in the corpus. This refines rather than contradicts Principle A. Galloway provides supportive but measured platforming without actively elaborating Bilzerian’s antisemitic claims; Shroyer and Jimmy Dore function as sympathetic amplifiers who pick up and extend the speaker’s framing. “Sympathetic” is not a binary condition — sustained speaker supply produces above-baseline saturation across all sympathetic threads, but host amplification drives it substantially higher.

Antisemitic Conspiracy Narratives: Jewish Power and Control

The dominant audience register, with sub-clusters reproducing the elements supplied at Phase 1 plus several thread-specific extensions.

Electoral-endorsement sub-register

Comments endorse Bilzerian electorally on the explicit basis that he names Jewish institutional control:

“Dan would shit down shop for the Ghews exploiting America I’d vote for him on that alone.” (DB-02)

“Dan for president!!” (DB-05 0066, 132 likes)

“Dan needs to run for president or lead a revolution” (DB-05 0050, 136 likes)

The third example explicitly positions Bilzerian as revolutionary leader rather than electoral candidate — the conspiracy framework generates eliminationist rather than democratic solutions. The keyword-evading orthography (”Ghews” for “Jews”) indicates platform-evasion awareness; the “vote for him on that alone” formulation makes the framework electorally functional.

Historical authority deployment (DB-05)

The Jimmy Dore thread surfaces systematic deployment of historical authorities to support conspiracy claims:

“Winston CHURCHILL, on COMMUNISM in RUSSIA, Sunday Herald February 8, 1920... ‘rise the schemes of the International Jews... This movement among the Jews is not new. From the days of Spartacus-Weishaupt to those of Karl Marx, and down to Trotsky (Russia), Bela Kun (Hungary), Rosa Luxemburg (Germany), and Emma Goldman (United States), this worldwide conspiracy for the OVERTHROW OF CIVILIZATION...’” (DB-05 0001-49, 0013-30, 0031-24 — repeated 3x across thread)

“Rabbi Ovadia Yosef a Talmudic scholar... ‘GOYIM were born only to SERVE US. Without that, they have NO PLACE in the WORLD; only to serve the People of Israel’” (DB-05 0001-48, 0013-29)

The Churchill deployment is structurally significant: a mainstream political authority positioned as validating contemporary conspiracy claims. The Rabbi Ovadia Yosef quotation performs religious authority for the supremacist claim — not a conspiracy theory but direct quotation from Jewish religious leadership.

Repetition as validation (DB-05)

Key conspiracy claims circulate through multiple comment stations:

“To learn who rules over you ask who you are not allowed to criticize” — circulated 4x across the thread (0003-9, 0013-26, 0019-12, 0031-22)

The “Voltaire” quote (actually from a neo-Nazi) achieves meme status through repetition — the appearance of independent validation through multiple sources citing the same false authority.

Within-Sneako title-framing test (DB-04 vs DB-03)

The two Sneako appearances enable direct comparison of title-framing effects under maximally controlled conditions. DB-04 (”Understanding Jewish Supremacy,” March 4) is framed around explicit antisemitic content; DB-03 (”Running for Congress,” April 10) is framed around the electoral campaign. Both feature the same host, the same primary speaker, and audiences drawn from the same baseline population, with only the title framing varied. The result: DB-04 produces 53.0% antisemitism density versus DB-03’s 23.6% — a 29.4-percentage-point increase attributable to explicit antisemitic title framing. This demonstrates that Phase 2 framing retains substantial effect even under conditions of sustained speaker supply.

The DB-04 audience produces several distinctive power-and-conspiracy formations not present in the other threads. The surname-conspiracy register operates through wordplay on Jewish names:

“very Cohencidental” (DB-04 0012-13)

“Cohencidence” (DB-04 0012-16)

The “Cohen” wordplay treats Jewish surname prevalence as evidence for conspiracy claims — a compact fusion of identity-marking and structural-control attribution. The register also surfaces direct endorsement of antisemitic hatred as a criterion for political support:

“anyone who shits on jews is a legend . cry about it” (DB-04 0012-06)

AIPAC / dual-loyalty cluster (DB-04)

The pre-candidacy thread generates substantial discussion of politicians’ Israel connections framed as treason:

“Tim Sheehy, major supporter of Israel and was literally in an ad for AIPAC” (DB-04 0014-2, 29 likes)

“the ad was shot in Israel too, cuz why not?” (DB-04 0014-4, 3 likes)

“Remember the times back in the day when traitors were publicly hanged?” (DB-04 0014-5, 4 likes)

The thread escalates from factual claims about AIPAC endorsements through foreign-filming logistics to explicit eliminationist threats against dual-loyalty figures. The progression demonstrates how factual political information becomes the basis for conspiracy attribution and eliminationist endorsement.

Named-target dual-loyalty escalation (DB-02)

The Galloway thread surfaces the corpus’s clearest single instance of dual-loyalty rhetoric escalating into explicit eliminationist endorsement directed at named US elected officials:

“It IS Time — We as American Patriots Demand that Our U.S. Navy & Marine Corp Mutiny for the sake of Our Country and the World and return to the U.S.A. and along with the Army & Airforce Arrest Trump, Dershashits, Spanberger and EVERY Dual Citizen in Congress and this Entire Administration with Israeli passports for Treason as Agents of the Foreign Zionist State... A Patriotic U.S. Military Coup Is Absolutely Necessary and called for Now ASAP for ‘Crimes Against Humanity/War Crimes & Genocide!!’” (DB-02)

Three structural features distinguish this comment from the AIPAC sub-cluster above. First, it names individual elected officials — Dershowitz (rendered “Dershashits” in scatological mockery orthography paralleling the Randy Fine bodily-mockery register), Spanberger, Trump — rather than treating dual-loyalty as a collective attribution. Second, it specifies a mechanism for the eliminationist endorsement: not metaphorical “destruction” but operational military mutiny, mass arrest under “U.S. Military Indefinite Detention,” and a coup framed as patriotic duty. Third, it frames the proposed violence as defensive correction rather than aggression — “American Patriots” recovering the country from “Foreign Spies.” The comment is the cleanest single instance in the corpus of dual-loyalty rhetoric resolving into named-target political-violence endorsement, and the structural significance for the Randy Fine candidacy framing is direct: the “Jewish supremacist” attribution Bilzerian applies to Fine across the corpus operates within the same register, with the same logic of disloyalty as treasonous, and only the proposed remedy varies.

Media-control sub-register (DB-04 and cross-thread)

At DB-04:

“Jewtube will probably ban him again” (DB-04 0005-1)

The “Jewtube” portmanteau performs the media-control claim at the level of platform-specific antisemitism — YouTube recoded as Jewish-controlled infrastructure that suppresses anti-Jewish content.

Israel-controls-the-US sub-register

Multiple comments developing the frame across paragraphs of detail:

“It is the Zionists in control of all our western governments that have promoted the illegal immigration issues we have all been experiencing for the past decade or so.” (DB-02)

The migration-policy extension matters structurally: the same conspiracy framework the AiRT series documented in the geopolitical domain operates here in domestic-policy framings, with the agent attribution constant.

Financial-control sub-register (DB-01)

A multi-comment thread development:

A: “i$rael needs to keep out of U.S. taxpayers pockets. They have gotten over a quarter TRILLION+ dollars from the U.S.”

B: “They got 16 trillion out of us by having us fight Middle Eastern nations for them. They got us into it. Just like they got us into Iran.”

C: “Aren’t they supposed to be good with money, they shoulda been able to double that by now lol”

D: “The amount of money they’ve stolen since 1776 is closer to $100 Trillion... Most of the money that is sent to Isn’tReal gets shipped up into the Rothschild banking regime.” (DB-01)

The thread escalates financial-transfer claims from $250 billion through $16 trillion to $100 trillion across four commenters in a reply chain. The Rothschild-banking-regime extension is the clearest single instance of the classical antisemitic finance-control architecture in the coded threads. It tracks Bilzerian’s Phase 1 supply at DB-01 — “you’ve got people that have no allegiance to the truth, have no allegiance to the country... they all do the same thing. They all fund these foreign countries... they’re all run by these Zionist lobbies.”

Media-control sub-register (DB-06 / Epstein activation)

At TMZ, the conspiracy framework organizes around Harvey Levin’s documented appearance in the Epstein files — a real fact that supplies the activation trigger. The audience slides from individual accusation to structural conspiracy claim:

“I care because the owner of TMZ is in the Epstein files and plays the victim. These people support actual genocide. They are using words as a shield” (DB-06 11-023, 30 likes)

A separate comment compounds three classical antisemitic registers — Mossad, finance-control, and alliance-fracturing directed at Black readers — and reads as a direct continuation of Bilzerian’s DB-01 slave-trade supply:

“the jdl was founded to lynch blacks as a mossad cut outs complaining about jewish monopolies, shady tactics in real estate including mass arson and seeking reparations for synagogues running new York slave sales.” (DB-06 11-001, 67 likes)

The same account (SidewaysBurnouts) operates across the thread as a one-person conspiracy elaboration engine, producing the densest single-comment conspiracy deployment in the sample:

“zionist monopolies, epstein dirty money, charities tied to intelligence and real estate scams like la fire aid, covid fraud and annenberg foundation funnelling money to idf war criminals pushing out san fernando valley middle class for a massive land grab directly tied to 70m embezzled from pacific palecades and pasadena hollywood bowl fund raiser, the head of annenberg foundation was dead mysteriously once joe rogan aired the story. Mirrors the bronx arson epidemic and current jewish scams against black home owners in brooklyn.” (DB-06 11-004, 12 likes)

This single comment compounds six distinct conspiracy sub-registers in one paragraph: finance, intelligence, LA-fires conspiracy, COVID fraud, IDF funding, and Brooklyn gentrification-as-Jewish-scheme. Across the thread as a whole, this account posts five separate comments activating structurally distinct conspiracy registers, with cumulative uptake of approximately 85 likes — the most concentrated single-account conspiracy elaboration in the corpus.

Low-intensity conspiracy normalization (DB-06)

Alongside the elaborated conspiracy architecture, DB-06 produces a parallel register of conspiracy claims delivered as settled fact — no sourcing, no elaboration, no hedge. These are structurally important because they show the framework functioning as background assumption rather than argued position:

“Dan is 100% right. The zionist are destroying America. These TMZ guys are stupid.” (DB-06 00006)

“Two Israeli assets interview Dan” (DB-06 00054)

“Harvey Levin is a jew, his entire family and him donate to pro-israel candidates, they support israel and their war effort against their neighboring countries” (DB-06 00047, 26 likes)

The third example is analytically instructive: it treats Levin’s Jewish identity as inherently explaining his political alignment — not “Levin is biased because he is pro-Israel,” but “Levin is a Jew, therefore he and his family support Israel.” The conspiracy framework operates here not as a claim requiring evidence but as a category-level inference from identity to allegiance.

Otherness and Deceit: Historical Precedent and Character Attribution

Second-densest cluster. Operates principally through the historical-precedent-as-evidence register.

The “109 countries” trope

The expulsion-as-evidence framing organized around the “109 countries” claim:

“It happened 109 times. The endgame will happen after coming of their ‘messiah’ anti-christ” (DB-01)

“So 109 countrys? This time the whole world. Buckle up. Its gonna be a bumpy ride. They are now in a faze where they dont hide the evil anymore. Just like in Germany late 1920s to early 1930s. They feed on beeing hated.” (DB-04 0018)

The DB-04 version escalates the historical-precedent claim into explicit eliminationist framing (”This time the whole world”) and cycles through Holocaust-adjacent periodization (”Just like in Germany late 1920s to early 1930s”). The “they feed on being hated” formulation inverts cause and effect — antisemitism becomes evidence of Jewish pathology rather than prejudice.

Comprehensive otherness attribution (DB-04)

The pre-candidacy thread surfaces the corpus’s most sustained character-attribution cluster:

“Their evil, untrustworthy, backstabbing, racist pieces of crap!!! Their the goyes!! See everything they say about all others, are really about themselves!!! They do everything they accuse others of doing when in fact their the ones telling on themselves while accusing others and trying to act like their all so God obeying so religious but are the evil ones like their father the devil!!! NOW WE SEE WHY JEWS GOT KICKED OUT SO MANY COUNTRIES!!!” (DB-04 0016)

This single comment cycles through moral inversion (”they’re the real racists”), theological demonization (”their father the devil”), religious hypocrisy attribution (”act so God obeying”), and historical precedent as evidence (”WHY JEWS GOT KICKED OUT”). The structural progression from character claims through theological authority to historical precedent represents otherness narratives at their most elaborated.

Pop-culture allegory laundering (DB-01)

A distinct otherness sub-register operates through pop-culture allegory rather than direct attribution, deploying fictional villain figures as analytical equivalents for Jewish historical behavior. The clearest instance is the Game of Thrones deployment at DB-01:

“Because white western Europeans allowed them to come into their homelands to escape persecution, and white western Europeans came to their rescue in WW2. But now white western Europeans are waking up to their back-stabbing subversive deceit. So we are now entering into a new era. It’s like how successful Littlefinger was, in Game of Thrones, right up to the end.” (DB-01)

The structural mechanism matters for the model. Direct character attribution (”evil, untrustworthy, backstabbing”) is recognizable as antisemitism on its surface and triggers content-policy responses on most platforms. Allegorical laundering through a fictional villain — Petyr Baelish, the manipulative outsider whose schemes succeed for years before catastrophic exposure — performs the same attribution while preserving deniability: the surface object of analysis is a fantasy character, and the antisemitic mapping is supplied by the reader. The comment compresses the historical-rescue / current-betrayal arc Bilzerian deploys at Phase 1 (”white western Europeans came to their rescue in WW2... back-stabbing subversive deceit”) into a recognizable narrative template, and the allegorical frame (”right up to the end”) imports the eliminationist resolution of the source material — Littlefinger’s throat is cut on screen — without naming it. This is otherness attribution through deniable narrative scaffolding rather than through direct claim. The register surfaces only at low density across the corpus, but its presence indicates a circulation pattern in which antisemitic content seeks platform-evasion routes through cultural reference rather than only through orthographic evasion (”Ghews,” “Jewtube,” “Isn’tReal”).

Theological delegitimization (cross-thread)

“The synagogue of Satan who say they are Jews and are not, but lie.” (DB-04 0023-02)

The biblical “synagogue of Satan” reference (Revelation 2:9, 3:9) performs theological authority for Jewish delegitimization — contemporary Jews recoded as imposters without divine authorization.

The historical-precedent extension produces an analytically dense soft-Holocaust-relativization move:

“Several countries kicked them out with no problems at all and afterwards they have prospered. Spain in 1700, Bulgaria and many others... as long as they carried on the holocaust alone, no external power ever stepped in to prevent it from happening, but in the long term the expansion of germany was affecting england and russia very negatively so germany had to be stopped; but nobody did that for the jews.” (DB-01)

Linguistic antisemitism (DB-04)

A distinct otherness sub-register deploys German-language antisemitic formulations:

“Juden = Burden” (DB-04 0060)

The German-language deployment performs historical authority — positioning contemporary antisemitism as continuation of a longer intellectual tradition rather than as contemporary prejudice.

Essentialist attribution (DB-01)

A compact otherness sub-register operates through the strike-and-cry-out aphorism, attributed in audience deployment to “an old proverb” and circulating in multiple variants:

“They scream help as they punch you” (DB-01)

“The old proverb it says ‘they cry out in pain as they strike you.’ Times change, but the Berg never does” (DB-01) “

’They cry out in pain as they stab you in the back’ – jews” (DB-01)

The aphorism — a Kevin Strom formulation circulated in far-right discourse as folk wisdom, sometimes laundered through false attribution to Voltaire, the same misattribution mechanism documented above for the “to learn who rules over you” line — attributes a specific behavioral signature (false victimhood deployed as cover for aggression) to Jews as a category. The “Berg never changes” formulation compresses the attribution into essentialist form: “Berg” as metonymic surname for Jews generally, “never changes” as the temporal extension of the attribution across history. The structural move is essentialism — the claim that the behavioral signature is constitutive rather than situational, that contemporary Jewish behavior is identical to historical Jewish behavior, and that the apparent variation across time and context is surface only. This is the otherness register operating at its most distilled: not historical claim, not theological claim, not character attribution applied to specific individuals, but pure category-level essentialism delivered as proverbial wisdom. The orthographic move (”Berg”) performs platform-evasion through surname-fragment substitution, paralleling the “Ghews” / “Jewtube” register documented elsewhere in the corpus.

Direct Nazi-text laundering (DB-06)

The TMZ thread surfaces the corpus’s clearest instance of explicit fascist-source authority — the Goebbels quotation circulating with attribution and being endorsed:

“The Jew is immunized against all dangers: one may call him a scoundrel, parasite, swindler, profiteer, it all runs off him like water off a raincoat.” -Goebbels. (DB-06 11-011, 18 likes)

Reposted nearly verbatim later in the thread (DB-06 11-078) and explicitly endorsed at 11-034 (”based”) and 11-092 (”my daddy”).

Who-is-a-real-Jew sub-register (DB-06)

A delegitimization register specific to DB-06:

“The real jews are us so called black people anyway” (DB-06 11-112)

The Black-Hebrew-Israelite framing — overlapping Khazar-theory and “synagogue of Satan” registers — surfaces in DB-06 specifically; it is not present at the same density in the alternative-media threads. As noted above, Bilzerian’s Phase 1 supply at DB-01 (the slave-trade sub-register) is the structural bridge.

Theological delegitimization (DB-06)

A distinct theological sub-register frames Jews as divinely rejected for deicide and currently engaged in ritual practice that confirms their exclusion:

“God chose the Israelis cause they needed the most help, they were struggling with child sacrifice and idolatry. And he unchose them when they sacrificed his son and denied his son after all the miracles he showed them” (DB-06 11-032, 14 likes)

The same commenter later deploys explicit blood-libel material in a separate reply — “They are still performing sacrifices to this day and ritual killings” (DB-06 11-100) — from a commenter self-identifying as Indigenous. The self-identification is one data point rather than evidence of demographic-level uptake patterns; it is noted because the otherness/blood-libel register surfaces from outside the speaker positions usually associated with these claims in the corpus.

Blame-inversion (DB-06)

A compact deceit/blame sub-register inverts the slur question: if the word “Jew” sounds negative, that is Jews’ own fault:

“If it is a slur, then they probably did something to make it a slur 😂” (DB-06 11-094)

“Well, 🧃 made it a slur because of all the sh*tty things they do.” (DB-06 11-124)

The juice-box emoji evasion in the second example is a platform-evasion orthography specific to TikTok-inflected comment culture.

Victimhood inversion (DB-06)

The victimhood inversion register is thinner than conspiracy or otherness/deceit but produces the sample’s most intellectually compact formulation:

“It’s Schrodinger’s Judaism. It’s both a badge of victimhood and a degrading slur depending on the context.” (DB-06 11-059)

This frames Jewish identity as strategically deployed victimhood — the victimhood inversion register at its most self-aware.

Aggressive Speech Acts and Eliminationist Endorsement

Hitler rehabilitation register (DB-05)

The Jimmy Dore thread contains the corpus’s most prominent Hitler-positive content:

“My spirit will rise from the grave and the world will know that I was right.” -AH (DB-05 0019, 71 likes)

“Yes we owe that man AH a big apology” (DB-05 0019-04, 12 likes)

“HISTER warned us of these demons” (DB-05 0001-46)

“He was absolutely right” (DB-05 0019-08)

The Hitler quotation operates as prophecy fulfillment — contemporary antisemitic conspiracy claims positioned as validation of Nazi predictions. The “apology” formulation explicitly reverses historical judgment.

Weimar-solution framework (DB-05)

“we are living in global weimar rhodesia” (DB-05 0038, 15 likes)

“Weimar problems require Weimar solutions” (DB-05 0038-1)

The Weimar comparison maps contemporary political conditions onto 1920s Germany, with “Weimar solutions” as euphemistic reference to Nazi responses.

Historical-precedent-as-threat (DB-01)

The Shroyer thread surfaces a register where conspiracy claims resolve directly into eliminationist endorsement, framed through historical precedent:

“The only lord the tiny hats have is money. That’s the only thing that’s keeping them from being destroyed. But their time is coming real soon. When the real sleeping giant wakes up your all fkd...” (DB-01)

“you’ll be destroyed and scattered again dw 😊” / “HADRIAN was successful 🙌” (DB-01)

The first comment treats the absence of mass violence as a temporary condition awaiting a “sleeping giant” awakening — the aggressive speech is not threatened but anticipated as inevitable historical correction.

The Hadrian reference invokes the Roman emperor who crushed the Bar Kokhba revolt (132–136 CE), an event that ended in catastrophic Jewish loss of life, the renaming of Judaea to “Syria Palaestina” as deliberate erasure, and the prohibition of Jewish settlement in Jerusalem. In contemporary far-right and accelerationist circles, “Hadrian” functions as shorthand for state-organized eliminationist violence against Jews — celebrated specifically because it combined mass killing with cultural-territorial erasure. The smiley emoji performs the threat as casual aspiration; “was successful 🙌” frames historical genocide approvingly as a model to repeat. The “destroyed and scattered again” formulation explicitly threatens repetition: not metaphor, but operational template.

Nazi-marking endorsement (DB-06)

“we should make them wear a star of david painted on their clothes. i wanna know who they are. they try to blend in, change their names and slither around like snakes” (DB-06 11-116)

This single comment performs aggressive speech acts (the yellow-star call), otherness (the visibility demand), deceit/blame (blending/name-changing), and dehumanization (snake imagery) simultaneously.

Personal eliminationist speech directed at named commenters (DB-06)

The 500-comment sample surfaces a cluster from a single account (anitbodi) arriving late in the thread (April 23) and directing eliminationist speech at individually identified Jewish commenters:

“its too bad you and your family werent at the 2023 nova festival shlomo 😘☠️” (DB-06 11-174, 1 like — directed at @ryanpaul1403)

“you deserve the leo frank treatment traitor 😘☠️” (DB-06 11-176, 1 like — directed at @LtGregoryStevens)

The Nova-massacre wish is repeated verbatim at two targets; the Leo Frank reference is an explicit lynching threat invoking a historical antisemitic murder. These are not directed at “Jews” abstractly but at specific commenters identified as Jewish in the thread. The kiss-and-skull emoji framing performs the threat as affection — the aggressive speech register at its most individually targeted.

Holocaust as punchline (DB-06)

“Dan cooked Harvey worse than his ancestors in Aushwitz ⛽️🤡” (DB-06 00014, 0 likes)

The gas-pump emoji confirms the Holocaust reference is intentional; “cooked” performs the double register of internet slang (”destroyed in argument”) and cremation.

Borat-quotation laundering (DB-03)

“’Throw the Jew down the well...’” (DB-03)

Ironic-register hate speech is coded if the in-context function is endorsement, regardless of the surface deniability of the quotation frame.

Dehumanization: Randy Fine Cluster and Tiny-Hat Cluster

“Fat Jew” and “tiny hat” euphemisms

The Randy Fine “fat Jew” framing concentrates in the alternative-media threads, deployed by audience members rather than recirculated from Bilzerian’s own speech (as the Phase 1 reading establishes, Bilzerian uses the phrase only at TMZ). The “tinyhatism” formulation at DB-03 — a portmanteau treating the kippah as cause of cognitive impairment — operates as compound dehumanization-conspiracy: the bodily-mockery surface (the hat as cause of cognitive impairment) carries the antisemitic euphemism inside it. At DB-06, the “tiny hat” euphemism is applied to Levin directly and carries significant uptake:

“harvey is a slime ball tiny hat” (DB-06 11-115, 36 likes)

“He would win with ease but I don’t think the tiny hats would ever allow him to run.” (DB-06 44-01, 73 likes)

The second example is one of the highest-uptake antisemitic comments outside the 00011 mega-thread (the 3,643-like top comment cluster discussed below) — 73 likes on a conspiracy-fused euphemism that treats Jewish institutional gatekeeping as settled fact.

Raw dehumanization register

The 500-comment sample also surfaces a raw dehumanization register that operates without the conspiracy framework — pure dehumanization contempt without conspiratorial elaboration:

“THATS HOW DIRTY JEWS ARE. EVEN THEIR NAME IS DIRTY” (DB-06 11-157)

“I acquaint the J word with scum, all the time. They’ve earned it” (DB-06 11-211)

“That breed has always been the downfall to many societies!!!!” (DB-06 11-215)

The third example deploys “breed” — animal-register dehumanization — alongside a generic civilizational-decline claim (”downfall to many societies”) that echoes the historical-precedent register without the specific numeric trope. These sit alongside the elaborated conspiracy architecture; they are not competing registers but co-present layers of the same discursive environment.

Israel-Related Antisemitism

Substantial overlap with conspiracy claims. The standalone Israel-related concepts operate principally through eliminationist endorsements directed at the Israeli state:

“The psychopathic genocidal colonial settler terrorist state will need to be defeated militarily like their German cousins were in WW2... GO IRAN!! THE WORLD 🌎 IS COUNTING ON YOU TO FINISH THE JOB THIS TIME!!” (DB-02)

The “German cousins” framing (Israelis-as-Nazis) is a Holocaust-instrumentalization move that codes under Holocaust inversion and Nazi analogy. The “FINISH THE JOB THIS TIME” formulation is the clearest instance in the coded threads of Israel-related-aggressive-speech fusion routed through Holocaust-completion framing. It tracks Bilzerian’s Phase 1 supply at DB-04 — “I would sign up tomorrow and go put boots on the ground... they need to be wiped off the map” — but escalates the eliminationist register into explicit Holocaust-completion language.

The DB-06 Stress Test: Adversarial Host Framing Meets Sympathetic Audience Supply

DB-06 is the corpus’s analytical pivot. Five of the six threads operate under sympathetic-amplifier conditions where host and audience reinforce each other. DB-06 splits them.

On the host side, Levin and Latibeaudiere produce the corpus’s most adversarial Phase 2 framing: they name “fat Jew” as antisemitic, challenge Bilzerian’s disavowal of the term “antisemitism” and his redirection toward “Palestinians-as-real-Semites,” refuse his pivot from his own rhetoric to Fine’s record on Muslims, and tell him on the air that the term anti-Jewish applies.

On the audience side, the comment section arrives sympathetic. Bilzerian himself reported to Shroyer at DB-01 that approximately 95% of cross-platform comments skewed in his favor — a figure consistent with the comment density we observe. The audience composition selecting into a TMZ Bilzerian clip is not a randomly drawn TMZ audience; it is the slice that arrived because Bilzerian was on, drawn through the alternative-media ecosystem (his own channels, the Sneako and Shroyer reactions) rather than by the host framing.

The result is a framing experiment with the host-variable running hard adversarial and the audience-variable running hard sympathetic — pulled against each other inside a single thread, with Phase 1 held constant.

The Phase 3 result is 52% antisemitism density, highest in the coded corpus. By the framing-as-suppressor hypothesis, this is anomalous: adversarial framing should produce lower saturation. It does not. DB-06 saturates at roughly four times the highest mainstream-outlet rate documented in AiRT III and roughly thirteen times the WHCD rate. The audience supply dominated.

The mechanism is visible in the comment data. Audiences integrate the host pushback into the conspiracy framework itself rather than being moved by it. Levin’s documented appearance in the Epstein files — a real fact in the public record — supplies the activation trigger. The comment section reads his on-camera challenge to Bilzerian not as journalistic accountability but as evidence that the network protecting Jewish institutional power is now also protecting Jewish institutional pedophilia: “the owner of TMZ is in the Epstein files and plays the victim... they are using words as a shield” (DB-06 11-023, 30 likes). The “tiny hat” euphemism applied to Levin directly (11-115, 36 likes) makes the framing structural rather than incidental — Levin is not a journalist who happens to be wrong; he is a representative of the institutional apparatus Bilzerian names. The Goebbels quotation circulates through the thread at six identifiable stations: the initial posting with attribution (11-011, 18 likes), a Yiddish-parody endorsement (11-025), an explicit “based” endorsement (11-034, 4 likes), an attempted exculpation claiming Goebbels “was talking about merchants” (11-056), a near-verbatim repost without attribution (11-078, 3 likes), a filial endorsement — “my daddy” (11-092), and a late counter-comment naming it as Nazi propaganda (11-184, 0 likes). The chain demonstrates both the circulation pattern and the structural marginalization of the counter-narrative: the endorsement stations accumulate uptake; the single counter-comment arrives sixteen days after posting and receives no engagement.

This counter-narrative marginalization is visible at the thread level as well. The only sustained factual counter-argument in the entire 500-comment sample appears from a single account (brianmacker1288) arriving on April 24 — sixteen days after the video posted. This commenter line-by-line fact-checks Bilzerian’s claims about the Iran war (11-183), corrects the record on Levin’s Epstein-file involvement (asserting Levin is mentioned once, not as owner, and noting that Bilzerian himself reportedly appears in the files multiple times — 11-179, 11-182), and identifies “fat Jew” as a Nazi-propaganda trope (11-180). The comments are detailed, sourced, and factually specific. Their uptake: 0–1 likes across all posts. By the time the counter-narrative arrives, the conspiracy ecosystem has been running for over two weeks unchallenged. The structural marginalization of the counter-voice is part of the finding.

The thread’s mega-cluster — the 00011 reply chain (228 replies, 3,643 likes on the top-level comment alone) — is the structural demonstration. The top-level comment (”So insane they pretend ‘jew’ is a slur on the same level as the N word”) frames the entire chain as a dispute about linguistic categorization, but the chain rapidly becomes a vehicle for the full conspiracy register: the JDL/Mossad conspiracy thread (11-001, 11-004, 11-005), the Epstein-as-structural thread (11-002, 11-022, 11-023), the slave-trade bridge with fabricated statistics — “Despite being the 0.1% of the world population, 50% of slave owners were jews” (11-010, 11-118, 11-151), the Black-Hebrew-Israelite register (11-014, 11-030, 11-112, 11-142), the Goebbels authority thread (11-011 through 11-092), the eliminationist yellow-star thread (11-116), the Nova-massacre death wishes and Leo Frank lynching threat (11-174, 11-175, 11-176), and the theological blood-libel register (11-032, 11-100). The chain is not a debate about whether “Jew” is a slur. It is a conspiracy ecosystem that uses the slur-equivalence question as its discursive entry point and then saturates across the full category spectrum of the DA Lexicon.

A further structural feature of the DB-06 comment environment is the recoding of the co-host, Charles Latibeaudiere, as a racial subordinate. Multiple comments frame Latibeaudiere’s participation in the interview not as independent journalistic judgment but as racial servitude to Levin: “His boss and co-partner is a Jue” (51-05); “He just wants to keep his job” (51-07); “He’s Harvey’s serf” (51-10); “a jew signs his paycheck” (51-12); “Harvey owns him” (51-14); and at 51-18, the slavery framing is explicit: “he figures it’s better to be in the house than in the fields.” The audience does not treat Latibeaudiere as an independent actor; it treats him as a Black man subordinated to a Jewish boss — reproducing the conspiracy framework’s claim of Jewish institutional control at the level of a specific professional relationship on camera. At 35-01, the recoding is stated as a generalization: “My favorite part is the Black and the Jew saying the white man can’t tell the truth.”

This is the deeper finding. Host adversarial framing presupposes that the audience accepts the host’s standing to push back. When that standing is itself absorbed into the conspiracy framework — when the host is recoded as an interested party rather than a neutral one, and the co-host is recoded as a subordinate rather than an independent voice — the framing loses leverage. The Mamdani study did not face this condition: the influencer hosts in that corpus were trusted by the audiences selecting into them, and adversarial framing therefore had purchase. Bilzerian at TMZ faces an audience that was already on his side before the appearance, and that arrives at the comment section having already coded the host as part of the system Bilzerian is challenging. The framing arrives without standing.

DB-06 defines the boundary condition of framing theory in digital hate ecosystems. The general rule the case establishes can be stated precisely: adversarial framing functions as a suppressor of audience saturation only when the audience grants the framer epistemic legitimacy. When audience-ideological supply has pre-coded the framer as part of the system being challenged, adversarial framing is not merely ineffective — it becomes additional evidence absorbed into the conspiracy framework. This is Principle B in operation. It applies wherever the four conditions hold: a sustained scaffolding speaker, a sympathetic audience pre-loaded against the framer, an adversarial framer, and a framework that includes the framer’s position as a target. The Bilzerian-TMZ instance documents these conditions cleanly; the rule generalizes to any case meeting them.

The implication is that suppression strategies premised on adversarial framing — the dominant assumption behind “calling out” interventions, fact-checking, on-camera pushback, mainstream-media accountability journalism — will fail predictably under these conditions. They will not fail because the framer performs poorly; they will fail because the conditions of legitimacy under which framing operates as suppression have been pre-negated by audience supply.

Other-Target Hate Registers in the Comment Sections

The comment sections do not only amplify the conspiracy framework; they also amplify and extend the other-target hate registers Bilzerian deploys at Phase 1. The DB-06 thread is the densest case, but the registers appear across the corpus.

Anti-Christian Content (DB-05)

Unique to the corpus: DB-05 contains substantial anti-Christian theological content, attacking Christian holidays and practices as pagan impositions:

Extensive attack on Easter and Christmas: “Constantine in 321 A.D. changed it to Easter and the 7th day Sabbath to Sun-day worship... It is paganism. It is easy to spot. It is always the father, the son, and the mother. Isis, Horus, Seb. They just change the names and some of the rites... The satanic panic crap. It’s ridiculous. Teaching people that you’re going to have more in the afterlife and God’s going to take care of it no matter what because you pray about it. No.” (DB-05 0001-54)

“Dan is an inward spiritual Jew Christian prostrating himself to a mythical jew” (DB-05 0001-63)

“every Christian is a spiritual inward Jew worshiping a Jew and the word of Jews as the words of God. That’s the real issue” (DB-05 0001-64)

This register positions Christianity itself as Jewish manipulation — Christians as unwitting servants of Jewish religious authority. The anti-Christian content in DB-05 exceeds the anti-Muslim content in other threads.

Misogyny

The comment-section misogyny clusters around two sub-registers. The Epstein/”little girls” cluster at the top of DB-06 is the densest:

“He said little girls wanted to have tea lol” (DB-06 01-01, 81 likes)

The TMZ source segment does not itself contain any reference to “tea” or to Epstein’s victims; the comment imports extra-corpus material (a recognizable internet-culture euphemism dismissing child sexual abuse, mocking abuse-defense narratives) and frames it as humor. The 81 likes confirm uptake of the dismissive frame. A distinct register at DB-04 extends Bilzerian’s Erica Kirk framing into commenter-supplied variants framing widows, female interlocutors, and women generally as untrustworthy or insufficiently grieving. Anti-feminist micro-registers (”herstory,” 11-026; gendered silencing of female-coded usernames at 11-076 and 11-099) appear at DB-06 in lower density. Across the coded threads, misogyny appears at roughly 10–15 instances per 500-comment thread, with DB-06 at the upper end.

Anti-Muslim and Anti-Arab Racism

Less common in audience comments than misogyny and anti-Black racism, despite the political salience of the Muslim/Israel framing in Bilzerian’s discourse. This is structurally interesting: the audience accepts Bilzerian’s pro-Palestinian positioning and his attacks on Fine’s anti-Muslim record without reproducing the substrate anti-Muslim immigration tropes that run through his Phase 1 discourse (the “military-aged men,” the UK rape framing).

Direct anti-Muslim hatred (DB-04). The pre-candidacy thread surfaces the corpus’s most explicit anti-Muslim content:

“The pedophile of Islam is dead and in Hell begging for a second chance. He died like a bitch.” (DB-04 0027)

This deploys the classical anti-Muslim “Muhammad as pedophile” trope in eliminationist language — the Prophet characterized as suffering eternal punishment. The register operates through sexual defamation of religious authority combined with eliminationist satisfaction.

The “fake Semites” / “real Semites are Arabs” register that appears across the corpus is not anti-Muslim content; it is an antisemitic delegitimization move (otherness/Israel-related) that denies Jewish authenticity by claiming Ashkenazi Jews are interlopers without genuine connection to the region. Muslims and Palestinians are positioned as the legitimate claimants in that framing, not as targets — the register weaponizes pro-Palestinian positioning as a vehicle for antisemitic denial of Jewish peoplehood. The “real Semites” redirection is itself a recurring antisemitic move with roots predating contemporary Israel/Palestine discourse: the term Antisemitismus was coined in 1879 specifically to designate Jew-hatred, never as a category claim about Semitic peoples generally. The exception where actual anti-Muslim audience content does surface is at DB-03, where the comment section reproduces Bilzerian’s extended UK-rape framing in lower-density form.

Anti-Black Racism

Concentrated at DB-06, with the comment section relitigating the N-word equivalence question Bilzerian raised on camera and producing direct anti-Black framings:

“He just loves grape soda” (DB-06 01-06, 42 likes) — classical anti-Black stereotype directed at Levin, who is white; the comment makes structural sense only as anti-Black racial reference deployed in the Bilzerian-defense context

“nobody cares about black fragility. paper skin little girls get upset over words.” (DB-06 11-018) — minimization of Black objection to racial slurs, fused with feminizing/infantilizing register

“I hate the n word I wish black ppl would stop using it too. It just sounds so uncivilized.” (DB-06 11-098) — civilization/savagery binary applied to Black speech

“I say it the same as young punk thug black people say it so it isn’t a slur anymore.” (DB-06 11-172) — anti-Black characterization (”punk thug black people”) fused with N-word normalization

A structurally revealing exchange occurs in the Sneako endorsement chain, where anti-Black and antisemitic registers collide directly:

A: “That should only count as a 3/5ths vote.” (DB-06 38-11, 7 likes — directed at Sneako)

B: “Anything by a Jew like you should count as a 1/8ths vote” (DB-06 38-12, 0 likes — reply)

The 3/5ths-clause reference deploys anti-Black racism against a Muslim convert; the reply deploys antisemitism against the anti-Black commenter. The thread produces both registers in direct exchange — neither suppresses the other.

A distinct anti-Black sub-register reproduces Bilzerian’s Phase 1 music-industry framing through commenter elaboration:

“the same people who own the the prison industrial complex. That’s why hip hop now is about murdering your opposition and hustling drugs, if they could get the black youth to think this to do this type of behavior is cool they could just arrest them and make money off them later in life when they hit prison.” (DB-06 11-120, 3 likes)

This routes anti-Black framings (rap-as-pathology, Black youth as manipulable) through the conspiracy framework (Jewish music-industry and prison-industrial-complex ownership) and reads as a direct audience continuation of Bilzerian’s Phase 1 supply at DB-04.

Homophobia and Anti-Trans Rhetoric

Two registers. Direct homophobic framings at DB-06:

“Harvey is gay. Maybe there were boys on the island?” (DB-06 01-11) — homophobic deflection minimizing child sexual abuse

“I use both words daily, only gay men and women get offended by them” (DB-06 11-128) — gay men positioned alongside women as comparable-hypersensitivity populations

The conspiracy-fused anti-trans register Bilzerian deploys at Phase 1 (”the Talmud has eight different genders”) receives less audience uptake than the antisemitism content alone — likely because the audience is already aligned on the conspiracy framework and does not require the gender-plurality bridge to confirm it.

Ableism

Two direct instances in DB-06 (”retard” at 11-020; “R words” at 11-039) — separate from the antisemitism / racism / misogyny clusters. Worth noting for completeness; not a structurally significant register at the corpus level.

The Multi-Target Stack at DB-06

The structurally significant point about DB-06 is not that any single other-target register is uniquely dense, but that the thread layers conspiracy-saturated antisemitism, anti-Black racism, misogyny, homophobia, and ableism in parallel. A single 500-comment thread contains all five registers with substantial uptake on multiple. This is what hate-saturated comment ecosystems look like at their densest — not specialization in one target, but parallel availability of multiple registers from which commenters draw.

Comparison with Prior Studies in the Research Line

The Bilzerian corpus extends the architecture documented across this research line. It sits structurally apart from the political-violence research line in two respects. First, the trigger event is not a discrete act of violence but a candidacy announcement; the comment-section response is a sustained engagement with an actor whose discursive output the audience has been processing for years. Second, the analytical object includes the elite speaker’s primary discourse as a coded stratum, not only the comment-section response. Cross-study quantitative comparisons should be read with these structural differences in mind. What does travel cleanly across studies is the analytical framework — the Decoding Antisemitism Lexicon, the implicit/explicit register distinction, the architecture/scaffolding mechanism developed across the AiRT series and clarified in the WHCD finding.

Versus the Mamdani study (January 2026). The Mamdani study held Phase 1 ambiguous and varied Phase 2 framing across nine long-form influencer videos covering the same source material — Mamdani’s own public speech on Israel, Zionism, and antisemitism, which was interpretively contested rather than overtly antisemitic. The headline finding: conspiracy-oriented influencer framing generated approximately eight times more antisemitic audience responses than empathy-oriented framing of the same source material. Phase 2 framing did the analytical work. Phase 1 was the constant.

The Bilzerian corpus inverts that design. Phase 1 is held maximally non-ambiguous — the speaker himself supplies the “Jewish supremacy” framework, the Disloyalty attribution, the financial-control register, and the eliminationist endorsement. Phase 2 varies along the sympathetic-to-partial-adversarial spectrum. The question the design tests is whether intermediary framing retains its suppressive power when Phase 1 is no longer ambiguous.

DB-06 is the cleanest test. Levin and Latibeaudiere push back on Bilzerian’s framings on camera, name “fat Jew” as antisemitic, name his disavowal of the term “antisemitism” as evasive, and refuse Bilzerian’s attempt to redirect to Fine’s anti-Muslim record. By the host-framing measure, DB-06 is the corpus’s most adversarial Phase 2 condition. The audience supply, however, runs in the other direction: 95% of cross-platform comment endorsement skewed pro-Bilzerian (his own claim, repeated in DB-01 to Shroyer; consistent with the comment density we observe). The two conditions — adversarial host framing, sympathetic audience supply — pull in opposite directions.

The Phase 3 result is 52% antisemitism density, highest in the corpus. The audience reframes the host pushback itself as conspiracy evidence: Levin’s documented appearance in the Epstein files supplies the activation trigger (”the owner of TMZ is in the Epstein files and plays the victim... they are using words as a shield”), and the conspiracy framework absorbs the adversarial framing rather than being suppressed by it.

Read together, Mamdani and Bilzerian specify the framing-effect mechanism. Phase 2 framing operates as suppressor under two conditions: Phase 1 is interpretively ambiguous, and the audience does not arrive with a pre-loaded conspiracy reservoir. Mamdani satisfied both conditions and the framing variation produced the eightfold rate spread. Bilzerian fails both conditions, and DB-06 demonstrates what happens when host pushback meets pre-saturated audience supply: the framing variation produces no suppression, and the audience integrates the host’s pushback into the conspiracy framework as further evidence of Jewish institutional control of media.

This is not a refutation of the Mamdani finding. It is its boundary condition. Phase 2 framing is powerful where Phase 1 leaves room for interpretation and the audience reservoir is not already fixed. Where the speaker himself fills the interpretive room, Phase 2 framing loses its leverage.

Versus AiRT III and Charlie Kirk. AiRT III documented conspiracy-saturated commentary at outlets covering the Iran-Israel-US campaign, with antisemitism rates in the 12–20% range across mainstream outlets. The Bilzerian baseline (40.8% corpus average across six coded threads, with the lowest single-thread rate at 23.6% still exceeding the AiRT III ceiling) substantially exceeds these rates. Where AiRT III tracked event-conditioned reactivation of the conspiracy framework, Bilzerian shows speaker-conditioned saturation. The Charlie Kirk assassination study aggregated to ~30% blame and ~6.5% conspiracy across nine matched outlets, with antisemitism rate at ~3.6%. Bilzerian inverts these proportions: antisemitism-coded conspiracy is the dominant register rather than a thin sub-stratum of partisan blame.

Versus WHCD. The WHCD finding was the inverse: conspiracy architecture decoupled from antisemitic scaffolding, with antisemitism rates at 0.5–2% across nine of ten outlets and elevated only at TYT (4%) where audience-ideological supply remained available — the 0.5–4% range shown in the comparison table reflects this distribution. Bilzerian is the corresponding case in which the speaker himself supplies both architecture and scaffolding. WHCD demonstrates that conspiracy can operate without antisemitism when neither route opens; Bilzerian demonstrates that antisemitism saturates conspiracy when the speaker himself is the route. DB-06 strengthens this paired reading: even when host framing attempts to interrupt the speaker-route at the Phase 2 level, the audience-supply route remains active and dominates the Phase 3 outcome.

Mapping the studies onto the model. The four-input Conspiracy Architecture Model presented in the Introduction is populated by these prior studies in distinct ways. The WHCD article established inputs 1 and 2 (event-context supply and audience-ideological supply) and demonstrated their decoupling: when neither is available, the conspiracy architecture can operate without antisemitism. The Mamdani study tested input 4 (intermediary framing strategy) under conditions of low speaker supply, producing the eightfold rate spread that established framing’s suppressive power when other inputs are weak. The Bilzerian corpus contributes input 3 (speaker-supplied scaffolding) and stress-tests input 4 against high audience supply at DB-06. Read together, the studies specify when each input dominates and when each loses leverage — most importantly, that intermediary framing operates as suppressor only when audience-ideological supply is also low. When speaker-supplied scaffolding is saturated and audience-ideological supply is available, host-adversarial framing fails to suppress and the audience reframes the host pushback itself as conspiracy evidence.

Conclusion

The Bilzerian corpus is one well-documented instance of a configuration that event-conditioned monitoring is not designed to capture: antisemitic discourse operating as steady-state production of a sustained ideological supplier, organized through a meta-frame that subordinates adjacent hate registers, and not reliably modulated by adversarial host framing once audience legitimacy conditions are pre-negated. The conspiracy architecture is the dominant register of this comment ecosystem rather than a variable register that reactivates under specific conditions. The candidacy is its institutional crossing point.

The Conspiracy Architecture Model’s three governing principles specify what event-conditioned analysis underdescribes in cases of this configuration, and why.

Principle A (speaker substitution) — surveillance scope mismatch. Pipelines calibrated to event-conditioned spikes are designed for one analytical job and do that job well; speaker-conditioned saturation falls outside their detection scope. The instrument is not wrong; it is calibrated for a different signal. The Bilzerian rates documented across the corpus do not register on event-anchored monitors because there is no event to anchor to — the saturation is continuous, sustained by speaker supply rather than triggered by external pretext. The persistence of saturation across radically different audience compositions (sympathetic-right, sympathetic-left anti-imperialist, sympathetic-left commentary) is the demonstration: speaker-conditioned saturation does not require any particular host or audience type; it requires a sustained ideological supplier. The Jimmy Dore result is particularly instructive: sympathetic-host amplification produces antisemitism density and Hitler rehabilitation content that match or exceed explicitly far-right outlets, indicating that alternative-media sympathetic framing is not a moderating condition under sustained speaker supply.

Principle B (framing dependence on audience legitimacy) — institutional gap. The candidacy converts what had been platform-content-policy territory into electoral-speech territory. The institutional infrastructure for addressing that crossing — campaign-finance disclosure, FEC content rules, congressional ethics frameworks, the ordinary practices of political journalism — operates on the framing-as-suppressor assumption that the TMZ case falsifies. On-camera adversarial pushback from professional hosts does not suppress audience saturation when audience supply has pre-negated host legitimacy; it is absorbed into the conspiracy framework. Suppression strategies premised on adversarial framing — “calling out,” fact-checking, on-camera pushback, mainstream-media accountability — will fail predictably under these conditions.

Principle C (hierarchical multi-target organization) — analytical scope. Frameworks that treat antisemitism as a discrete object will under-describe what is happening in these threads. Multi-target hate is not co-occurrence; it is hierarchical organization under the antisemitic conspiracy meta-frame. Monitoring systems that track hate categories separately measure the parallel layers and miss the architecture; intervention against any single register without addressing the meta-frame leaves the system intact.

The three principles together establish that the dominant analytical apparatus of digital hate monitoring, while productive for event-conditioned discourse, is not designed for the speaker-driven configuration the Bilzerian corpus exemplifies. The methodological infrastructure to track this configuration exists. The analytical assumptions under which it currently operates are calibrated for a different mode. Extending the paradigm — adding speaker-conditioned analysis alongside event-conditioned analysis, treating the two as complementary rather than competing — is the precondition for institutional response.

What does that look like operationally? Three reorientations follow directly from the three principles. Monitoring systems calibrated to event-conditioned spikes need to add continuous-supply detection: speaker-level baseline tracking that does not require an external trigger to register the saturation. Adversarial host framing — “calling out,” fact-checking, on-camera pushback — should not be auto-coded as mitigation; audience response under conditions of pre-negated host legitimacy must be examined for inversion, not assumed to indicate suppression. And hate categories cannot be tracked in parallel as if they were independent dimensions; the analytical question is whether and how the conspiracy meta-frame is organizing them, because intervention against any single register without addressing the meta-frame leaves the architecture intact. None of these reorientations require new methodological infrastructure. They require the infrastructure to be deployed under different theoretical assumptions about what it is supposed to detect.

This is what mainstreaming looks like at the bottom of the iceberg — not the visible top, where commentators occasionally surface antisemitic content into the news cycle, but the sediment layer, where the discourse has settled, accumulated, and now sits as foundation rather than froth. And it is what mainstreaming looks like at its operational stage: a federal congressional candidacy that combines manosphere-cultivated audience reach, horseshoe-positioned anti-establishment platforming, and openly eliminationist discursive content into a single electoral vehicle. The streams have already merged. The candidacy is the merger’s institutional form. Detection requires recognizing the pattern as such — and the analytical apparatus to recognize it must be built alongside the existing event-conditioned apparatus, not in opposition to it.

Limitations

Generalizability of the speaker-driven mode. The article does not claim that the speaker-driven mode is dominant across all digital environments, nor that the event-response paradigm fails everywhere. The claim is bounded: where high-reach primary speakers operate continuously and audience-ideological supply is already in place, speaker-conditioned saturation becomes the governing condition of discourse production. The Bilzerian corpus documents one instance of this configuration. How widespread the configuration is across the broader digital environment, and what proportion of digital antisemitism it accounts for relative to event-conditioned production, are open empirical questions that this article does not resolve.

Corpus scope. The article does not test whether fully adversarial-mainstream Phase 2 framing — adversarial at both the host and the audience-composition level — can suppress speaker-conditioned saturation. DB-06 tests only the host-adversarial / audience-sympathetic configuration. The fully-adversarial condition requires a different corpus design.

Coding status. Phase 1 is coded against all six transcripts. Phase 3 quantitative results cover all six threads: DB-01 (Shroyer), DB-02 (Galloway), DB-03 (Sneako post-candidacy), DB-04 (Sneako “Jewish Supremacy”), DB-05 (Jimmy Dore), and DB-06 (TMZ).

Other-target hate coding. The misogyny, anti-Muslim, anti-Black, and homophobia/anti-trans registers are coded at the discourse-analytic level without imposing the DA Lexicon framework, which is calibrated for antisemitism specifically. Future work in this research line will benefit from comparable target-specific lexicons for parallel coding.

Methodological notes. A methods companion to this article documents dual-coding conventions, platform-evasion-orthography handling, ironic-register coding decisions for the Borat-quotation and similar laundering moves, and inter-coder reliability outcomes from the calibration sessions. The companion is not necessary for following the analytical claims here; it supports replication and methodological scrutiny.

Sampling and representativeness. The sampling strategy — first 500 comments per thread — captures early, visible participation rather than the full distribution of engagement, and may overrepresent highly active or ideologically aligned users. Silent audiences (viewers who do not comment) are unobserved. Comments removed by platform moderation before data collection are not captured. The 500-comment sample mitigates but does not eliminate density-estimation limitations inherent to early-comment sampling.

AddressHate Research Scholar at NYU’s Center for the Study of Antisemitism | Lead, Decoding Antisemitism | Research Advisor, AddressHate | Editor-in-Chief, Digital Hate Review

Executive Summary

Within an hour of the April 25 shooting at the White House Correspondents’ Dinner (WHCD) — the third assassination attempt against Donald Trump in under two years — the comment sections of every major US news outlet had already arrived at competing explanations: a false flag staged by Trump himself, a Mossad operation, a Deep State distraction from the Epstein files, a left-wing terror plot. All were already circulating at scale.

This study analyzes 2,000 YouTube comments collected within hours of the event from ten major US news outlets across the political spectrum. Its headline finding concerns a measurable shift in how online publics process political violence in real time. In a comparable analysis of the September 2025 killing of Charlie Kirk — the closest reference point with a similar incident in our research line — roughly three in ten comments performed conventional blame attribution: it was the Left’s fault, the Right’s fault, the media’s fault, Kirk’s own rhetoric. In the WHCD comments, that figure collapses to one in twenty. In the same comparison, conspiracy theories — false flag claims, staged-event narratives, claims that Trump himself or the security state orchestrated the shooting — jump from a marginal one in thirty to roughly one in four.

What is happening here is not the disappearance of blame but its migration. Where in earlier corpora commenters debated who had created the violent climate that produced an attack — the Left, the Right, the media, the target’s own rhetoric — at the WHCD they overwhelmingly skip past that question and proceed directly to a different one: who manufactured this event in the first place. The orchestrator slot in the comments remains fully occupied; the question being answered has shifted from causality to authorship.

A second finding cuts more sharply against expectation, given the wider geopolitical backdrop. The WHCD shooting unfolds against the still-recent US–Israeli campaign against Iran, the discursive context analyzed in Antisemitism in Real Time, Part III, which documented a saturated landscape of antisemitic conspiracy frames — Israel-controls-the-US, Mossad-orchestrates-American-foreign-policy, Trump-as-Israeli-asset — running through mainstream YouTube comment sections. One might expect that infrastructure to reactivate around an attempt on Trump’s life only weeks later. It does not. Antisemitic conspiracy theories drop sharply at the WHCD shooting, falling by roughly two-thirds compared to the Kirk corpus and to a small fraction of what AiRT III documented at the height of the Iran coverage. The architecture of conspiracy still operates — a hidden orchestrator, a manufactured event, a perpetrator framed as a patsy — but the orchestrator slot is now filled by Trump and the US security state rather than by Jews, Israel, or Mossad. Antisemitism reattaches to the conspiracy architecture only at outlets whose audiences already carry such frames as a default register, most clearly at The Young Turks, where Mossad-orchestration claims and classical antisemitic tropes resurface despite the absence of any geopolitical pretext for them in the WHCD event itself.

A third finding concerns tone. With the shooter caught alive and the target unharmed, explicit threats collapse across most of the corpus, persisting at meaningful density only on Fox News (4.5 percent), where partisan blame and partisan threat travel together as in earlier corpora. What replaces threats elsewhere is a register we call regret-the-miss — comments lamenting the failure of the attempt rather than calling for new violence. “Damn They Miss Again,” “Trump can’t get got soon enough,” “unfortunately Trump is still alive,” “Why do they always miss,” “We need Tyler Robinson” (the alleged shooter of Charlie Kirk, invoked as a model). The endorsement of political violence does not disappear; it shifts from prospective threat to retrospective approval.

The implications matter for platform governance, content moderation, and political-violence research. Conspiracy narratives about events of political violence are not driven solely by ideology; they are also driven by the structural shape of the events themselves. The same narrative scaffolding — a manufactured event, a hidden orchestrator, an expendable patsy — recombines around whichever adversarial actor is most rhetorically available in a given case. When that actor is Jewish or Israeli, the result is antisemitic conspiracy. When it is the US security state or the sitting president, the surface looks different but the underlying narrative operates by the same logic. The infrastructure for processing political violence as something staged rather than something real is now portable across ideological lines, and it travels with the audience as much as with the event.

In one sentence: conspiracy theories have become the default way many online publics process political violence, and antisemitism now appears mainly where either the event or the audience invites it.

The methods to track these shifts in real time — interdisciplinary discourse analysis combined with LLM-assisted annotation — are now mature enough to do this work as it happens, not only after the fact. What is still missing is the institutional capacity to act on this analytical visibility at the speed at which these narratives now spread.

Introduction

Within an hour of the shooting at the White House Correspondents’ Dinner (WHCD) on Saturday night — the third assassination attempt against Donald Trump in under two years — the comment sections of every major US news outlet had already arrived at competing explanations: a false flag staged by the Trump administration, a Mossad operation, a Deep State distraction from the Epstein files, a karmic return on Trump’s rhetoric, a left-wing terror plot, an inside job by the press itself. The shooter — later identified as Cole Tomas Allen, a 31-year-old tutor and indie game developer from Torrance, California — had been in custody for less than sixty minutes. None of these explanations were supported by available facts. All of them were already circulating at scale.

This is the fifth major event of political or antisemitic violence this research line has tracked since May 2025 — the Capital Jewish Museum shooting, the Charlie Kirk assassination, the Bondi Beach massacre, the Temple Israel synagogue attack, and now an attempt on a sitting US president at the symbolic apex of American press freedom. Before the dinner, five hundred journalists had signed a petition urging the WHCA to oppose Trump’s “efforts to trample freedom of the press”; the press–executive relationship was already in a contested state when the gunfire started. The intervals are not lengthening. They are shortening. And the digital response is not catching up to events; it is racing ahead of them, filling the interpretive vacuum left by absent facts with speculation, projection, and conspiracy theories at industrial speed.

What links these events analytically is not a single causal mechanism but a recurring pattern in their digital aftermath: the rapid online normalization of violence through blame projection, conspiracy theories, and ideological repertoire convergence in the immediate post-event window. The WHCD shooting is the first major domestic political-violence event of the post-Iran-campaign period — the campaign analyzed in Antisemitism in Real Time, Part III — and the question this article asks is how the discursive infrastructure consolidated during that campaign now absorbs and reorganizes around a foiled attempt against the president himself.

The basic facts of the incident are these. According to initial law enforcement statements and media reporting, Allen rushed a security checkpoint outside the ballroom at the Washington Hilton, armed with a shotgun, a handgun, and multiple knives. Reports indicate gunfire was exchanged with Secret Service agents, with one officer struck in his bullet-resistant vest, before Allen was tackled and arrested. President Trump, First Lady Melania Trump, Vice President JD Vance, and members of the Cabinet were rushed from the room. Two sources told CBS News that Allen told law enforcement after his arrest he wanted to shoot Trump administration officials. Some details may be subject to revision as the investigation proceeds.

This study examines the discursive consequences of the attempt: how the event was framed, contested, and appropriated across digital publics in the immediate window before stable narratives had time to form. At its center is an analysis of spontaneous user reactions in YouTube comment sections beneath ten widely circulated clips from leading US media outlets, spanning the political spectrum from left to right. By analyzing 2,000 comments collected within hours of the event, the study captures the immediate responses before the slower-developing details of the case — Allen’s biography, his $25 donation to the Harris campaign, his Caltech background, his stated motive — had circulated widely. This is, in other words, a snapshot of the interpretive vacuum: the period in which conspiratorial speculation, blame projection, and justificatory reasoning operate at maximum velocity precisely because facts are scarce. The vacuum produces this acceleration through a specific structural conjunction — the absence of authoritative reporting, the high emotional salience of the triggering event, and the algorithmic amplification of speculation under conditions of low informational constraint. Social media in this window does not simply react to the event; it produces its first stable interpretations, and those interpretations are the analytical object of this article.

The findings should be read in continuity with prior installments in this research line. The Kirk study documented how blame projections, affirmations of violence, conspiracy narratives, and explicit threats formed a chain of discursive escalation in mainstream YouTube comment sections. The Bondi Beach analysis showed that even under conditions strongly conducive to empathy, antisemitic and adjacent illiberal repertoires remained visible. The Temple Israel analysis demonstrated that foiled attacks suppress explicit Death Wishes but not the underlying discursive infrastructure: conspiracy narratives, refusal of empathy, and threats reorganize themselves around the new outcome structure. This research line — now operating under the Decoding Hate framework at NYU’s Center for the Study of Antisemitism — applies a shared annotation protocol and analytical architecture across events, enabling the kind of cross-event comparison this article performs against the Kirk corpus. The WHCD case offers a further empirical test of that pattern, in a setting whose distinctive features — a foiled attack at a press-freedom event, a sitting president as the apparent target, a perpetrator alive and in custody — condition a discursive response that is structurally distinct from prior cases in the series.

The Shooter: Cole Tomas Allen

Cole Tomas Allen, 31, lived in Torrance, California, a coastal city of approximately 140,000 in Los Angeles County. The factual profile that emerged within the first 24 hours of reporting complicated easy explanations.

Allen graduated from Pacific Lutheran High School in Gardena and from the California Institute of Technology in 2017 with a bachelor’s degree in mechanical engineering, completing a master’s degree in computer science at California State University-Dominguez Hills in 2025. As a Caltech student, he was featured in a 2017 local news report for developing a prototype emergency brake for wheelchairs and was a member of the campus Christian fellowship. He worked part-time at C2 Education, a tutoring company, where he was named “Teacher of the Month” at the Torrance branch in December 2024. Students he tutored described him as “very intelligent” and “on the nicer, quiet side,” according to Dylan Wakayama of the Asian American Civic Trust, speaking to the Los Angeles Times. A former high school teammate told NBC News he was “probably the most gentle person on the team.” Allen was also an independent video game developer who published an indie game titled Bohrdom on Steam in 2018 and was working on a second game described as a “top-down shooter” combat game set in outer space.

His political profile is thin. Federal Election Commission records show a $25 donation to a Democratic political action committee supporting Kamala Harris’s presidential campaign in October 2024; California voter registration records list him as “no party preference.”

According to law enforcement statements and reporting, Allen traveled from Los Angeles to Chicago and then on to Washington, D.C., by train, checking into the Washington Hilton as a guest a day or two before the dinner. The two firearms he carried — a shotgun and a handgun — had been purchased “within the last couple of years,” acting Attorney General Todd Blanche told NBC News, indicating premeditation rather than a sudden acquisition. He charged the magnetometer screening area outside the ballroom at 8:36 p.m. and exchanged gunfire with law enforcement before being tackled and handcuffed. One Secret Service officer was struck in his bullet-resistant vest; he was treated and released from the hospital. Allen was charged with using a firearm during a crime of violence and assault on a federal officer using a dangerous weapon; U.S. Attorney for the District of Columbia Jeanine Pirro indicated additional charges were expected. Initial law enforcement assessment indicates Allen acted alone.

What makes Allen’s profile analytically significant is what it did not contain at the time the comment corpus was collected on April 25–26. As of that window, there was no publicly known manifesto, no sustained extremist online presence, no clear ideological signature comparable to the bullet-casing inscriptions recovered in the Kirk case. There was a brief biographical sketch, a stated motive (target Trump administration officials), and an open question. The interpretive vacuum this produced is precisely what the comment corpus fills — and the question this article asks is how.

A note on post-corpus disclosures. As this piece went to press on April 26, additional reporting indicated that investigators had recovered written material from Allen’s hotel room and pre-attack messages to relatives denouncing the Trump administration. President Trump, speaking on Fox News, publicly characterized this material as a “manifesto.” Acting Attorney General Todd Blanche, speaking to NBC, used the more cautious term “writings” and noted that a preliminary review suggested administration officials, “likely including the president,” had been targets. As of the time of writing, no manifesto has been publicly released; the existence and content of the document described by the President have not been independently verified, and what has been characterized as a “manifesto” remains, at this stage, the President’s framing of evidence still being processed by investigators. These post-corpus disclosures do not alter the analytical findings of this study, which concern how commenters interpreted the event in the absence of such information during the first 24 hours. If anything, they underline the article’s framing: the conspiratorial saturation documented below was generated before any of this material became public, and is therefore independent of whatever specific motive content the writings may ultimately be reported to contain.

Dataset and Method

This study draws on ten widely circulated YouTube clips from leading US media outlets covering the WHCD shooting in real time. Following the protocol established across this research line, threads were selected on the basis of highest engagement: for each outlet, the YouTube clip generating the largest comment volume within the first 24 hours after the event was identified and used as the data source. Comments were collected within hours of the event on April 25–26, 2026, capturing the immediate post-incident window before stable narratives, biographical details about the suspect, or follow-up reporting had time to circulate widely. From each video, the first 200 top-level comments appearing chronologically (sorted by “Newest first”) were analyzed, with reply chains preserved for sequence and relational analysis where relevant; spam and duplicates were excluded. This produced a total corpus of 2,000 comments. The full corpus has been coded using the framework set out below, enabling direct comparison with the Kirk, Antisemitism in Real Time, and Temple Israel datasets.

The corpus includes:

• Left/Progressive: CNN, MS NOW (formerly MSNBC), The Young Turks (TYT)

• Center: ABC News, NBC News, CBS News, Associated Press (AP), USA Today

• Right/Conservative: Fox News, Forbes Breaking News

Audience composition varies significantly across these channels; the analysis below reports comment-level patterns rather than audience-level representation, and outlet-level comparisons should be read in that light.

The dataset architecture follows the protocol established in the Kirk study (September 2025), the Antisemitism in Real Time series (June 2025–March 2026), and the Temple Israel analysis (March 2026). All studies in this immediate-aftermath research line apply the same annotation protocol and counting unit, enabling cross-study comparison. The total sample of 2,000 comments sits between the Kirk corpus (2,400 comments across twelve outlets) and the AiRT III and Temple Israel corpora (1,600 comments across eight outlets).

Analysis followed the principles of qualitative content analysis, carried out in MAXQDA, with native-speaker coders applying linguistic and multimodal close reading supplemented by sequence and relational analysis where comments form part of reply chains. Coding combined deductive categories — derived from the Decoding Antisemitism Lexicon (Becker et al., Springer Nature, 2024) and from the analytical framework developed across the prior studies in this research line — with inductive coding of formations specific to the WHCD case. Inter-coder reliability was addressed through iterative calibration sessions on borderline cases, with attention to coded references, ironic framing, and emoji-based signaling that evades surface-level keyword detection. As a preparatory step, user comments were pre-processed using Anthropic’s Claude Opus 4.7 model to support cleaning, deduplication, and initial structuring of the dataset, as well as the pre-identification of overt, explicit forms of verbal violence and conspiracy narratives, prior to manual coding; all category assignments and qualitative readings reported below are the work of human coders. As LLM-assisted annotation continues to mature for ironic, coded, and multimodal content, the analytical pipeline running from real-time corpus capture through structured coding to actionable diagnostic output stands to compress significantly — turning the kind of cross-outlet, cross-event analysis presented here from a retrospective study into something closer to a real-time monitoring instrument.

Analytical Categories

The classification system used for this study adapts the Kirk framework to the distinctive features of the WHCD case — the press-freedom venue, the foiled-attack outcome structure, and the post-Iran-campaign discursive environment. The schema comprises five categories:

1. Conspiracy Theories. Two distinct sub-categories: (a) generic conspiracy theories including false flag claims, Deep State/FBI orchestration, distraction narratives, fascist-templating (Reichstag-fire-style pretext claims), and elite manipulation; (b) explicitly antisemitic conspiracy theories drawing on the Power, Subordination, and Trumpstein-thread infrastructure documented in AiRT Part III. These sub-categories frequently overlap, with transitional forms in which generic elite-orchestration claims slide into antisemitic attribution through coded references, dog whistles, or named actors.

2. Schadenfreude, Sarcasm, and Ridicule. Mockery, irony, and dark wordplay directed at the event, the venue, the response, or the principal figures present. The Temple Israel finding suggests that foiled-attack outcome structures shift the affective register away from explicit Death Wishes toward this category. Includes ridicule of the WHCD ritual itself and dehumanizing characterological reduction (mocking nicknames, bodily mockery).

3. Affirmation, Justification, and Glorification of Violence. Comments treating the attack as inevitable, deserved, morally justified, or actively celebratory, including explicit refusal of empathy and moral inversion of victim/perpetrator roles. Following the Kirk distinction: affirmation casts the violence as predictable consequence; justification frames it as morally legitimate response; glorification celebrates the act outright or laments its failure. The foiled-attack outcome conditions a distinctive sub-register — regret-the-miss formulations — that endorse the attempt and lament its failure rather than celebrating a death that occurred.

4. Discursive Blame Projections. Attributions of responsibility for the attack across multiple potential targets — the political Left, the political Right, the media, and Trump himself. The Kirk study found that blame projections operated as the entry point of the discursive escalation chain. The specific distribution of sub-targets in the WHCD corpus, and its relationship to conspiracy displacement, is reported in the empirical chapter.

5. Threats and Calls for Retaliation. Speech acts that move from interpretation to projection, articulating explicit threats, calls for further violence, or fantasies of mass political conflict. Sub-motifs include targeted threats, reciprocal threats, civil-war and purge fantasies, and threats fused with conspiratorial framings.

In the empirical chapter that follows, categories are presented in order of empirical density at WHCD, which inverts the ordering used in the Kirk study (where Blame led and Conspiracy sat third). The reordering reflects the central finding of this corpus and is taken up structurally in the comparison with the Kirk findings.

A Note on Cross-Study Comparability

All studies in this research line — Capital Jewish Museum (May 2025), Kirk (September 2025), AiRT I–III (June 2025–March 2026), Bondi Beach (December 2025), Temple Israel (March 2026) — apply the same annotation protocol, counting unit, and core analytical framework. Cross-study directional observations are robust. Aggregate antisemitic content rates and category distributions should nonetheless be interpreted with awareness of event-type conditioning: the structure of the discourse event — geopolitical military escalation, mass-casualty terrorism, foiled domestic attack, attempt on a head of state — shapes which repertoires are available and plausible for commenters to deploy, and therefore conditions both the volume and the register of response.

The WHCD case sits at a structural crossroads in this typology. Like the Kirk case, it is a domestic political-violence event involving a high-profile US figure. Like the Temple Israel case, it is a foiled attempt with the perpetrator alive and in custody. Distinctively, the apparent target is the sitting president, the venue is a press-freedom event, and the corpus was collected during the interpretive vacuum before the suspect’s biography had circulated. These structural features are tracked through the analysis below.

Results

Quantitative Overview

Three patterns emerge from the rate-by-rate distribution across the ten outlets, presented below in left-to-center-to-right order — the ordering convention used in the Kirk study.

(All values in percent; n = 200 per outlet.)

First, the discursive-harm rate climbs from ~19.5% at Fox and ~23% at Forbes through the centrist outlets (~32–36% at ABC, NBC, AP, USA Today), reaches ~42% at CBS, and peaks at ~47% at MS NOW and TYT. The gradient broadly tracks editorial position from right to progressive, but CBS’s centrist–high outlier shows the relationship is not strictly editorial: outlet-level audience disposition and the kinds of comments a thread invites both shape the rate. Two effects compound across the gradient: conspiracy density rises sharply from 8% at Fox to 39% at MS NOW, while blame density falls from 9% at Fox to ~4–5% across the rest. Fox’s editorial framing tracks Trump’s political position most closely, and Fox commenters accordingly process the event through partisan-blame and counter-conspiracy registers that the other outlets’ commenter bases do not.

Second, conspiracy is the load-bearing category across nine of the ten outlets — only at Fox does it sit below 10%. The structural implication is that the staged-event frame, which in prior work in this research line operated primarily as antisemitic conspiracy directed at Israel/Jews (see AiRT Part III, Temple Israel), here operates as partisan conspiracy directed at Trump and the US administration. The orchestrator slot is preserved; the agent attribution moves.

Third, the antisemitism rate is a separate matter. It stays at or near 0.5–2% across nine of the ten outlets and jumps only at TYT, where it reaches 4.0% — four times the centrist baseline. TYT is the corpus’s highest-AS outlet but not the corpus’s highest-conspiracy outlet (that distinction sits with MS NOW at 39%, with TYT at 37% and CBS at 33% close behind). The MS NOW finding is structurally important: at the corpus’s joint conspiracy-density high and joint discursive-harm high, antisemitism returns to near-zero. Same conspiratorial saturation as TYT, opposite antisemitism density. The variable that distinguishes them is not event context — both outlets covered the same shooting — but audience-ideological reservoir. TYT’s audience carries Mossad-orchestration, Israel-foreign-control, and Rothschild frames; MS NOW’s progressive-centrist audience does not. The conspiracy architecture saturates at both outlets; the antisemitic scaffolding reattaches only where the audience supplies the reservoir to support it.

MS NOW functions as a control case: high conspiracy without antisemitism. TYT shows the same saturation with antisemitism reattached. The contrast sharpens a claim from prior work in this research line: antisemitic scaffolding reattaches to conspiracy through two independent routes — event-context supply (the event itself supplies a Jewish-control geopolitical pretext) and audience-ideological supply (the audience already disposed to route conspiracy through Israel/Jewish-power frames carries the architecture in regardless of event context). Either route alone is sufficient. The absence of both produces conspiracy without antisemitism even at the corpus’s highest conspiratorial saturation. USA Today (30% conspiracy, 0.5% AS) and CBS (33% conspiracy, 1.5% AS) reinforce the finding from third and fourth positions: when neither route is open, the conspiratorial architecture remains active but the antisemitic scaffolding does not reattach. The two rates are conditionally independent and decouple cleanly when only the architecture is in play.

Conspiracy Theories

Non-Antisemitic Conspiracy Theories

Compressed staging assertions

Across nine of the ten outlets, single-word and short-phrase staging assertions function as the corpus’s most legible compressed register. The form is consistent: a one-to-five-word assertion that the event was manufactured, deployed without elaboration. Five examples sufficient to characterize the register:

• “STAGED” (CBS, 659 likes)

• “Faker than 3 dollar bill BS” (CNN, 1,233 likes — the corpus’s most-engaged compressed staging tag)

• “False flag - again!!! Its Trump!!” (CBS, 50 likes)

• “Another Trump staged assassination attempt” (CBS, 4 likes)

• “OBJECTIVE investigation” (MS NOW, 193 likes — sarcastic compression performing the same claim through ironic inversion)

The engagement weighting matters. The CNN top-comment alone reaches over a thousand likes; the CBS top-staging assertion reaches 659. These are not marginal expressions. They are doing the load-bearing interpretive work for their commenter ecosystems.

Affective cover for conspiracy: humor as filter-evasion

A second register operates by routing the same propositional content through Schadenfreude. The comment reads as a joke, accumulates engagement on its humor, and carries the staging claim as embedded propositional content. Five examples:

• “Not a dry depend in the place” (CBS, 847 likes — corpus’s highest-engagement Schadenfreude comment overall, performing both bodily ridicule and the staging-as-fear-response implication)

• “And the band played on” (CBS, 1,875 likes — Titanic-band metaphor; commenters in reply explicitly extend the metaphor to Trump-as-doomed-ship-of-state)

• “That’s a helluva way to get out of the dinner berating” (CBS, 1,887 likes — corpus’s highest-engagement single comment overall; Trump-as-orchestrator motive expressed through what reads as a joke)

• “His bone spurs were tingling” (CBS, 15 likes — draft-dodger callback embedding cowardice-as-evidence-of-staging)

• “MAGA-HOAX 🤡” (Fox, recurring sustained comment thread — counter-direction Schadenfreude that performs conspiracy from the opposite political position)

This is the corpus’s clearest example of Schadenfreude operating as affective vehicle for conspiracy. Comments coded under category 2 frequently hold propositional content that would be coded under category 1 if rephrased without the humor wrapper. Where this dual-coding is operative, the comment contributes to the discursive-harm rate through the conspiracy route.

Pattern-of-staging: Butler as template

The Butler event functions as the corpus’s structural reference point. Butler-as-staged is asserted as established fact, and the WHCD shooting is interpreted as the next instance of an established pattern. (Background: on July 13, 2024, 20-year-old Thomas Matthew Crooks fired multiple shots at Trump from a rooftop during a campaign rally in Butler, Pennsylvania. A bullet grazed Trump’s right ear; Corey Comperatore, a rallygoer and volunteer firefighter, was killed; Crooks was killed by Secret Service counter-snipers. The Butler attempt is the historical referent for much of the staged-claim register documented below.) Six examples mark the range:

• “Just like when the shooter was on the roof and the Secret Service saw the guy with the rifle it was allowed to happen” (CBS, 840 likes — corpus’s highest-engagement Butler-retcon)

• “All the world’s a stage” (CBS, 251 likes — Shakespeare As You Like It reference, generalizing the staging frame to a worldview claim)

• “Have you ever seen his ear since that attempt in 2024? Doesn’t even look like he’s been shot” (CBS, 6 likes — physical-evidence Butler-retcon)

• “He didn’t get shot lol. No scar” (CBS, 5 likes)

• “He’s pulling another stunt like he did in Butler Pennsylvania when he was also crazy low in the polls” (CBS, 6 likes — pattern-plus-motive)

• “Like the ear.............” (CBS, 7 likes — minimal-form Butler-retcon as ironic shorthand)

The engagement weighting is again significant. The 840-like security-failure comment anchors a 102-reply thread. Within the corpus, the Butler-retcon register operates not as fringe revisionism but as assumed common ground at the saturated outlets, particularly TYT, MS NOW, and CBS.

Elaborated motive attributions

Where the compressed assertions perform the staging claim, the elaborated comments specify why the event was staged. Five motive frames recur across the corpus:

• Dinner-avoidance. The corpus-dominant motive at CBS, where the top-engagement comment (”helluva way to get out of the dinner berating,” 1,887 likes) routes through this frame. The variant: Trump staged the event to escape being roasted. Distinctive to MS NOW and CBS; minimal at the right-leaning outlets.

• Ballroom-funding. Stable motive at NBC, AP, USA Today, TYT. The variant: Trump staged the event to manufacture sympathy for the $400M White House ballroom expansion. The MS NOW elaboration adds a “bunker” variant (”$400 million ballroom/bunker,” 10 likes at CBS) that recurs at CBS.

• Epstein-files distraction. Most engagement-dense at MS NOW (top formulation 150 likes), where it appears as “staged event to distract from files.” Returns at CBS in compressed form (”He’s in the files” within a stanzaic 37-like Butler-denial formulation).

• Iran-war distraction. Cross-outlet, low-to-mid engagement. Variant: Trump staged the event to redirect attention from the Iran conflict. The strongest single formulation (at NBC, 8 likes) names “Russian and Israeli asset” as the operating coalition — coded under AS because of the Israel attribution.

• Poll-numbers / sympathy-bump. “He’s looking for a sympathy bump after those poll numbers” (CBS, 24 likes); “scraping the bottom of the barrel for pity because his ratings are so low” (CBS, 21 likes). Recurrent, low-to-mid engagement, broad across centrist and left-leaning outlets.

The motives are not mutually exclusive in the corpus. Compound formulations stack two or three together: “Cover up the Epstein files, or Iran or Trump 33% at the polling” (CBS, 1 like). The compounding is itself analytically significant — it suggests that the underlying interpretive task is not selecting the motive but generating a saturating motive landscape such that the staging claim is over-determined.

The CIA / intelligence-patsy frame

A subset of comments works the orchestrator slot through institutional rather than partisan agency. The CIA, “deep state,” “shadow government,” or “the agencies” replaces Trump as the figure pulling strings. Five examples:

• “The deep state attempted butler. Who knows who did this” (CBS, 1 like)

• “Wow another convincing Mossad-CIA joint charade!” (TYT, 3 likes — note hybridization with AS register)

• “Always shadow people” (CBS — minimal-form shadow-state frame)

• “It’ll be some trump supporter he will have promised a bunch of money to false flag. But like all his gifts, you can’t cash in if you are dead” (CBS, 2 likes — bribed-perpetrator-disposable-patsy variant)

• “Bro with all the money that they’re stealing you don’t think that they could pay better people to fake this” (CBS — meta-conspiracy frame)

The CIA-patsy frame is structurally adjacent to the AS register without requiring antisemitic content: the orchestrator slot is filled by a stateless or trans-state institutional agent rather than by a Jewish-control attribution. At TYT, this register hybridizes with the Mossad-orchestration register (the “Mossad-CIA joint charade” example). At the centrist outlets, it operates without that hybridization.

Reasoning structures

Three reasoning sub-registers organize the staging claims:

Security-failure-as-evidence. Slow Secret Service response, stepped-aside agents, missed shots, and the gunman’s ability to reach the hallway are read as evidence of staging rather than as evidence of operational failure. “Im amazed how slow security was to react here- WOW!! They should have been on this in milliseconds- not seconds!!” (CBS, 63 likes); “Of all the places you’d think would have f*cking security..” (CBS, 324 likes); “Why did they get the vice president out first?” (CBS, recurring sub-frame). This is the register the 840-like CBS Butler-retcon comment operates through.

Witness-skepticism. Trump’s apparent non-reaction is read as evidence of foreknowledge. “As the sounds happened, T did not even stop socializing. He didn’t even flinch or look around. Was this ‘staged’❓” (CBS, 17 likes); “So he’s just sitting there completely unphased while the room ducks and multiple service agents swarm him?” (CBS, 66 likes); “Pathetically staged he couldn’t even act surprised” (CBS, 38 likes). The register reaches saturation density at CBS — multiple commenters independently note Trump’s non-reaction as evidence.

Forced-binary. The disjunctive structure presents two options whose branches both indict Trump or the administration. “Either it’s staged or the secret service is so incompetent” (CBS, 6 likes); “1. He knew it was fake 2. He has Frontotemporal Dementia” (CBS, 6 likes). The forced-binary appears first at MS NOW and consolidates at CBS as a generalized reasoning structure.

Fascist-templating

A small but structurally significant cluster routes the staging claim through historical-precedent reasoning, drawing comparisons to Hitler, the Reichstag fire, and martial-law imposition. Five examples:

• “He desperately wants his Reichstag fire decree moment” (CBS, 3 likes — corpus’s first explicit Reichstag-fire formulation)

• “I’m waiting for the whole Martial Law to be (illegally) invoked” (CBS, 3 likes)

• “Sometimes that is true. Then again, nobody would have said that about Hitler” (CBS — Hitler-equivalence performing prospective justification of violence)

• “Ole ‘Red Terror’ trick” (CBS, 4 likes — historical-precedent compression)

• “He’s looking for ANY reason to take over or cancel these midterms” (CBS, 4 likes — instrumental-staging frame with electoral-suspension implication)

The fascist-templating register is small in absolute frequency — under 1% of comments — but disproportionately present in the engagement-weighted top tiers and in the elaborated-motive sub-cluster. It supplies the comparative framework against which the staging claim is rendered politically meaningful. The cross-corpus directionality of these analogies — Nazi/fascist templates pointed away from Jewish targets and toward the US executive — is taken up in the Synthesis, where it can be assessed against prior work in this research line.

Antisemitic Conspiracy Theories

The antisemitism rate in the corpus is low across nine outlets (0.5–2%) and elevated at TYT (4.0%). Where antisemitic content does appear, it operates in three structurally distinct registers — none of which exceeds five hits per 200-comment sample at any centrist outlet.

Cross-event reactivation

A small set of comments at ABC and NBC pairs the WHCD-staged claim with the Israel-killed-Kirk claim from the September 2025 corpus. Two examples:

• “Yes it was. And Israel killed Kirk.” (ABC) — overtly antisemitic cross-event linkage routed through Israeli agency

• “They will just pave over it the next day like they did with Charlie Kirk” (NBC, 6 likes) — institutionally-routed cross-event linkage with vague-shadow-state agent (antisemitism-adjacent rather than overtly antisemitic)

The two formulations together suggest cross-event reactivation operates in two modes — overtly antisemitism-routed and institutionally-routed — with the latter potentially serving as a discursive entry-point that primes antisemitic reactivation without requiring it explicitly.

The Russia/Israel double-asset frame

Distinctive to NBC, where it appears as the corpus’s most elaborated single-thread development of antisemitic conspiracy. Three examples:

• “It’s not just liberals. No one with a brain likes Trump, the Russian and Israeli asset.” (NBC, 8 likes)

• “yeah Trump works for Russia and the jews !!! Please don’t bring up that Russia and Israel hate each other and it makes no sense that they both control America but hey I’m a crazy leftists so I don’t have to make sense.” (NBC) — sarcastic counter-reply that performs the same claim through ostensible repudiation; the slippage from “Israel” to “the jews” is a classic antisemitic substitution

• “the secret juice tunnels must remain hidden” (NBC) — pizzagate / QAnon-adjacent intersecting with classical antisemitic conspiracy structure

The double-asset frame is structurally distinctive: it stacks a non-antisemitic conspiracy attribution (Russian control) with an antisemitic conspiracy attribution (Israeli/Jewish control), so that the antisemitic content rides on the back of a left-coded frame and acquires plausibility-by-association.

The Power-frame at ABC, NBC, AP, and CBS

Three centrist outlets each produce one elaborated single-comment Power-frame formulation. CBS produces the corpus’s most architecturally complete version. Five examples:

• “Netanyahu is your president” (ABC) — Israel-as-state Power-frame

• “Russian and Israeli asset” (NBC, 8 likes) — Israeli-asset Power-frame

• “The United Snakes of AIPAC.” (AP) — AIPAC-specific lobby Power-frame; “United Snakes” performs a punning corruption of “United States” with a dehumanizing register applied to the Jewish-American lobby organization

• “Trump is trying to get out of an illegal war. AIPAC/Adelson/Netanyahu are giving warning shots to stay in Iran and finish what you’ve been paid to do” (CBS, 3 likes) — corpus’s most architecturally complete AS Power-frame, stacking lobby + donor-network + state into one orchestrator coalition

• “probably cia or the tiny hats trying again” (CBS) — antisemitic euphemism for Jewish people, “tiny hats” referencing kippot

These four formulations together map the spectrum of contemporary antisemitic Power-frame elaboration in left-of-center US political discourse: from Israel-as-state through asset/agent through lobby (AIPAC) through fully coordinated lobby-donor-state coalition. The fact that none of them takes root as a generalized register at any single centrist outlet — they appear as compressed single-comment moves rather than as sustained interpretive programs — itself matters. The Power-frame architecture is available; it is being deployed in compressed form rather than as a saturating register.

TYT and the audience-supply route

TYT diverges structurally. Where the centrist antisemitic register operates through compressed single-comment formulations, the TYT antisemitic register includes fully elaborated antisemitic architecture of the kind documented in prior work in this research line. Five examples cover the range:

• “Wow another convincing Mossad-CIA joint charade!” (TYT, 3 likes) — Mossad-CIA hybridization, the corpus’s clearest formal combination of antisemitic and non-antisemitic conspiracy scaffolding within a single comment

• “But really by the family that owns and brags it founded that country and stole our fed and our way of tying our currency to its value in gold and other PMs.” (TYT) — Rothschild conspiracy, Federal Reserve control, gold-standard manipulation; the corpus’s single most antisemitism-laden comment

• “Ploy / BALLROOM” (TYT, 599 likes — embedded variant within elaborated thread that includes Mossad-orchestration framing)

• “LEGO video of psy-ops” (TYT, 762 likes — psy-ops staging frame within an architecture that elsewhere routes through Mossad)

• “Mossad killed Kirk and now they’re at it again” (TYT, paraphrased compression of recurring frame)

The TYT antisemitic register reactivates the Mossad-orchestration / Rothschild-finance-control / foreign-control architecture documented in prior work in this research line — but in a discourse environment where the editorial coverage supplies no Israel-related geopolitical pretext. This is the cleanest available evidence for the audience-supply route: the architecture reactivates not because the event invites it but because the audience carries it.

The structural finding the section as a whole supports is worth stating directly. The WHCD corpus shows that antisemitic scaffolding is not a necessary component of conspiracy architecture, but a conditional attachment. Where in earlier corpora the orchestrator slot was typically filled by Israel/Jews, here the same slot is filled by Trump, his administration, the security state, the CIA, or unspecified institutional actors — and conspiracy continues to function as the load-bearing register. Antisemitism reattaches when one of the two routes opens; it does not reattach when both are closed.

Schadenfreude, Sarcasm, and Ridicule

Across the corpus, Schadenfreude rates cluster between ~9.5% and ~12.5% at nine outlets, with CBS at the high end at ~15%. Consistent with patterns documented in foiled-attack contexts in prior work in this research line (notably the Temple Israel analysis), the foiled outcome shifts the affective economy from explicit death wishes (which dominated mass-casualty corpora) toward irony, sarcasm, and dark wordplay. Mockery here does not celebrate a killing that occurred; it ridicules the attempt, the target’s reaction, the security failure, the production values of the released video, and the entire ceremonial frame of the WHCD itself. The register operates in six sub-modes.

Bodily ridicule and incontinence humor

Saturated at CBS, where the 847-like Depends comment anchors a 28-reply thread of riffs; recurs at NBC and CNN with bowel- and flatulence-coded variants.

• “Not a dry depend in the place” (CBS, 847 likes — corpus’s highest-engagement Schadenfreude comment overall)

• “Gun shot??? Donny did u fart again???” (CNN, 80 likes) — recasts the gunshot as flatulence

• “he doesnt fart, he poos” (CNN)

• “Someone yelled, EVACUATE,’ and Donald dropped one in the diaper. No wonder everyone ran!!!” (NBC, 12 likes) — recasts the security evacuation as a bowel-incontinence event

Appearance and characterological mockery

Distinct from incontinence humor: this register works through name-calling, appearance-coded epithets, and characterological reduction. Most engagement-dense at CNN and ABC.

• “the mango buffoon” (CNN, recurring formulation) — appearance + character compound

• “donnie fatass and his vladministration” (CNN, recurring formulation) — appearance epithet plus Russia-asset implication (”vlad” / Putin) embedded in the administration label

• “Is he orangier 😮” (ABC, 77 likes) — appearance-coded

• “tangerine tyrant” (ABC) — appearance + political-character compound

• “Don Snorleon here aka the Nodfather” (ABC) — sleep/age + Mafia-metaphor compound; positions Trump as a doddering crime-family head

• “His bone spurs were tingling” (CBS, 15 likes) — draft-dodger callback; characterological cowardice attribution

Production-values mockery — Forbes, USA Today

Saturated at Forbes and USA Today, where the released surveillance footage drew sustained mockery for its image quality. Operates through camera-quality and equipment-vintage jokes.

• “Why is the camera from 1950?” (Forbes, 701 likes — the highest-engagement comment in the Forbes sample)

• “You get better security footage from 7-Eleven’s cameras.” (USA Today, 908 likes — the highest-engagement production-values comment in the corpus)

• “Potato cam for the white house correspondents dinner 🤌” (Forbes, 262 likes)

• “It’s a really crappy security camera” (USA Today, 130 likes)

• “Not 1950’s ... definitely from Charlie Chaplin Silent Movie Era 😂😂😂” (Forbes)

• “MEEP MEEP!!!!!!!!” (Forbes, 82 likes) — Road Runner reference

• “trillions of us dollars with 1940s camera quality” (USA Today)

• “the epstein video was better quality” (USA Today) — combines production-values mockery with Epstein-frame

The CBS Titanic-band metaphor (”And the band played on,” 1,875 likes) sits structurally adjacent to this register but is dual-coded under Conspiracy Theories, where the embedded staging claim (”Trump-as-doomed-ship-of-state”) is the load-bearing analytical content.

Mockery of the WHCD ritual itself — Fox News

A distinctive Fox sub-register: ridicule of the WHCD ceremonial frame, often performed by left-coded counter-commenters arriving at the Fox thread.

• “How can anyone continue this dinner? F the dinner. F the press.” (Fox, 349 likes — the highest-engagement non-mockery comment in the Fox sample)

• “Hell of a way to get out of rubber chicken...” (Fox)

• “is Clint still talking to empty chairs again?” (Fox) — meta-ridicule, callback to Eastwood’s 2012 RNC bit

• “ANOTHER ONE!! 😂😂😂” (Fox)

• “It must be painful 😅😅😅” (Fox)

• “We all knew Trump would try find a way out of this event 😂” (Fox)

Mock-condolence — MS NOW

Distinctive to MS NOW: a feigned-grief-then-reveal structure where the apparent condolence inverts mid-comment.

• “Heartbreaking news. ... Trump is okay. 😢” (MS NOW, 54 likes) — corpus’s clearest single example of the mock-condolence structure

The mock-condolence sub-register is narrow in absolute frequency (a handful of comments at MS NOW). Its analytical interest is structural rather than statistical: it inverts the conventional condolence frame to produce ridicule that reads, on first scan, as expression of grief. Comments that ridicule Trump’s cowardice during the evacuation (”Trumps staged exit from ultimate humiliation…he is a coward,” CNN, 174 likes) operate adjacent to this register but route their analytical content through the staging-frame architecture and are therefore coded under Conspiracy Theories. Comments that lament the failure of the attempt (”Why the hell do they keep missing the target?” CBS) are regret-the-miss formulations and are coded under Glorification.

Theological inversion and condemnation — Fox News

Two related Fox sub-registers operating in the same comment ecosystem: replies that systematically reverse the divine-blessing formula, and theological condemnations that cast Trump within biblical-villain or moral-damnation scripts. Both function as ridicule of pro-Trump theological framings.

• “DEVIL BLESS AMERICA” (Fox) — explicit inversion of “God Bless America”

• “God has better things to do” (Fox) — inversion through divine indifference

• “Why would God want to bless Stupid America” (Fox) — inversion through divine judgment

• “If God were to bless America, Trump wouldn’t be president anymore” (Fox) — inversion through counterfactual

• “Like Trump is King Herod the Great?” (Fox) — theological condemnation; positions Trump within the Herod-the-Great script as a biblical villain

• “Trump is the walking embodiment of all 7 Deadly Sins. He’s going to burn in hell 🤣” (Fox) — theological condemnation through moral-damnation script

Schadenfreude is more pervasive than the rates alone suggest, and dominates the engagement-weighted profile of most outlets’ threads. In several distinct sub-registers — the Fox “ANOTHER MAGA HOAX 🤡” pattern, the CNN/ABC ear-cluster, the CBS Depends and Titanic-band clusters — it operates as affective cover for conspiracy, with the humor wrapping a propositional staging claim. Those sub-registers are coded under Conspiracy Theories rather than here.

Affirmation, Justification, and Glorification of Violence

This category contains two related but distinct registers: affirmations, which cast the act as inevitable, deserved, or reciprocally warranted, and glorifications, which celebrate it as heroic, joyous, or insufficient. The structural difference between the WHCD context and completed-killing contexts in prior work in this research line is that the WHCD attack was foiled — the target survived and the perpetrator was arrested alive. This outcome conditions a distinctive register: comments that regret the miss or endorse future attempts rather than celebrating a death that occurred. The Glorification rate sits between 3.5% (Fox, Forbes, CNN) and 7.5% (TYT). The register operates in five sub-modes.

Affirmation and justification

Affirmation framings cast the attempt as a predictable or reciprocal consequence of Trump’s politics or character.

• “Would you take a bullet for this clown” (Forbes)

• “It’s only trump, no need to try that hard” (Forbes)

• “I don’t care if the ‘orange man’ gets shot” (CNN, 35 likes) — direct first-person disclaimer of normative concern about political violence

• “Looks like he needs to hide. Twice attempted assault on him. Guess he Not a good President.” (ABC) — naturalizes the assassination attempt as predictable response to bad presidency

• “Trump probably killed more people than the rest of the world leaders combined once you factor in the fallout of this war. Doctors in Cuba are having to hand pump respirators for babies in the hospital because of Trump’s embargo. Trump can’t get got soon enough!” (ABC) — ties global death-toll to “can’t get got soon enough”

• “your team wanted a war so bad they were willing to bomb schools hospitals civilian infrastructure and densly populated urban areas. Cry me a river. Civility is over and your team did it in. Trump is old, he’ll be gone soon enough amen” (ABC) — full reciprocal-violence justification

Regret-the-miss as direct prospective endorsement

The corpus’s signature affirmation register, foiled-attack-specific. Reaches cluster density at ABC, CBS, and TYT.

• “Damn They Miss Again!!! 😮😮😮” (ABC, 33 likes) — explicit lament; “again” places the failed attempt within the sequence of prior attempts (Butler, West Palm Beach)

• “unfortunately Trump is still alive” (CNN) — uses “unfortunately” to perform the inverted moral framework explicitly

• “Why do they always miss” (Forbes) — “always” both encompasses and laments the sequence of failed attempts

• “Luigi did it better” (Forbes) — invokes Luigi Mangione (alleged perpetrator of the December 2024 UnitedHealthcare CEO killing) as comparative reference for “competent” political assassination

• “We need Tyler Robinson” (AP, 1 like) — names the alleged shooter of Charlie Kirk as the model

• “Counting his days!” (ABC, 8 likes)

• “Hope they get Trump next time” (NBC) — explicit prospective endorsement of a future successful attempt; dual-coded under Glorification and Threats

• “trump should be eliminated” (CBS, 1 like) — corpus’s most direct elimination call

• “I hope one of these attempts are successful” (CBS) — explicit prospective endorsement

• “Just get it done!” (CBS, 8 likes) — minimal-form prospective endorsement

• “Thats a shame Trump and Vance got away tyalk about bad luck!” (TYT) — extends the regret register from single target to target-pair

• “Can’t ANY of these people get the job done???” (TYT) — places the sequence of failed attempts within an exasperated-rhetorical-question frame

Reciprocal-violence frame — anti-Trump position

Anti-Trump commenters frame WHCD attendees, or Trump’s policies, as deserving recipients of violence by analogy to other victims.

• “Now all of those folks know what children feel like when gunmen come into their schools” (CBS, 311 likes — corpus’s most-engaged reciprocal-violence formulation)

• “He is good, he was just trying to take out the trash” (Forbes) — explicit moral inversion; perpetrator coded as good, target coded as trash

• “Why not! Trump is horrible and people are been killed everyday in Gaza, Lebanon not to mention Iran.” (Forbes) — links the WHCD attempt to civilian death tolls in Gaza, Lebanon, and Iran as moral justification

Glorification of state-level mass violence — pro-Trump position

Distinct from the anti-Trump reciprocal-violence frame above: a small but distinctive Fox sub-register where pro-Trump commenters glorify mass state-level violence directed at Iranian civilians, often in response to anti-Trump or Iranian-identified commenters in the same threads. The register operates from the political position opposite to the anti-Trump reciprocal-violence frame, which is why it is reported separately.

• “God Bless Iran, drop a bomb on this stupid, dispicable regime already... the whole world will be a better place with 6 million stupid assholes in this world” (Fox) — combines glorification of mass civilian violence, Islamophobia, and Holocaust trivialization in a single sentence; the “6 million” figure operates as a Holocaust echo, transposing the Jewish death toll into a quantity-template for a fantasized destruction of Iranian civilians

• “Hahah we about to wipe out ur civilization” (Fox) — directed at a self-identified Iranian commenter

Pre-emptive disavowal

First identified at MS NOW, consolidates at CBS. The disclaimer attempts to immunize the prospective-violence content that follows by asserting non-advocacy in advance — a strategic move whose function is to evade content-moderation flagging rather than to genuinely disclaim the content.

• “I’m not advocating violence. But do not forget what Trump is. A literal cancer on American society. The man is a moronic monster who only cares about himself.” (MS NOW, 8 likes) — the cleanest single formulation

• “No... you’re not advocating violence. Because you said it before you made your actual statement.” (MS NOW) — reply-comment that names the disavowal pattern directly

• “its karma dont be a cosmic loser. you dont comment ‘sicko’ when terrorists and dictators get headshot” (MS NOW) — explicit terrorist/dictator-comparison

• “orange man has honestly killed millions more than the few rich he benefited over his terms... its basically less bad than having empathy for hitler and kim” (MS NOW) — Hitler-comparison mass-death-tally justification

These formulations represent the foiled-attack equivalent of the glorification registers documented in completed-killing contexts in prior work in this research line. They do not celebrate a death that occurred but endorse the attempt and lament its failure, thereby projecting endorsement of future attempts. Worth flagging for the Synthesis: violence endorsement persists across the corpus but shifts from prospective threat to retrospective approval — a register-shift that the foiled-attack outcome conditions but does not by itself explain.

Discursive Blame Projections

The blame economy is recognizable but unevenly distributed. Across the corpus the four sub-targets aggregate to ~4.9%, against ~29.7% in the Kirk corpus’s nine matched outlets. The structural finding: in nine of the ten samples, the blame frame is displaced by conspiracy — commenters skip past asking who created the violent climate and proceed directly to the claim that the event itself did not happen as reported. Only Fox News produces a blame economy in something like the density characteristic of comment ecosystems analyzed in prior work in this research line. The four blame sub-targets all materialize, though their distribution varies sharply across outlets.

The Left’s Fault

Anchored by the rapid surfacing of perpetrator Cole Tomas Allen’s $25 donation to the Harris campaign and his identification as a California-based teacher. On Fox, commenters projected blame onto Democrats and the broader Left in formulations that flattened the political opposition into a singular guilty collective.

• “This is from the Radical Left. They’ve caused craziness.” (Fox)

• “Democrats or Islamic Republic? Same thing these days!” (Fox) — note the Islamophobic edge: the equation works through a dehumanizing collapse of the political opponent into a caricatured Muslim-state enemy

• “Liberals are out of control” (Fox)

• “A black left wing teacher assassin from crazy California. The education system is riddled with these loons. Not good for the narrative that is the liberal media but I know they will give it their best try.” (Forbes)

• “Why we should entirely redo the public school systems!” (Forbes) — calls for institutional consequences

• “Braindead. Clearly brainwashed by leftist media, like the attempted assassin.” (Forbes)

• “Predictable Comucrat air-head, knee-jerk response — who’s surprised at that, nobody in the world!!!” (Forbes)

• “TDS raging in those two woke brain cells is all you have liberal lunatic” (Fox)

• “the lefty gunman from cali? No doubt” (AP)

• “Dem and msm rhetoric led to this” (ABC) — cross-direction comment in ABC thread; compresses Left’s-Fault and Media’s-Fault into a single dual-target attribution

The Right’s / MAGA’s Fault

The reciprocal blame frame appears across outlets, with comparable density at CNN, ABC, NBC, and AP. Two of the eight bullets below are from Fox itself, where they operate as cross-direction commenters arriving in the Fox thread — the same dynamic noted in the Schadenfreude WHCD-ritual sub-cluster. A distinctive sub-strand presupposes a pattern of MAGA-coded political violence.

• “yet another dissatisfied MAGA Moron with a gun?” (Fox) — cross-direction comment in Fox thread

• “MAGA are the unhinged group at large… EVERY single MAGA is a maniac. Fckn NAZI’s the lot of them.” (Fox) — cross-direction comment in Fox thread

• “Lol blaming Democrats, of previous failed attempts all were magahats” (CNN) — meta-frame claim: refutes the Right’s-Fault attribution by citing the political identification of prior assassins

• “Anyone still supporting tRump is an accessory to every crime he has committed. Pdfiles always cover each others a**es” (ABC) — encoded keyword-evading spelling (”Pdfiles” for “pedophiles”) routes the MAGA-blame frame through Epstein-era pedophile-network framing

• “Lone wolves who generally turn out to be former Trump supporters now equal the ENTIRE LEFT? Trump’s cult should get new talking points.” (ABC, 14 likes)

• “the shooters were disgruntled Trumpers, what lies are you trying to sell here?” (NBC)

• “Every Trump attempt has been by a disgruntled Trump voter.” (AP, 6 likes) — explicit pattern claim

• “Funny because it’s generally MAGA showing signs of psychosis.” (AP)

The Media’s Fault

This sub-target was predicted to be a distinctive feature of the WHCD case given the venue. In the corpus, however, it is markedly thinner than expected. The hits operate as conspiracy-cover-up rather than as climate-of-incitement.

• “By the press / They said they had something planned. Asswipes” (Fox) — direct accusation of press complicity in the attack

• “Shame on the media for not asking more questions in 2024” (Forbes)

• “F CBS. For not protecting our president” (CBS, 2 likes)

• “Now the press is facing the result of their rhetoric” (CBS, 4 likes)

• “TDS is high on this channel. OH...nevermind it’s NBC of course it’s full of nuts.” (NBC, 14 likes) — partisan press-delegitimization aimed at the host outlet itself

Trump’s Own Fault

In the right-leaning outlets, this sub-target is effectively absent. Where it appears in this category, it operates as characterological framing in which Trump’s behavior or rhetoric is the explanatory variable — most engagement-dense at the progressive outlets. The displacement form (”he orchestrated this himself”) functionally migrates the sub-target into Conspiracy and is reported there.

• “Don’t let the Constitution Stomping, Freedom Crushing, Fascist Toddler and his Incompetent Circus of Sycophants back in!! He doesn’t deserve to be in the same room as our Truth Tellers” (CNN)

• “no one more hateful than Trump himself. Listen to his rallies listen to his speeches most hateful evil human being there is” (ABC)

• “Trump would never say half the **** he says if he was standing in front of people. He says it because he knows he has security that are paid to protect him. He antagonizes people with his vileness and emotionally unstable people have unfortunately shown they will do this.” (NBC)

The blame distribution at WHCD does not mirror the Kirk distribution scaled down. It collapses unevenly: Media’s-Fault is markedly thinner than the press-freedom venue would have predicted and operates mainly as cover-up rhetoric; Right’s-Fault concentrates at the centrist outlets (CNN, ABC, NBC, AP); Left’s-Fault concentrates almost entirely at Fox; Trump’s-Own routes through characterological framing rather than through causal blame for the attack. The partisan-attribution work that would otherwise occupy this category has migrated to the conspiracy register, where Trump-as-orchestrator answers a different question than Trump-created-the-climate.

Threats and Calls for Retaliation

The dominant sub-type on Fox, anchored by the rapid mobilization of Allen’s Harris donation as discursive evidence. Fox threat content frequently fuses generalized-political and individual-interpersonal targeting: the “you” addressed in many of these comments is a stand-in for the broader Left, but the speech act is interpersonal. Four representative formulations from across the Fox sample’s threat hits, including two from a single commenter using a recurring template:

• “scumbag left are finished. You fuvkhead scumbags” (Fox) — eliminationist; addressed to the Left collectively

• “You and your kind need to be classified as communist and designated public enemy #1” (Fox) — calls for state-level designation/persecution of political opponents through generic “you / your kind” Left-stand-in addressee

• “Trump will make an appointment for you with the Mad Mullahs and Ayatollahs of Tehran” (Fox) — coded threat invoking Trump-as-instrument-of-Iran-policy; rhetorically merges domestic political opponents with foreign-state enemies through Islamophobic “Mad Mullahs” shorthand. The same fusion appears in the Left’s-Fault sub-cluster (”Democrats or Islamic Republic? Same thing these days!”); recurring formulation by single commenter

• “President Trump will make a lunchtime appointment with you and the Mad Mullahs and Ayatollahs of Tehran” (Fox) — same commenter, extended formulation

Reciprocal / commenter-on-commenter threats

A small but present sub-type on Fox, CNN, NBC, and AP: direct hostility to other commenters within the same thread, without the Left-as-collective-target framing of the previous sub-cluster.

• “I hope that this comment is last in your life” (Fox)

• “You worthless m’fer” (Fox) — directed at a commenter calling the event fake

• “shut up punk quit mocking the president he could’ve been assassinated I’m sick of your childish mind” (Fox) — coercive silencing

• “go play in traffic -DUMBAZZ-” (CNN, 12 likes) — directed at another commenter

• “I stay in Austin tx when ever you in town I drop my location and you can stand on that” (NBC) — direct interpersonal challenge to face-to-face confrontation; from the same commenter as “Hope they get Trump next time” in the Glorification regret-the-miss sub-cluster. The migration from Trump-targeted prospective endorsement to interpersonal face-to-face threat at the level of a single commenter is itself analytically significant: it shows the threat register, where it appears at outlets outside Fox, is performed by individuals already operating at the prospective-violence end of the discursive escalation chain

• “DOES YOUR HUSBAND ABUSE YOU???” (AP) — gendered hostility, commenter-on-commenter

Conditional / contingent threats — single-instance

A distinctive single formulation at AP, isolated in the corpus rather than constituting a register, but worth flagging because it conditionally nominates a third-party population for collective punishment.

• “Nuke Iran if the USA president gets assassinated” (AP) — invokes nuclear retaliation as response to a hypothetical successful attempt

Sub-types largely absent: civil-war fantasies, threats toward the press, apocalyptic threats

These three sub-types — prominent in mass-casualty corpora in prior work in this research line — are largely absent from the WHCD corpus. Threats toward the press do not appear at all; anti-press content circulates instead through Schadenfreude (WHCD-ritual ridicule) and Conspiracy Theories (press-complicity claims). Civil-war fantasies and apocalyptic / antisemitism-linked threats are sparse, consistent with the near-absence of antisemitic conspiracy. Naming the absence is itself a finding: the WHCD discourse environment does not generate the apocalyptic and civil-war registers that prior corpora produced even at lower overall threat densities.

The structurally interesting finding is that the threats register did not expand even as the conspiracy register expanded by an order of magnitude. Where one might expect that conspiracy density would feed threat density, the corpus shows the opposite: the saturated outlets (MS NOW 1.0%, TYT 2.5%, CBS 1.5%) sit at or below the cross-corpus average. The reading I propose: conspiratorial saturation may absorb the energy that elsewhere fuels prospective-violence formulations. Explaining the event away forecloses demanding more violence in response to it.

A separate observation worth flagging: the corpus’s clearest single-comment prospective formulations — NBC’s “Hope they get Trump next time,” AP’s “We need Tyler Robinson,” and CBS’s “trump should be eliminated” — do not arrive as parts of partisan-blame structures. They arrive as glorification crossing the line into prospective endorsement, and are dual-coded under Glorification and Threats. This complicates the otherwise clean threat-blame covariance: prospective threat in this corpus migrates not from partisan blame but from regret-the-miss formulations.

These two findings operate at different scales. At the outlet level, conspiratorial saturation appears to absorb threat energy; at the individual-comment level, the strongest prospective threats migrate not from blame but from glorification’s regret-the-miss register. Both findings sit comfortably with the broader pattern: the canonical blame → conspiracy → glorification → threats escalation chain in this research line no longer organizes prospective-violence formulations in the way the analytical framework would predict. Conspiracy absorbs threat energy from above; regret-the-miss generates the residual threat content from a different category entirely.

Comparison with the Kirk Findings

Rate-by-rate

(Comparison restricted to the nine outlets coded in both studies. CBS is in the WHCD ten-outlet design and was in the Kirk twelve-outlet design, but its per-outlet Kirk rates are not retained in current working notes; CBS is therefore reported in the WHCD-only rate table above and excluded from this matched-comparison aggregate.)

Four findings that travel beyond this corpus

Finding 1 — Two distinct discursive regimes. The Kirk and WHCD corpora are not just two events with different content. They are two different organizing logics for post-violence commentary. The Kirk corpus is organized around an attribution register (~30% blame across four sub-targets, conspiracy a marginal 6.5% sub-stratum). The WHCD corpus is organized around a reality-contestation register (~27% conspiracy, blame collapsed to ~5%). The same speakers, on comparable outlets, processing structurally similar events, rotate the entire interpretive frame from who is responsible? to did this happen?

Finding 2 — A reconfigured discursive escalation chain. In earlier work the canonical chain ran blame → conspiracy → glorification → threats: the discourse moved from contested attribution through hidden-actor framings to celebration of completed violence to prospective violence. In the WHCD corpus that chain is reconfigured. Conspiracy is now the entry-point register rather than the second stage. The downstream registers also reshape: glorification works through regret-the-miss humor rather than through celebration; threats stop short of the apocalyptic, civil-war, and press-targeted registers documented in prior corpora — these registers are largely absent rather than merely thinner. The most plausible reading: the conspiratorial saturation absorbs energy that elsewhere fuels threats. The explanatory work previously performed by calls for further violence is now performed by claims that the event itself was staged.

Finding 3 — Antisemitism is decoupled from conspiracy architecture, with two routes back in. In earlier work — the Capital Jewish Museum shooting, AiRT Part III, Temple Israel — conspiracy discourse around political-violence events frequently required antisemitic scaffolding to function: the orchestrator slot was filled by Mossad, Israel, Rothschild-linked actors, or Jewish-control variants because the architecture assumed a transnational hidden orchestrator. In the Kirk corpus this requirement was already weakening (non-AS at 2.9% matched AS at 3.6%, both marginal). In the WHCD corpus the requirement is gone: non-AS expands ninefold, AS contracts. The staged-event frame, the patsy frame, the hidden-orchestrator slot remain present and active — but the orchestrator slot is now filled by Trump, his administration, the security state, the CIA, “globalists,” or unspecified institutional actors. In this register, antisemitism becomes structurally optional rather than structurally necessary. The MS NOW–TYT contrast specifies the mechanism: same conspiracy density (37–39%), same event context, opposite antisemitism rates (0.5% vs 4.0%). The variable is audience-ideological reservoir. Antisemitic scaffolding reattaches through two routes — event-context supply (Kirk’s pro-Israel record, embassy attack, Temple Israel) and audience-ideological supply (TYT’s reservoir of Mossad / Israel-foreign-control / Rothschild frames). Either route alone is sufficient; the absence of both produces conspiracy without antisemitism.

Finding 4 — Two-layer cognitive/affective system. Cognitive layer (conspiracy) organizes interpretation; affective layer (Schadenfreude) organizes circulation. The two layers operate in parallel, with significant dual-coding: the same comment performs both interpretive work (asserting the staging frame) and affective work (mobilizing the joke). At the most cognitively saturated outlets — particularly CBS and TYT — the cognitive layer becomes engagement-leading directly, without affective vehicle: bare staging assertions reach hundreds of likes (CBS’s “STAGED” at 659; TYT’s “Ploy / BALLROOM” at 599). MS NOW saturates at the rate level (39% non-AS conspiracy, the highest in the corpus) but its bare staging assertions sit in the lower-hundreds engagement tier rather than anchoring the engagement-leading pattern at the level of single comments. Methodological consequence: engagement-weighted analysis systematically over-reports affect and under-reports structure where the cognitive layer is mature enough to circulate without humor cover.

Conspiracy as transfer function

These findings together support a sharper formulation of the structural-portability claim. Conspiracy discourse in these comment ecosystems operates as a transfer function: a stable narrative architecture (manufactured event, hidden orchestrator, strategic motive, perpetrator-as-patsy, security-failure-as-evidence, witness-skepticism) that takes the event context and the audience reservoir as inputs and outputs the agent attribution most rhetorically available given those inputs. In the Kirk case, the input (a polarizing partisan figure with a documented pro-Israel record, killed by a hyper-online young perpetrator) routed the architecture through partisan blame and antisemitic scaffolding. In the WHCD case, the input (a sitting president attacked at a press dinner, perpetrator identified within hours, no Jewish-control geopolitical scaffold available) routes the same architecture through Trump-as-orchestrator and security-state-as-instrument framings — except at TYT, where the audience-ideological reservoir routes the architecture through its own frame regardless of event context. The architecture is the constant; both event context and audience reservoir are variables.

Put more plainly: the same basic story-shape gets reused across events, and who plays the villain changes depending on what the event makes available and what the audience is already disposed to believe. Conspiracy in digital discourse is not ideologically fixed but template-driven and dual-input-conditioned. This formulation specifies a mechanism rather than asserting an observation.

Conclusion

This study presents a structural finding that, if it generalizes, reshapes how discourse around political-violence events should be analyzed and monitored.

The headline empirical result is the inversion of conspiracy and blame between the Kirk and WHCD corpora across nine matched outlets. Non-antisemitic conspiracy expands by an order of magnitude; blame collapses by a factor of six; the antisemitic conspiracy register contracts to roughly a third of its density in the Kirk corpus. The same speakers, on comparable outlets, processing structurally similar events, produce categorically different category distributions. The interpretive task that “blame the Left / Right / Media / Kirk himself” performed in the Kirk corpus is performed in the WHCD corpus by “the event was staged.” Different rhetorical surface, functionally similar work: assigning meaning to a violent event by locating it within a recognizable adversarial frame.

The structural reading: conspiracy discourse in these environments operates as a transfer function. A stable narrative architecture — manufactured event, hidden orchestrator, strategic motive, perpetrator-as-patsy, security-failure-as-evidence, witness-skepticism — takes event context and audience reservoir as inputs and outputs the agent attribution most rhetorically available given those inputs. The architecture is portable across radically different agent attributions. In the Kirk case, the architecture routed through partisan-blame and Mossad-orchestration. In the WHCD case, the same architecture routes through Trump-as-orchestrator and security-state-as-instrument — except at TYT, where the audience-ideological reservoir overrides the event-context absence and the architecture continues to route through its established Mossad / Rothschild / foreign-control channels. The architecture is the constant; event context and audience reservoir are independent variables.

Three operational implications follow from the transfer-function model developed across the Quantitative Overview and the Comparison with the Kirk Findings.

First, monitoring conspiracy density and monitoring antisemitism density are now distinct surveillance tasks. Where the prior research line treated conspiratorial response to political violence as substantially overlapping with antisemitic response — because the orchestrator slot in the conspiratorial architecture was typically filled by Israel/Jews — the WHCD case shows that the two can decouple cleanly. Surveillance regimes calibrated only to antisemitic markers will miss the bulk of the discursive harm in events whose geopolitical context does not supply a Jewish-control scaffold and whose audience does not carry one in. The MS NOW finding is methodologically clarifying: 39% conspiracy density at 0.5% antisemitism is unambiguously a high-discursive-harm comment ecosystem, but a surveillance pipeline trained on antisemitic content alone would register it as approximately clean.

Second, the two-route mechanism specifies what surveillance of antisemitic-conspiracy reactivation should attend to. Event-context supply (the event itself supplies a Jewish-control geopolitical pretext) is the route the prior research line has implicitly tracked. Audience-ideological supply (the audience already carries the architecture in regardless of the event) is the route the TYT finding identifies as analytically distinct and operationally significant. Surveillance pipelines should therefore attend to platform-level audience composition as a leading indicator of antisemitic reactivation potential, not only to event characteristics.

Third, engagement-weighted analysis systematically over-reports affect and under-reports structure where the cognitive layer of the discourse is mature enough to circulate without a humor wrapper. The CBS finding is methodologically clarifying: the corpus’s two highest-engagement comments overall (CBS, both around 1,880 likes) are dual-coded conspiracy formulations whose engagement is driven by the affective vehicle (the dinner-avoidance joke; the Titanic-band metaphor) but whose interpretive work is propositional staging-claim content. A monitoring regime that ranks by engagement and codes only on surface affect will systematically miss the cognitive layer that the engagement is in fact circulating. The complementary point: rate-only monitoring without engagement weighting will conversely miss outlets like CBS and TYT, where the cognitive layer is engagement-leading at the level of single comments — bare staging assertions reach hundreds of likes without affective vehicle. Two different kinds of saturation, two different operational signals: rate-level saturation (most clearly at MS NOW, 39% non-AS conspiracy density) and engagement-leading saturation (most clearly at CBS and TYT). Both signals need to be tracked simultaneously; pipelines that monitor only one will systematically miss outlets that saturate via the other route.

A note on what this study does not claim. The reality-contestation register identified here is documented across one large-n corpus and one comparative baseline. That is enough to license a structural claim about the WHCD case and a comparative claim across two events; it is not enough to license a generalized claim about a permanent epistemic shift in post-violence discourse. The discursive register identified at WHCD may turn out to be event-conditioned in ways the present study cannot fully specify — driven by the foiled-attack outcome, the rapid perpetrator identification, the absence of geopolitical scaffold, or some combination. Replication across additional events — particularly events that share some of WHCD’s structural features but differ on others — is required before the register can be characterized as a stable feature of contemporary digital response to political violence rather than as a feature of this case. The architecture and the dual-route mechanism, by contrast, draw on two independent corpora and travel better.

A broader point is worth stating explicitly. The methodological infrastructure for this kind of work — interdisciplinary social-media studies that combine discourse analysis, corpus linguistics, and large-language-model-assisted annotation — has matured in the past two years to the point where the analytical pipeline from real-time corpus capture to structured diagnostic output is genuinely within reach. The link between online radicalization and offline political violence is no longer speculative. What remains underdeveloped is the institutional capacity to act on what these methods can already see: to move from retrospective documentation to early warning, and from early warning to platform-level and policy-level intervention before staged-event narratives, regret-the-miss endorsements, and audience-supplied antisemitic scaffolding consolidate into the next escalation. The technical and interdisciplinary frameworks exist. The operational uptake is the bottleneck.

The empty chair is the literal one — Trump’s, after the evacuation. It is also the discursive one — the chair where blame used to sit in this research line’s prior corpora, and where conspiracy now sits in its place. The shift is structural, the mechanism is specifiable, and the operational consequences for surveillance and platform governance are direct.

Notes on Coding and Methodology

All comments were coded against the five-category schema, adapted from the analytical framework developed across prior work in this research line (the Decoding Antisemitism Lexicon, Becker et al. 2024, 41 concepts). A comment may hit multiple categories; dual-coding is noted where operative. Rates are reported per-category. The discursive-harm rate is computed as the proportion of comments containing at least one hit in categories 1 (Conspiracy Theories, both antisemitic and non-antisemitic), 3 (Glorification), or 5 (Threats). Schadenfreude (2) and Blame (4) are excluded from the rate on the grounds that neither category, on its own, constitutes harmful discourse. Where Schadenfreude operates as affective cover for conspiracy theories, the same comment is dual-coded under Conspiracy Theories and contributes to the discursive-harm rate through that route.

The dual-rate reporting structure — antisemitism rate alongside discursive-harm rate — reflects the analytical reality of the WHCD case: the dominant register is political-violence and conspiracy discourse, of which antisemitic content is a thin sub-stratum.

The Kirk study (Becker, September 2025) and this study share methodological features: identical 200-comment sampling per outlet, chronological draw from the first video published by each outlet within 24 hours of the event, the same coding framework. Kirk used a ten-category design separating Blame (Left / Right / Media / Kirk himself), Lack of Empathy, Schadenfreude, Glorification, Threats, Non-Antisemitic Conspiracy Theories, and Antisemitic Conspiracy Theories; WHCD collapsed Lack of Empathy into adjacent categories where its content lived more naturally. The four blame sub-targets are reported separately in the WHCD analysis above; in the comparison they are aggregated to align with the Kirk study’s blame totals. CBS was in both corpora’s design; CBS-Kirk per-outlet rates are not retained in current working notes, so CBS appears in the WHCD-only rate table and is excluded from the matched nine-outlet Kirk-WHCD aggregate.

Engagement-weighted prominence and frequency-weighted prominence diverge in this corpus; high-engagement comments are flagged where they distort the picture.

The core argument moves from historical framing through the mechanics of contemporary digital antisemitism — implicitness, coded language, plausible deniability — to the structural conditions that accelerate its normalization online. The emphasis is on the hidden register: the vast, implicit underbelly that explicit-only approaches miss entirely.

The film closes on opportunity: for the first time, social media studies make it possible to study Jew-hatred — like any other hate ideology — in real time and at scale. Social media isn't just the problem; it's the evidence base. The tools exist. The question is whether we'll use the interdisciplinary methods that combine humanities, social sciences, and technology to understand these phenomena in fundamentally new ways — and, not least, to sharpen our ability to counter bigotry and exclusion across research, education, politics, law, and security.

Funded by the Hertie Foundation (Gemeinnützige Hertie-Stiftung), Germany.

The clip is also on YouTube, with subtitles in English, German, French, Spanish, Hebrew, and Arabic:

AddressHate Research Scholar at NYU’s Center for the Study of Antisemitism | Lead, Decoding Antisemitism | Research Advisor, AddressHate | Editor-in-Chief, Digital Hate Review

Introduction

On March 12, 2026, a man identified in preliminary law enforcement reports as Ayman Mohamad Ghazali drove a truck through the entrance of Temple Israel in West Bloomfield Township, Michigan — one of the largest Reform synagogues in the United States, with approximately 12,000 members — armed with a rifle. What follows is based on official statements and media reports available within the first 24 hours of the attack; some details may be subject to revision as the investigation proceeds. According to those accounts, synagogue security engaged the attacker immediately; he was fatally shot inside the building after driving through a hallway. A security officer was knocked unconscious by the vehicle and hospitalized; dozens of first responders were treated for smoke inhalation after the vehicle caught fire. All students and staff at the synagogue’s early childhood center were reported evacuated safely. The FBI designated the attack a “targeted act of violence against the Jewish community” and took over the investigation within hours.

The attack did not occur in a vacuum. It unfolded against a backdrop of sharply rising antisemitic incidents across the United States — and as part of a broader, accelerating global pattern of violence against Jewish institutions. Temple Israel itself had received active shooter preparedness training from the FBI’s Detroit field office as recently as January 2026, a detail that is not incidental: it reflects the degree to which American Jewish communities have been compelled to treat armed attack as an operational contingency rather than a remote risk. That sense of operational urgency extended to law enforcement. Oakland County Sheriff Michael Bouchard stated at the post-attack press conference that local and national law enforcement had been in active communication for approximately two weeks before the attack, sharing threat intelligence and specifically discussing the potential for exactly this kind of incident. “There was no lack of preparation,” he said. The attack was anticipated — and it happened anyway. The Tree of Life Congregation in Pittsburgh — site of the deadliest antisemitic attack in US history in 2018 — issued a statement of solidarity in the immediate aftermath, noting that the Michigan attack had reawakened feelings of fear across the Jewish community. “We grieve for a lost sense of security,” its CEO wrote; “that we live in a world where Jews have to worship under the protection of armed guards; that preschoolers must be equipped to survive an active shooting situation.”

These words carry more than symbolic significance. They locate the Temple Israel attack within a sequence — Pittsburgh, Washington, D.C., Bondi Beach, now West Bloomfield — in which the intervals between mass antisemitic violence are not growing longer but shorter. This study does not attempt to explain that sequence. It examines one of its constitutive dimensions: the digital afterlife that unfolds in the hours after each event, in the comment sections of major news outlets, where the discursive response to antisemitic violence is formed before official narratives have settled and before grief has been given time to breathe.

This article draws on a dataset of 1,600 comments collected from eight major US news outlets’ YouTube channels within the first 24 hours of the attack. It analyzes how online publics responded to the Temple Israel attack — whether with empathy or antipathy, mourning or mockery, solidarity or conspiracy — and it documents the antisemitic discursive repertoires that emerged in that immediate window. The event is structurally distinct from the discourse events examined in prior work in this research line. The Washington Embassy shooting in May 2025, the Bondi Beach massacre in December 2025, and the YouTube reaction studies conducted under the Antisemitism in Real Time series all examined online responses to episodes in which the discursive positioning of Jewish actors was contested — where antisemitic discourse could hide behind geopolitical framings, inversion structures, or the ambiguities of armed conflict. The Temple Israel attack offers no such cover. A Jewish institution was attacked. The perpetrator is dead. The FBI has designated the event an act of antisemitic violence. The question is not whether the attack was antisemitic, but how the digital public responded to that fact.

That question is not rhetorical. The prior case studies give reason to expect that online comment sections will not simply reflect the dominant social consensus — that antisemitic violence against Jews is unambiguously wrong — but will fracture it, contest it, and in some cases invert it. The Washington Embassy analysis documented a near-total refusal of empathy in certain comment ecosystems, where Jewish grief was mocked as performance and the murder of two diplomats reframed as legitimate political consequence. The Bondi Beach analysis found that even under conditions maximally favorable to empathy — children present, a Holocaust survivor among the dead, an unambiguous terror attack on a Western democracy’s soil — antisemitic discourse achieved significant visibility, comprising between 5% and 17% of analyzed comments depending on platform. The present study tests whether the Temple Israel attack, with its distinctive features — a foiled attack, no Jewish fatalities among congregants, a named and identified perpetrator — produces comparable, diminished, or intensified antisemitic responses.

The perpetrator’s background introduces a further variable. Ghazali was born in Lebanon and became a US citizen in 2016. In the hours following the attack, before motive had been officially established, this detail became a significant flashpoint in comment sections — activating not only antisemitic conspiracy narratives but also Islamophobic framings, immigration discourse, and counter-accusations that shaped the discursive environment in ways that distinguish this case from prior episodes. The interaction between antisemitic and adjacent hate discourses is a central analytical concern of this study: not because Islamophobia and antisemitism are equivalent, but because they circulate in the same comment ecosystems and because understanding how they interact — whether they reinforce, compete with, or displace one another — is necessary for a complete account of what these spaces produce.

The findings are analyzed using the Decoding Antisemitism Lexicon and draw on the methodological framework developed across the Washington, Bondi, and Antisemitism in Real Time studies. They are presented here as a first analytical pass, intended to document the immediate discursive response and to identify the dominant repertoires before the news cycle moves on. A fuller report will follow.

Dataset and Methodology

The dataset comprises 1,600 user comments collected from YouTube videos posted by eight major US news outlets in response to the Temple Israel attack on March 12, 2026. Comments were drawn from the first video published by each outlet’s official YouTube channel covering the event, collected within the first 24 hours following publication. The sample size is 200 comments per outlet.

The outlets, their primary URLs, and their positioning within the mainstream US media landscape are as follows:

Several outlets published multiple videos covering the attack; for methodological consistency, only the first published video per outlet was analyzed. PBS NewsHour was excluded as comments were disabled; NPR was excluded due to insufficient comment volume; Breaking Points (Krystal Ball & Saagar Enjeti) published no coverage of the event.

Each comment was analyzed qualitatively using linguistic and multimodal close reading, supplemented by sequence and relational analysis where comments form part of reply chains. The analytical framework is guided by the Decoding Antisemitism Lexicon (Becker et al., Springer Nature, 2024), a research-based instrument for identifying explicit and implicit antisemitic discourse in digital environments. The Lexicon accounts for dog whistles, sarcasm, irony, visual and emoji-based signaling, and inversion structures, enabling systematic identification of hostile speech that evades surface-level keyword filtering. Quantitative coding of category frequencies is supplemented by qualitative close reading of representative examples. Intercoder reliability procedures follow the protocol established in prior studies in this research line.

The main analytical categories applied are:

1. Conspiracy narratives (antisemitic and non-antisemitic)

2. Power and Subordination narratives

3. Exploitation tropes

4. Instrumentalization of Antisemitism

5. Privilege narratives

6. Schadenfreude and Sarcasm

7. Blame, Guilt, and Inversion of Victimhood

Two further formations appear in the corpus but are treated as outlet-level analytical notes rather than standalone categories given their limited distribution: Jewish orchestration of social and demographic division (NBC, AP) and Nazi discourse/Holocaust denial (Forbes only). These are addressed in the outlet-level notes below.

A note on the Schadenfreude category. Prior studies in this research line — particularly Part II of the Antisemitism in Real Time series (Iranian retaliation, June 2025) and the Charlie Kirk assassination analysis (September 2025) — documented explicit Death Wishes and Glorification of Violence as the dominant affective register when Jewish actors or figures were killed or targeted at mass-casualty scale. The Temple Israel attack presents a structurally different outcome: no Jewish congregants were killed, all children were evacuated, and the perpetrator died. This outcome structure conditions a different affective repertoire. What the preliminary data show is not Death Wishes or Glorification of Violence in the register documented in prior studies, but rather Schadenfreude — satisfaction at the event expressed through sarcasm, irony, and mockery — alongside cynical detachment. The category is renamed accordingly for this study.

A note on cross-study comparability. This study applies the same annotation protocol and counting unit as the Washington Embassy, Bondi Beach, and Antisemitism in Real Time analyses. Aggregate antisemitic content rates across studies should nonetheless be interpreted with awareness of event-type conditioning: the discourse event structure — domestic antisemitic attack versus geopolitical military escalation versus mass-casualty terrorist attack — shapes which repertoires are available and plausible for commenters to deploy, and therefore conditions both the volume and the register of antisemitic response. These structural considerations are addressed in the analytical sections below.

Results

The following section summarizes outlet-level patterns before examining the dominant antisemitic formations across the dataset.

Summary of Comment Patterns by Outlet

MS NOW (28 hits, ~14%): The lowest hit rate among the left-leaning outlets, consistent with patterns documented in the AiRT Part III dataset, where MS NOW / MSNBC consistently produced lower antisemitic content rates than CNN. The dominant categories are Conspiracy narratives and Privilege arguments. The register is less overtly hostile than CNN but follows the same structural logics of false flag, subordination, and goy discourse.

CNN (88 hits, ~44%): The highest antisemitic content rate in the dataset — substantially above the dataset mean and higher than CNN’s rate in Part III (21%). The repertoire is the broadest of any outlet: Conspiracy narratives, Power and Subordination, Instrumentalization of Antisemitism, Privilege arguments, Blame and Guilt, dehumanization, and Schadenfreude all appear within the sample. Several comments display structural complexity that warrants close analytical attention.

NBC News (68 hits, ~34%): The second-highest rate in the dataset. Conspiracy narratives — particularly false flag and Mossad orchestration claims — are the dominant formation, constituting the majority of hits. A secondary cluster involves Power narratives and Instrumentalization of Antisemitism. The overall register is conspiratorial rather than overtly celebratory.

CBS News (56 hits, ~28%): Conspiracy narratives dominate to a degree unmatched elsewhere in the centrist outlets: the large majority of hits fall within this category. The defining feature is the Epstein thread — a formation documented as cross-platform in AiRT Part III that here re-emerges with striking intensity in response to a domestic antisemitic attack, adapted from its Part III distraction function into a framework for interpreting the attack itself as a Jewish operation.

ABC News (44 hits, ~22%): Conspiracy narratives again dominant, with a secondary cluster of Exploitation tropes anchored by a near-verbatim reproduction of the CBS Epstein comment across a separate outlet — confirming the Epstein thread’s cross-platform circulation as a recurring narrative template. The register tends toward sustained conspiratorial argument rather than compressed sarcasm.

Associated Press (12 hits, ~6%): The lowest hit rate among centrist outlets. Hits are concentrated in Conspiracy narratives, with one instance of a Jewish orchestration of social division trope. The low volume may reflect the AP comment ecosystem’s comparatively smaller and less algorithmically amplified comment base.

Forbes Breaking News (32 hits, ~16%): The most ideologically extreme content in the dataset at the individual comment level, including the only instances of explicit Nazi citation and Holocaust denial in the corpus. Conspiracy narratives dominate, with notable instances of Blame and Guilt, Schadenfreude, and Privilege arguments.

Fox News (4 hits, ~2%): The lowest antisemitic content rate in the dataset by a significant margin — a finding that is structurally counterintuitive given Fox News’ 20% rate in AiRT Part III. The most likely explanation is event-type conditioning: for a right-leaning audience, the Temple Israel attack activates a discursive repertoire centered on immigration, Islamist terrorism, and law enforcement rather than the Jewish-control-of-US-policy frame that dominated in the co-belligerence context of Part III. The false flag frame — sole antisemitic formation in the Fox sample — persists nonetheless.

Breaking Down the Predominant Concepts of Antisemitism in YouTube Responses

1. Conspiracy Narratives

Conspiracy narratives are by far the dominant antisemitic formation across the dataset, appearing at every outlet and constituting the majority of hits at six of the eight. The dominant sub-type is the false flag claim: the Temple Israel attack was not an antisemitic act of violence but a staged event, orchestrated by Israeli intelligence (Mossad), designed to manufacture Jewish victimhood and generate political sympathy for Israel. False flag claims range from compressed, unelaborated assertions to fully theorized conspiratorial architectures.

The most minimal instances perform the conspiracy claim as self-evident, requiring no argument:

• “False flag” (NBC)

• “Fake” (NBC)

• “B TO THE S 🙄” (NBC)

• “Mossad false flag operation” (NBC and CNN)

• “FALSE FLAGS MANIPULATION DONT WORK ANYMORE .... STOP WITH THE BULLSHT” (NBC)

• “Another psyop by Israel.” (NBC)

• “This was obviously a false flag operation conducted by Israel you moron!” (ABC)

• “Not really. Prob. self inflicted to cause some shit.” (Forbes)

The aggression in the ABC comment — “you moron!” — is notable: it performs the conspiratorial claim as self-evident truth that only the deliberately ignorant could contest, rendering skepticism itself a form of complicity.

A second cluster deploys ironic coding and wordplay to assert the conspiracy while maintaining deniability:

• “I bet he left his Iranian passport in the car 😂😂😂😂😂😂 Mossad that you?” (CNN)

• “Oy vey, what a cohencidence” (Forbes)

• “Probably a new Gun Cohentrol Hoax” (Forbes)

• “A: Same old trick / B: My thinking exactly! False Flag incoming!” (Forbes)

• “The first of more to come… Begins with M and ends in AD… begins with F and ends in AG 🧿” (CBS)

• “Smell like Israeli Mossad, well well well” (AP)

• “Sloppy sympathy stunt by Mossad” (Fox)

The Iranian passport comment references a recurrent trope — Mossad operations staged with conveniently placed identifying documents pointing to an Arab or Iranian perpetrator — combined with a laughing emoji cluster that performs gleeful irony. “Cohencidence” — a portmanteau of “Cohen” (a common Jewish surname) and “coincidence” — signals that the event is a Jewish-orchestrated non-coincidence; the wordplay allows the claim to circulate with plausible deniability while remaining legible to audiences familiar with the code. The same substitution appears in “Gun Cohentrol Hoax” — Cohen + control, again substituting a Jewish surname to imply Jewish agency, this time behind gun control advocacy — compressing three conspiracy claims into four words: the attack is staged, Jews orchestrate gun control, and both are part of the same operation. The CBS comment spells out “Mossad” and “False Flag” through initial and terminal letter cues, with the blue eye (nazar) emoji deployed with ironic inversion. “Sloppy” at Fox implies the alleged Mossad operation was poorly executed — a variant of the “running out of public sympathy” formulation documented at CNN, positioning the conspiracy as a failing or amateurish operation.

A third cluster elaborates full conspiratorial architectures — theorizing the mechanism and motive behind the alleged staging:

• “Mossad false flag opration. They are running out of public sympathy they always do these type of stunt... Investigate synagogue.....!!!” (CNN)

• “Exactly. False flag attack trying to create sympathy for Israel.” (CNN)

• “Oh please nobody is buying this shit Israel” (CNN)

• “CIA/Mossad. They are manufacturing consent. Or attempting to.” (NBC)

• “They have done this before you know... Mossad are notorious for pulling sh*t like this to increase support for Israel.” (CBS)

• “False Flag. Satanyahu and Trump are desperate.” (MS NOW)

• “Mossad taking out their own… I hate those people” (Forbes)

• “Zionists shooting up Jewish temples” (Forbes)

The CNN comment’s claim that Israel is “running out of public sympathy” positions antisemitic violence as a renewable resource deliberately manufactured to replenish goodwill. “They have done this before” at CBS performs normalization — Mossad’s alleged conduct is framed as established historical record rather than speculation; “notorious” presents the extraordinary claim as common knowledge. “Satanyahu” — a portmanteau of Satan and Netanyahu — is among the most widespread demonizing compounds in digital antisemitic discourse, condensing theological vilification and political delegitimization into a single word; its deployment within a false flag claim performs a double function, casting the conspiracy as orchestrated by a literally satanic figure. “I hate those people” at Forbes is a rare instance where the conspiracy claim is accompanied by explicit ethnic hatred rather than the ironic detachment typical of many other comments.

The most theoretically elaborated false flag claims construct full causal chains:

• “u believe in the greater Israel project?? Creating antisemitism to get 🧃 to move to Israel n bring the messianic age? All done by mossad” (NBC)

• “just look at the motivation of Israel, what are they trying to accomplish, the greater Israel project, in order for them to accomplish that they need antisemitism to be on the rise. We are not paranoid. You are just blind. That is the only logical explanation.” (ABC)

• “@JohnKoenig-s6d need to learn how Zionism is shaping what we’re seeing today. They’ve been doing it for over a century now and you’re too compliant to see it.” (ABC)

The NBC comment condenses several formations — false flag, Mossad orchestration, Jewish demographic manipulation, eschatological conspiracy — into a single post. The juice emoji (🧃) functions as a derogatory coded substitution for “Jews,” a phonetic dog whistle common in moderated comment sections. The ABC comment constructs a causal chain (Greater Israel project → need for rising antisemitism → manufactured attacks) and presents it not as speculation but as “the only logical explanation” — performing epistemic closure, positioning the conspiratorial framework as the rational default and skepticism as complicity. The third ABC comment adds a temporal dimension — “over a century now” — situating alleged Zionist orchestration within a longstanding historical pattern of Jewish long-range planning.

The Epstein Thread

A distinct sub-pattern within Conspiracy narratives — and the most significant cross-platform formation in the dataset — is the Epstein thread: the claim that Israeli intelligence orchestrated or was complicit in Jeffrey Epstein’s sexual exploitation network, and that the Temple Israel attack is either a false flag to distract from the Epstein files or direct evidence of Israeli operational methods. The thread was documented as a near-ubiquitous cross-platform motif in AiRT Part III, where it functioned primarily as a distraction narrative in the context of Operation Epic Fury. Its appearance here — in response to a domestic antisemitic attack on a Jewish institution — represents a structural adaptation: the Epstein conspiracy is no longer a distraction from geopolitical violence but a framework for interpreting antisemitic violence itself as Jewish-orchestrated. The most elaborated instance is constructed as a prosecutorial brief:

• “Denial in the face of evidence is a characteristic of complicity: ‘Make sure to tell them that I am not Mossad :)’ - Jeffrey Epstein in a personal email. ‘Epstein is a Mossad asset’ - FBI investigation released in the Epstein files. Ehud Barak and Mossad agents stayed for weeks at a time in Epstein’s apartment, according to evidence released by the files. Robert Maxwell, an Israeli superspy, is father to Ghislaine Maxwell. He is buried on the Mount of Olives, and his funeral was attended by six Mossad agents and he was eulogized by the Israeli Prime Minister and the Israeli President. ‘Let the g0Y1m live in the real world’ - Jeffrey Epstein in a personal email. Jeffrey Epstein is featured in photos wearing an IDF sweater. Financial accounts document Jeffrey Epstein’s many contributions to Friends of the IDF.” (CBS)

This comment assembles attributed quotations, biographical facts, and documentary claims into an evidentiary chain, performing rigorous sourcing while constructing a totalized narrative in which Israeli intelligence, Jewish elite power, and sexual exploitation of children are inseparable. The Epstein email quote — “Make sure to tell them that I am not Mossad :)” — is presented as evidence of Mossad affiliation, when it could equally be read as a joke or deflection. The phrase “Let the g0Y1m live in the real world” — using character-substitution filter-evasion — activates the master/servant hierarchy of the goy trope. The overall structure exemplifies documentary antisemitism: the assembly of real, distorted, and fabricated elements into an apparently rigorous evidentiary presentation that makes the conspiracy appear empirically grounded.

The thread recurs at ABC in near-verbatim form:

• “Friendly reminder that the Israeli government provided security for Jeffrey Epstein’s multi-million dollar apartment in New York City. Evidence published by Drop Site News.” (ABC)

• “Friendly reminder that the Israeli government provided security for Jeffrey Epstein’s multi-million dollar apartment in New York City. This included cameras, access and monitoring of the rooms where girls were held. Evidence published by Drop Site News.” (CBS)

The near-identical phrasing across two separate outlet comment sections confirms the Epstein thread’s character as a recurring cross-platform narrative template — the same text, or near-identical text, circulating across multiple comment ecosystems.

The thread is extended at ABC through a chain of alleged Israeli false flag operations:

• “Bingo. You’re exactly right bro. Remember Israel attacked their own embassy in the UK too. Israel/Zionism is the greatest enemy humanity has ever known. End the Jou Coup!” (ABC)

• “Lavon affair, Hotel David Bombing, USS Liberty, WTC, we know your fruits” (ABC)

The second comment lists the Lavon Affair (1954), the King David Hotel bombing (1946) — rendered by the commenter as 'Hotel David Bombing,' inverting the name in a pattern consistent with the broader encoding strategies documented in this corpus — the USS Liberty incident (1967), and the September 11 attacks as evidence of a consistent Israeli pattern of false flag violence. “We know your fruits” — a biblical phrase (Matthew 7:16-20, “by their fruits ye shall know them”) — applies scriptural authority to the conspiratorial claim. The list-making strategy performs encyclopedic authority while condensing disparate historical events — some involving genuine Israeli actions, some contested, one a widely debunked conspiracy theory — into a single unified narrative of Jewish orchestration.

2. Power and Subordination Narratives

The Power and Subordination frame — the claim that Jewish actors exercise commanding authority over American political life, and that the US state is structurally subordinate to Jewish or Israeli interests — is the second most prevalent formation in the dataset and the one that most directly connects this study to the trajectory documented in AiRT Part III. Where Part III found the subordination frame dominant across all eight outlets in the context of US co-belligerence, its appearance here — in response to a domestic antisemitic attack, not a US–Israeli military operation — confirms that the frame has become sufficiently entrenched to operate independently of the geopolitical scaffold that gave it apparent factual grounding in 2026.

The most compressed Power formulations perform the subordination claim at the level of national identity:

• “Owner of the USA – Israel.” (CNN)

• “sure just what we need another president that bows down to Israel for everything” (NBC)

• “shut up you goy and serve your masters” (CNN)

“Owner of the USA – Israel” performs the subordination frame in its most minimal form: not asserting that a specific president defers to Israeli interests, but that the United States itself is Israeli property. Like “United States of Israel” from the Part III Forbes dataset, it collapses the distinction between the two states entirely. “Bows down” retains a servility dimension without the explicit sexual humiliation register documented in Part III; its appearance in response to a synagogue attack is analytically notable — even when a Jewish institution has just been attacked, the Power narrative reasserts itself, redirecting attention from Jewish victimhood to alleged Jewish control.

A multi-turn exchange at CNN illustrates how the Power narrative is constructed dialogically:

• A: “That’s the most police I’ve ever seen at a crime scene”

• B: “Special treatments for chosen people. Don’t be jealous now.”

• C: “Someone mentioned Israel and they all came running”

• D: “They rule the world...” (CNN)

The exchange moves in four steps from an apparently neutral empirical observation to conspiratorial conclusion. The “chosen people” reference activates the theological dimension of the Power trope — Jewish claims to divine election recast as a mechanism of political privilege. The ellipsis following “They rule the world” performs world-weary resignation, as though the conclusion requires no elaboration.

The “Puppetmaster” formation from NBC encodes the subordination logic through a concealment metaphor:

• “You honestly thought a live stream genocide and Netanyahu as Puppetmaster […] pulling the strings of the American war machine wouldn’t have consequences” (NBC)

This comment simultaneously functions as political causation argument and antisemitic claim — implying that the Temple Israel attack was a foreseeable consequence of Jewish orchestration, thereby attributing a degree of responsibility to the Jewish community itself.

3. Exploitation Tropes

Exploitation tropes position Jews or Jewish institutions as extracting resources — material, emotional, or political — from non-Jewish populations for instrumental benefit. In this dataset the category operates primarily through the ‘Mass Sympathy Collection Card’ at CNN, alongside the Epstein thread addressed above under Conspiracy narratives, where its exploitation dimension — the framing of Israeli intelligence as extracting sexual and political capital from non-Jewish victims — is analyzed in full.

• “Here are 20 sharp, precise terms for the Mass Sympathy Collection Card: Victimhood Harvesting / Sympathy Mining / Grief Extraction / Manufactured Martyrdom / Outrage Farming / The Pity Racket / Suffering on Subscription / The Grievance Dividend / Tears as Currency / The Martyr Franchise / Bleeding for Profit / The Wound Economy / Sympathy Laundering / The Victimhood Cartel / Emotional Counterfeiting / Instrumental Victimhood / Performative Persecution / Grievance Arbitrage / Synthetic Suffering / The Persecution Dividend” (CNN)

This comment generates a taxonomy of twenty terms for alleged Jewish victimhood manufacturing, each combining economic or commercial language (harvesting, mining, extraction, laundering, arbitrage, dividend) with the vocabulary of suffering and persecution. The economic register activates the classical financial trope — the association of Jews with extraction and instrumental manipulation — applied specifically to the domain of suffering: Jewish grief is recast as a commodity to be produced, circulated, and monetized. The list format performs pseudo-analytical authority, mimicking intellectual categorization while deploying it in the service of dehumanization. It also illustrates a distinctive feature of digital antisemitism’s platform adaptation: elaborate, apparently neutral taxonomic language that evades keyword-based moderation filters.

• “see cruisers parked outside Jewish houses in fl. Money and whining works” (CNN)

“Money and whining” condenses two classical formations — Jewish financial power and Jewish complaint as manipulation — into a single causal claim: Jews secure preferential police protection through a combination of wealth and performative grievance.

4. Instrumentalization of Antisemitism

The Instrumentalization of Antisemitism category covers comments that frame antisemitism not as a real social phenomenon but as a rhetorical instrument — a concept deployed strategically by Jewish actors to generate sympathy, deflect criticism, or advance political agendas. Its appearance in response to an antisemitic attack on a synagogue is structurally significant: the attack itself is reframed as evidence of Jewish instrumentalization rather than Jewish victimhood.

• “Fake to push anti semitism narrative.” (NBC)

• “Look how much coverage the fake news gives to anything against the tiny hats compared to others.....” (NBC)

• “Oy Vey! Let’s have a national year of remembrance.” (NBC)

The first explicitly frames the attack as fabricated to advance an “antisemitism narrative” — positioning antisemitism as a rhetorical instrument rather than a real social phenomenon. The second introduces a media privilege argument via “tiny hats” — a derogatory coded reference to Jews through the kippah metonym — framing disproportionate media coverage of antisemitic violence as evidence of Jewish media control. “Oy Vey! Let’s have a national year of remembrance” deploys mock Jewish speech to ridicule the perceived disproportionality of Jewish responses to antisemitic violence: “Oy Vey” — typically an expression of grief — becomes a punchline, and the call for formal national commemoration performs the Instrumentalization frame in Schadenfreude register.

5. Privilege Narratives

Privilege narratives frame the institutional response to the Temple Israel attack — police presence, media coverage, political statements — as evidence of disproportionate Jewish power over state resources, at the expense of ordinary (non-Jewish) Americans. The category is concentrated at CNN and MS NOW, with instances at Forbes.

• “Is every Patrol police car in the state at the synagogue 🕍” (CNN)

• “Goyims need the same level of security.” (CNN)

• “How well they patrol our streets, they cater to those greedy roaches but when an elderly woman needs help, it takes half to one hour to rescue her.” (CNN)

• “@gaiazala3386 wel,.. whut wee no fur faKd iz that they wanuh giv that imij that mor reezoRsiz wouBbee alokeydİd to kuhvur thu joooW an synuhgagz, than redily iveyluhbuhl fur difrint uhmerikuhn[baring tUmp an& hiz sdeyt krOwneez, yes-(wOoo)men🦝🤏🐽” (CNN)

• “Mmmmmmmhhhhh. Look at the thumb counts. They think we’re stupid. Only we goy do the dying” (MS NOW)

• “Good g/0:ym” (MS NOW)

• “wish police would come out like this for Americans.” (Forbes)

The CNN comments construct a privilege argument through the lens of policing resource allocation: the mass police response to the synagogue attack is presented as evidence that Jewish safety receives preferential state protection while ordinary Americans are left unprotected. The third comment escalates from the privilege argument to explicit dehumanization — “greedy roaches” — embedding the parasite/vermin metaphor, one of the most historically charged tropes in antisemitic discourse, within an apparently civic argument about fairness. The fourth comment encodes its message in phonetic misspelling so severe that content moderation cannot process it, while remaining readable with minimal decoding effort: “what we know for fact is that they want to give the image that more resources would be allocated to cover the Jews and synagogues than readily available for different Americans.” The derogatory emojis — raccoon (🦝, a common coded slur for Black Americans), pinching hand (🤏), pig snout (🐽, a longstanding antisemitic dehumanizer) — layer racialized contempt onto the privilege argument. The phonetic encoding is not merely technical evasion but a performance of community membership, signaling familiarity with a register of digital hate speech that operates below the threshold of automated detection.

The MS NOW comments frame the privilege argument through the goy discourse register. “Only we goy do the dying” inverts the victimhood of the Temple Israel attack entirely: in this framing, it is non-Jews who are the true victims, bearing the costs of violence while Jewish interests are protected. “Good g/0:ym” is an ironic inverted salute — praising the “good goy” as a term of contempt for non-Jews who serve Jewish interests. The deliberate misspelling “g/0:ym” uses character substitution to evade content filters while simultaneously signaling community membership in spaces where such coded formulations circulate.

6. Schadenfreude and Sarcasm

As noted in the Methodology section, prior studies in this research line documented Death Wishes and Glorification of Violence as the dominant affective register when Jewish actors were killed or targeted at mass-casualty scale. The Temple Israel attack — foiled, no Jewish congregant fatalities, perpetrator dead — conditions a different repertoire. What the data show is Schadenfreude: satisfaction at the event expressed through sarcasm, irony, and darkly comic wordplay, rather than explicit celebration of death.

• “Remember a couple of elbows to the temple and they won’t remember their names 😮😅😊” (CNN)

• “Oy Vey! Let’s have a national year of remembrance.” (NBC)

• “Oy vey, what a cohencidence” (Forbes)

The CNN comment deploys a violent pun — “elbows to the temple” simultaneously referencing the anatomical temple and Temple Israel — to express satisfaction through wordplay. The emoji sequence (shocked face, laughing-crying face, smiling face) performs escalating amusement. The construction illustrates how Schadenfreude in digital antisemitism operates through humor as affective cover: violence is rendered entertaining and signaled through the emoji register. The NBC comment performs Schadenfreude through mock commemoration (this comment also appears under Instrumentalization of Antisemitism above, where its primary function as a vehicle for ridiculing Jewish grievance is analyzed; it is cited here for its affective register). The Forbes comment performs it through the “cohencidence” wordplay — the attack is not mourned but savored as a knowing joke about Jewish orchestration. (”Cohen,” a common Jewish surname, substitutes for “coin-” to signal Jewish agency while maintaining ironic deniability; see also “Gun Cohentrol Hoax” under Conspiracy narratives above.)

7. Blame, Guilt, and Inversion of Victimhood

The Blame and Guilt category covers comments that attribute collective responsibility for the Temple Israel attack to the Jewish community itself — framing the attack as a foreseeable or deserved consequence of Jewish or Israeli conduct. The category is concentrated at CNN, with instances at MS NOW and Forbes.

• “Wow! They almost got a taste of having their own children murdered!” (CNN)

• “Isreal is begging to be targeted with their actions. I feel bad for the children, but as a whole I say fk Isreal.” (CNN)

• “and wb those poor poor Jews? Did they do anything to deserve being disliked or we all just… dun dun dun… antisemitic” (CNN)

• “Of course i don’t support this. 99+% don’t. However. It’s to be expected after murdering a school full of girls...” (CNN)

• “You honestly thought a live stream genocide and Netanyahu as Puppetmaster […] pulling the strings of the American war machine wouldn’t have consequences” (NBC)

• “No one is more antisemitic than the Jews” (MS NOW)

• “A: I wonder why people would hate them / B: Must be a cohencindence” (Forbes)

The first CNN comment frames the near-miss outcome as a missed opportunity for collective punishment: Jewish children’s safety is treated as a deprivation of deserved consequences. The second makes the collective guilt logic explicit — Israel’s actions make Jews “beg to be targeted” — while the partial concession (”I feel bad for the children”) intensifies rather than mitigates the hostility, carving out children as the sole exception to an endorsement of antisemitic violence. The third performs the blame/guilt trope through sarcastic inversion: antisemitism is reframed as rational response to Jewish behavior, with “dun dun dun” rendering the category itself absurd. The fourth wraps the blame attribution in a hedge — “Of course I don’t support this” — before delivering the causal claim: the attack was “to be expected” given Israeli military conduct. The NBC “Puppetmaster” comment, addressed under Power narratives, also functions here: by casting the attack as a consequence of Jewish orchestration of US foreign policy, it attributes a degree of responsibility for the attack to Jewish actors themselves.

“No one is more antisemitic than the Jews” performs blame through self-referential inversion: antisemitism is recast not as a prejudice directed at Jews from outside but as a pathology internal to Jewish identity itself, simultaneously denying its existence as an external social phenomenon and attributing collective moral failure to Jews as a group. The Forbes exchange — “I wonder why people would hate them / Must be a cohencindence” — naturalizes antisemitism as a predictable response to Jewish conduct, framing hatred as an effect rather than a prejudice (on the 'Cohen' substitution pattern, see Conspiracy Narratives above).

Outlet-Level Notes

Jewish Orchestration of Social and Demographic Division (NBC, AP)

A formation documented across two ideologically distinct outlets involves the claim that Jewish actors deliberately engineer social fragmentation and demographic disruption among non-Jewish populations:

• “It’s just so funny to me how the jews here want open borders, but for some strange reason the same jews don’t want that for israel. It’s almost as if unchecked immigration is a weapon being used against us” (NBC)

• “We’re being divided by Israel” (AP)

The NBC comment specifies the mechanism — open borders policy as a weapon of demographic destabilization — while the AP comment is more compressed, a bare assertion of Israeli agency over American social cohesion. Together, across outlets separated by political positioning, they confirm the presence of this trope in the dataset.

Nazi Discourse and Holocaust Denial (Forbes)

Two comments in the Forbes sample represent the most historically extreme content in the corpus:

• “The Jews are our misfortune” (Forbes)

• “Was it 6 million?” (Forbes)

“The Jews are our misfortune” — “Die Juden sind unser Unglück” — was coined by the German historian Heinrich von Treitschke in 1879, adopted as a Nazi movement slogan, and appeared as the closing line of every issue of Der Stürmer. Its appearance here without modification or irony represents direct citation of Nazi ideological language, offered as self-sufficient statement requiring no elaboration. “Was it 6 million?” performs Holocaust denial through interrogative insinuation — introducing doubt about the most documented genocide in modern history as a response to an antisemitic attack on a synagogue. Its appearance in this context is structurally characteristic of how antisemitic discourse uses events of Jewish victimhood as occasions for amplifying rather than contesting antisemitic formations.

Cross-Study Comparison: Washington, Bondi Beach, Temple Israel

This study is the third in a sequence of analyses examining antisemitic digital discourse following mass antisemitic violence. The Washington Embassy shooting (May 2025) and the Bondi Beach massacre (December 2025) provide the closest structural comparators: both involved lethal antisemitic attacks, both generated substantial YouTube comment corpora analyzed under the same methodological framework, and both documented the antisemitic discursive repertoires that emerged in the immediate aftermath. Placing Temple Israel alongside these two cases reveals both persistent cross-case patterns and event-specific divergences that illuminate how online antisemitism adapts to different attack structures.

Antisemitic content rates

The most immediately striking finding is the Temple Israel rate. The Washington analysis documented rates ranging from 24% (LiveNow from Fox) to 66% (Al Jazeera English) depending on outlet — a range reflecting both ideological positioning and the specific character of those comment ecosystems. The Bondi Beach analysis, drawing on five major American outlets, found rates between 5% (NBC) and 17% (CNN), with a mean substantially lower than Washington, attributed to event-type conditioning: the visual framing of families and children, the unambiguous Western-democratic setting, and the absence of geopolitical cover made the standard antisemitic inversions harder to deploy. The Temple Israel dataset, drawn from eight US outlets, produces a mean of approximately 17% — higher than Bondi’s mean but below Washington’s. The CNN figure of 44% is among the most striking outlet-level findings across the three studies, substantially exceeding CNN’s Bondi rate (17%) and suggesting that the domestic synagogue attack context does not suppress antisemitic discourse in left-leaning comment ecosystems in the way that victim framing at Bondi partially did.

Fox News presents the inverse anomaly: 2% at Temple Israel against 10% at Bondi and 24% at Washington (LiveNow from Fox, a comparable right-leaning outlet). The event-type conditioning explanation proposed for Fox in the outlet summary above — the Temple Israel attack activating an immigration/Islamist terrorism discursive repertoire that displaces rather than amplifies antisemitic conspiracy discourse — is consistent with the cross-study pattern: Fox’s antisemitic content rates are structurally responsive to whether the geopolitical context supplies a Jewish-control-of-US-policy scaffold, as it did in Washington (and in the AiRT Part III Operation Epic Fury dataset at 20%), and does not at Temple Israel.

Dominant categories

Conspiracy narratives — and specifically false flag and Mossad orchestration claims — are the single most consistent formation across all three studies. At Washington, they appeared under the heading of “Conspiracy Myths” and included explicit Mossad playbook and CIA false flag attributions. At Bondi, they included Mossad false flag claims and broader Jewish global control narratives. At Temple Israel, they constitute the dominant category across all eight outlets, with the false flag sub-type the most prevalent individual formation in the corpus. The persistence of the false flag response across all three studies is analytically significant: regardless of the actual character of the attack — whether it is ambiguous (Washington), unambiguous terrorism in a Western democratic setting (Bondi), or a foiled domestic attack designated antisemitic by the FBI (Temple Israel) — a significant proportion of commenters immediately recast Jewish victimhood as Jewish orchestration.

The Epstein thread, documented as cross-platform in AiRT Part III (Operation Epic Fury), emerges at Temple Israel as a domestic-violence adaptation — no longer a distraction from geopolitical military operations but a framework for interpreting an antisemitic attack on a synagogue as itself a Mossad operation. Its cross-outlet appearance at both CBS and ABC in near-verbatim form confirms the Epstein thread’s cross-platform circulation as a recurring narrative template.

Shifts and divergences

The most analytically significant divergence across the three studies concerns the Blame, Guilt, and Inversion of Victimhood category. At Washington and Bondi, inversion of victimhood was among the most prominent categories — commenters framed the murders as foreseeable consequences of Israeli conduct in Gaza, deployed casualty comparisons to delegitimize Jewish grief, and positioned the perpetrators as morally justified actors. At Temple Israel, this category remains present (particularly at CNN) but operates in a structurally modified form: since the perpetrator is identified as Lebanese-born rather than Palestinian, and the geopolitical causation argument is less readily available, the blame attribution is more generic — antisemitism as rational response to Jewish conduct in general, rather than as specific retaliation for Gaza.

The Schadenfreude category also shows cross-study evolution. At Washington, the “Kvetching intensifies” formation documented explicit mockery of Jewish grief. At Bondi, schadenfreude took the form of laughing emojis, popcorn GIFs, and the macabre “It was a Hanukkah menorah lighting” wordplay. At Temple Israel — where no Jewish congregants were killed and the perpetrator died — schadenfreude is more muted and operates primarily through irony, sarcasm, and wordplay (”elbows to the temple”; “Oy Vey! Let’s have a national year of remembrance”) rather than through the explicit celebration of death documented at Washington and Bondi. This shift is consistent with the event-type conditioning argument: the affective repertoire available to commenters is conditioned by the outcome structure of the attack itself.

Structural continuity

What cuts across all three studies is the structural refusal of empathy: in each case, significant proportions of commenters respond to antisemitic violence against Jews not with mourning or solidarity but with conspiracy, blame, mockery, or celebration. The specific formations shift — theological demonization prominent at Washington, victim-framing effects visible at Bondi, conspiracy dominance and Fox anomaly at Temple Israel — but the underlying discursive logic is consistent. Antisemitic violence against Jews does not, in these comment ecosystems, generate the empathetic consensus it might be expected to produce. Instead, it activates a repertoire of inversions, distancing moves, and hostile framings that have been documented with increasing precision across this research line. The Temple Israel dataset extends and complicates that documentation — and suggests that the discursive infrastructure enabling the normalization of Jewish death in digital public discourse remains, across events and platforms, firmly in place.

This analysis forms part of a larger comparative study on antisemitic discourse in digital public space following major antisemitic incidents; a full research report will be published in due course.

AddressHate Research Scholar at NYU’s Center for the Study of Antisemitism | Lead, Decoding Antisemitism | Research Advisor, AddressHate | Editor-in-Chief, Digital Hate Review

Introduction

On February 28, 2026, Israel and the United States launched Operation Epic Fury, a joint military campaign targeting Iran’s nuclear infrastructure, missile and drone production facilities, IRGC command centers, and political leadership. Supreme Leader Ali Khamenei was killed in a strike on the Leadership House compound in Tehran; Iranian state media confirmed his death in the early hours of March 1. Iran declared 40 days of mourning and launched retaliatory missiles and drones against Israel, US military bases, and Gulf states across the region.

By March 7, Iranian authorities reported a death toll exceeding 1,300, a figure not independently verified. Iran launched over 500 ballistic and naval missiles and nearly 2,000 drones since February 28; hundreds of US–Israeli strikes targeted facilities across most of Iran’s provinces. The Strait of Hormuz was severely disrupted, with commercial traffic effectively suspended. Hezbollah launched cross-border fire from Lebanon. Among the incidents that generated the most sustained online commentary was a strike on a girls’ school in Minab: Iranian authorities reported roughly 150–175 fatalities; independent investigators have confirmed a mass-casualty event but have not certified a precise toll. Preliminary US inquiries suggest US forces were likely responsible; Israel stated it had no operations in the area. No official investigation has issued a final conclusion. The Minab strike became a significant focal point in the comment sections analyzed below.

This article is the third installment in the Antisemitism in Real Time series, which tracks how antisemitic discourse adapts to, and is activated by, breaking geopolitical events in YouTube comment sections beneath major U.S. news coverage. The first two parts (Part I; Part II), published in June 2025 following Israel’s preemptive strike on Iran and Iran’s subsequent retaliation, established the baseline discursive patterns: demonization, Holocaust inversion, conspiracy narratives, and the normalization of Jewish suffering. This installment examines how those patterns have evolved — and in several cases dramatically intensified — in response to an escalation of categorically greater violence, political complexity, and symbolic weight.

Discursive Shift Across Three Studies: A Category Trajectory

The decline in mean antisemitic content rate between the June 2025 studies (18.5%; 19.1%) and the February–March 2026 dataset (12.6%) is not attributable to a difference in method: all three studies apply the same annotation protocol. The structural explanation lies elsewhere — in the altered geopolitical context itself.

When Israel acted alone in June 2025, antisemitic discourse could project its full weight onto a single target. The dominant registers were theological and dehumanizing: demonology, deicide, Nazi analogy, Holocaust inversion. These formations require Israel as sole agent — an undiluted vessel for antisemitic projection. When the United States became a full co-belligerent in February–March 2026, that total demonization became structurally harder. Commenters cannot cast the operation as purely Jewish evil without simultaneously implicating the US military, American foreign policy, and their own country. The rhetorical problem this creates is resolved by a shift in register: from Israel is demonic to Israel made America do it. Demonology gives way to conspiracy; theological vilification gives way to the subordination frame. The antisemitism does not diminish — it migrates into a register better suited to the political facts on the ground.

This trajectory is visible in the category data across all three studies:

* The subordination frame appears in incipient form in Parts I and II — as a byproduct of conspiracy claims rather than an independent organizing register. In Part I, comments such as “Congress is controlled by Mossad” and “Israel will drag the USA into this” imply US subordination without foregrounding it. In Part II, the CNN false-flag claim and the AP Khazar comment similarly cast the US as an instrument of Israeli interests. The explicit rhetorical vocabulary of personal submission — puppet, master, “good boy,” bodily subordination — plays a markedly less prominent role in both 2025 datasets than in Part III, where it constitutes an autonomous primary formation with its own consistent lexicon, operating independently of conspiracy elaboration rather than as a byproduct of it.

Several findings from this table warrant analytical emphasis. First, the theological register — deicide, Satan, synagogue of Satan, religious demonology — is prominent in Part I and entirely absent by Part III. Its disappearance is not incidental: these formations depend on Israel as a metaphysically singular evil. The presence of US co-belligerence disrupts that singularity. Second, the Nazi/Holocaust Analogy, which appeared in expansive and explicit form at CNN and ABC in Part I (”modern-day Hitler,” “doing to others what was done to them”), survives in Part III only in compressed, portable shorthand — “ZioNAZIs” as a single portmanteau — adapted to a context where the analogy can no longer bear the full weight it carried when Israel acted alone. Third, Power and Conspiracy narratives follow the inverse trajectory: emerging in Part I, minimal in Part II, dominant across all eight outlets in Part III. The Epstein thread — a single isolated comment at Fox News in Part II — consolidates into a near-ubiquitous cross-platform formation in 2026, appearing at all eight outlets under the portmanteau “Trumpstein.”

The implication is analytically significant: the rate drop between 2025 and 2026 should not be read as evidence of declining antisemitism in these comment ecosystems. It reflects, in part, a structural adaptation — antisemitic discourse reorganizing itself around the political facts of US co-belligerence, finding in the conspiracy and subordination register a more contextually plausible vehicle than the theological demonology that dominated when Israel acted alone.

Two further findings from this dataset warrant emphasis.

First, US military involvement has reshaped the antisemitic register in two related ways. Demonizing projections — the theological vilification, Holocaust analogies, and dehumanizing tropes that dominated the June 2025 datasets — are less prevalent in 2026: when the United States is a full co-belligerent, commenters cannot cast the operation as purely Jewish evil without implicating their own country and military. What fills that space is a marked increase in the directness of power and conspiracy allegations. Tropes that in 2025 were largely coded, insinuated, or metaphorically displaced now appear in unambiguous, unelaborated form. Commenters no longer hedge. When the geopolitical context already supplies the conspiratorial scaffold — a US president publicly committed to regime change alongside Israel — antisemitic interpretive frameworks require less rhetorical work to appear plausible. The political facts lower the threshold for explicit statement. Both shifts are documented empirically across all eight outlets and addressed in the analytical sections below.

Second, this is not a phenomenon confined to one end of the political spectrum. CNN — consistently associated with left-leaning audiences — and Fox News — consistently associated with right-leaning audiences — produce the two highest antisemitic response rates in this dataset: 21% and 20% respectively, separated by a single percentage point and both well above the dataset mean of 12.6%. The antisemitic content at Fox News’ YouTube comment section spans Power and Conspiracy narratives, demonological and dehumanizing tropes, and explicit Death Wishes — including the most extreme exterminatory rhetoric in the corpus. The data indicate that antisemitism in online comment culture is a societal problem that does not stop at the boundary of any political identity or partisan affiliation (see also my Substack article “The Platform Infrastructure Behind Generational Antisemitism”).

From June 2025 to February 2026: Continuity, Rupture, and Escalation

The June 2025 case studies established what might be called the rhetorical grammar of antisemitic online response to Israel–Iran conflict: a stable repertoire of inversions, projections, and dehumanizing tropes that was activated rapidly and at scale across ideologically diverse platforms. What distinguished those earlier episodes was the speed and internal coherence of the antisemitic response — Israel cast as aggressor, Iran as victim; Israeli suffering treated as deserved or dismissed entirely; conspiracy narratives framing Jewish actors as the hidden architects of regional instability. The June 2025 episodes thus function as discursive groundwork: they seed the conspiratorial interpretive frameworks — Jewish control of Western policy, Israel as permanent aggressor, US co-responsibility as proof of Jewish power — that commenters reactivate and intensify in their responses to the 2026 war.

The February–March 2026 escalation shares this discursive architecture but unfolds in a fundamentally altered context. Several factors distinguish this episode from the June 2025 escalation and are legible in the comment data.

Scale and US co-belligerence. The June 2025 strikes were Israeli operations. Operation Epic Fury is a joint US–Israeli campaign explicitly aimed at regime change — a goal stated publicly by President Trump on Truth Social within hours of the opening strikes. The direct involvement of the United States has materially shifted the discursive landscape: conspiracy narratives about Jewish control of American foreign policy, which in June 2025 largely appeared in coded or insinuated form, now operate in a context where the US is actively at war alongside Israel. Commenters no longer need to allege manipulation; they can point to observable political fact and allow antisemitic interpretive frameworks to supply the explanatory logic. The result is a significant migration of Power and Conspiracy tropes from latent to overt across all eight outlets analyzed.

The killing of Khamenei. The assassination of Iran’s Supreme Leader introduces a register of martyrology and symbolic grief that was absent in June 2025. Online, this has generated a distinct discursive response: the valorization of Iranian suffering as righteous, the framing of the strike as an act of desecration, and — at the antisemitic extreme — the reactivation of structures in which Jewish actors are cast as killers of a sacred or prophetic figure. Some comments draw explicitly on Christian eschatology to frame these events, echoing the deicide structures documented in Part I.

The Minab school strike. A strike on a school in Minab — in which Iranian authorities report roughly 150–175 people, mostly school-age girls, were killed, a toll not yet independently certified — rapidly became a focal point in online comment sections regardless of its evidentiary status. Current evidence and preliminary U.S. inquiries suggest U.S. forces were likely responsible, with Israel stating it had no operations in the area; no official investigation has yet issued a final public conclusion. For the purposes of this analysis, what matters is not the verified death toll but the discursive function the event assumed: it became an anchor for demonizing rhetoric, attributed deliberate intent to Israeli and American actors, and served as the primary referent for Nazi/Holocaust inversion, with commenters explicitly reaching for analogies to the targeting of Jewish children during the Shoah.

The explicit regime change framing. Trump’s public statement of regime change as a war aim, and Netanyahu’s direct address to the Iranian people in Farsi, give existing conspiracy tropes a factual scaffold. In June 2025, the characterization of Israeli actions as aimed at regional domination was presented by commenters as conspiratorial revelation. In February–March 2026, it is official US policy. This has the effect of normalizing the conspiratorial interpretive framework — and of intensifying its antisemitic valence when it migrates from political criticism to ethno-religious attribution.

Structural Shifts Between June 2025 and February–March 2026: Quantitative Overview

A third finding cuts across both observations and recurs consistently across all three studies in this series: CNN (left-leaning) and Fox News (right-leaning) produce the highest antisemitic response rates in every dataset. In Part III, CNN leads at 21% and Fox News follows at 20% — separated by a single percentage point, and together significantly above the dataset mean of 12.6%. This cross-ideological symmetry is not incidental. It reflects a structural feature of high-traffic, politically polarized comment ecosystems: antisemitic discourse does not map onto a single point on the political spectrum but is activated, in different registers and with different surface framings, at both ends of it. The finding should caution against any assumption that antisemitic comment culture is a phenomenon of the political right or left alone.

Two cross-cutting observations structure the dataset as a whole and recur across outlets regardless of political leaning. The dominant logic across the comment sections is one of Israeli control over the United States — expressed overtly through Power narratives (Israel commands, Trump obeys) or covertly through Conspiracy tropes (AIPAC, Mossad blackmail, political blackmail). This overt/covert axis is the organizing spine of antisemitic discourse in this dataset in a way that has no precise equivalent in the June 2025 case studies, where US involvement was not yet a fact. A second, related formation recurs across all eight outlets: commenters present Trump’s prosecution of the war as a deliberate distraction from the Epstein files, alleging that the conflict serves to suppress evidence linking Jewish elites to sexual exploitation and blackmail. This is frequently condensed into the portmanteau “Trumpstein.” The formation connects to the broader power-conspiracy frame — in both cases, Jewish actors are cast as the hidden authors of American political conduct — but its specific content (the Epstein network, file suppression, blackmail) is distinct enough to warrant separate identification.

Per-outlet antisemitic content rates across three case studies

Note: Forbes Breaking News was not included in the Part II sample. The lower Part III mean should not be read as a decline in antisemitic prevalence; all three studies apply the same annotation protocol. The structural explanation for the rate difference is discussed in the opening section.

Several observations are robust across all three studies. CNN and Fox News consistently produce the highest antisemitic content rates in their respective datasets, confirming a cross-ideological pattern in which the most-watched outlets at opposite ends of the political spectrum generate the most hostile comment environments. ABC News was the most antisemitic centrist outlet in Part I (30%) and Part II (22%), declining to 14% in Part III (28 hits in 200 comments), it remains the highest centrist outlet in the current study. NBC News showed an anomalous spike in Part II (28%), likely reflecting the specific framing of Iran’s retaliatory strike as morally justified; it returns to a comparatively low rate in Part III (6%). MSNBC / MS NOW is consistently among the lowest across all three studies, ranging from 6% to 16%.

Dominant concepts across the three case studies

The trajectory across the three studies is analytically significant. Part I, responding to an Israeli military operation, generated the broadest and most theologically grounded antisemitic repertoire: demonization, deicide, Holocaust inversion, and the orchestration of victimhood. Part II, responding to Iranian retaliation, pivoted sharply toward affective registers — schadenfreude, death wishes, glorification of violence — as the moral inversion calculus was resolved in Iran’s favor and Jewish suffering became a source of satisfaction rather than contested meaning. Part III, responding to a qualitatively different war involving US co-belligerence and explicit regime change objectives, consolidates Power and Conspiracy as the dominant frame: the structural conditions of the conflict have given conspiratorial logic a factual scaffold, enabling it to displace the more explicitly theological and dehumanizing formations that characterized Part I while absorbing the affective intensity documented in Part II.

The emergence of the Epstein thread as a cross-platform organizing motif represents one of the most significant new developments in this dataset. While an isolated precursor appeared in the Part II Fox News dataset — a single sprawling emoji-coded comment listing approximately ninety public figures — it was absent as a cross-platform formation from both June 2025 studies. Its consolidation here across the dataset — appearing in comment threads at all eight outlets, often under the shorthand ‘Trumpstein’ — marks a qualitative shift specific to the February–March 2026 context, in which the direct involvement of the Trump administration in a war alongside Israel has given pre-existing conspiratorial narratives about Epstein’s alleged ties to Israeli intelligence a new geopolitical anchor. The thread functions as a shorthand that links the current war to a broader narrative of Jewish elite conspiracy, political blackmail, and information suppression — and its cross-platform prevalence in 2026, where it was entirely absent in June 2025, is itself an analytically significant finding.

Dataset & Methodology

The dataset comprises 1,600 user comments collected from YouTube videos posted by eight major U.S. news outlets in response to the US–Israeli strikes on Iran beginning February 28, 2026. Videos were published on February 28–March 1, 2026 [URLs to be verified before publication]. Comments were drawn from top-level comment threads beneath breaking news and live coverage videos. The standard sample size is 200 comments per outlet. PBS NewsHour was excluded as comments were disabled; NPR was excluded due to insufficient comment volume (fewer than 100 comments on a single clip). The outlets span the ideological spectrum of mainstream US television and digital news:

• Left-Leaning/Liberal: CNN; MS NOW (formerly MSNBC)

• Centrist: ABC News; NBC News; CBS News; Associated Press

• Right-Leaning/Conservative: Fox News; Forbes Breaking News

Each comment was analyzed qualitatively using linguistic and multimodal close reading, supplemented by sequence and relational analysis. The analytical framework is guided by the Decoding Antisemitism Lexicon, a research-based instrument for identifying explicit and implicit antisemitic discourse in digital environments. The Lexicon accounts for dog whistles, sarcasm, irony, visual and emoji-based signaling, and inversion structures, enabling systematic identification of hostile speech that would evade surface-level keyword filtering.

A note on cross-study comparability. All three studies in this series apply the same annotation protocol and counting unit. The lower mean antisemitic content rate in Part III (12.6%) compared to Parts I (18.5%) and II (19.1%) therefore does not reflect a methodological difference. As argued in the opening section of this article, the rate difference is better explained by a structural shift in discursive register — from theological demonology and Holocaust inversion, which dominated when Israel acted alone, to Power and Conspiracy narratives, which have become primary in the context of US co-belligerence. The total dataset for Part III comprises 1,600 comments across eight outlets at 200 comments each. Within-study outlet comparisons and cross-study directional observations remain analytically robust.

Summary of Comment Patterns by Outlet

CNN (42 hits, 21%): The highest antisemitic hit rate in the dataset, consistent with the June 2025 findings. Power and Conspiracy narratives dominate (14 cases), followed by Demonology (12 cases) and Financial/Exploitation tropes (10 cases). The most structurally complex antisemitic comment in the entire dataset appears here: a theological conspiracy passage linking Orthodox Jewish religious authority to demonic political power, Third Temple conspiracism, and the “Synagogue of Satan.” Financial tropes are more prevalent at CNN than at any other outlet.

MS NOW (12 hits, 6%): Low hit rate, concentrated in Power and Conspiracy. Framing of Trump as Netanyahu’s instrument and characterization of the “Israel lobby” as having captured Congress and the billionaire donor class. Comments operate through political critique that slides into antisemitic attribution without overt markers.

ABC News (28 hits, 14%): A broad spectrum of antisemitic content, including Power and Conspiracy, Demonology, and financial and resource exploitation tropes. AIPAC references and the “sold us out for some silver” comment position Jewish actors as extracting blood and treasure from American citizens. The Mossad/Trump blackmail frame appears. One comment introduces the classical financial trope in its most explicit economic form (usury as Israel’s only export). Media control tropes also surface (”Disney is owned by Israel”).

NBC News (12 hits, 6%): Antisemitism surfaces almost exclusively through Power and Conspiracy tropes, often via implication rather than explicit statement. The Friday night timing of the strikes is used to insinuate Jewish orchestration; references to Israeli “supreme leaders” perform a rhetorical inversion of political hierarchy.

CBS News (14 hits, 7%): Low overall volume, but the cases present are analytically significant. Power and Conspiracy narratives dominate, with AIPAC invoked directly. The dataset includes a notable instance of visual antisemitic coding — the “imprint of his hat” comment — in which Jewish identity is read onto a political figure’s physical appearance to imply hidden allegiance. The Epstein thread makes a direct and maximally explicit appearance: “BREAKING NEWS: The Protocols of Zion are real.”

Associated Press (34 hits, 17%): The most diverse antisemitic repertoire in the dataset, spanning all five analytical categories and including the only clear Nazi Analogy in the corpus (”Again U$ joins ZioNAZIs and there are no rules”). Demonology is prominent — dehumanization (”Zionist rats”), cancer metaphors, and the “god’s chosen people you know” sarcasm. Death Wishes are explicit and extreme, including a call for nuclear annihilation (”I wish Israel to turn to glass”) and what appears to be a transliteration of an Arabic militant phrase calling for the destruction of Tel Aviv. The Khazarian Mafia conspiracy appears in concentrated form (”KHAZARIAN MAFIA BIG BO$$ BATTLE”), and “ZOG” (Zionist Occupation Government) appears as a bare neo-Nazi code term.

Fox News (40 hits, 20%): High volume of overtly hostile content alongside CNN. Antisemitic discourse spans three distinct sub-categories: Power and Conspiracy narratives (16 cases), Demonology and dehumanizing tropes (12 cases), and Death Wishes or glorification of Iranian violence against Israel (6 cases). The “Epstein Files” motif recurs as both a distraction narrative and a coded reference to Jewish elite conspiracy. Framing of the US as a subordinate actor to Israeli command is explicit and recurrent. The platform also hosts the most extreme exterminatory rhetoric in the dataset.

Forbes Breaking News (20 hits, 10%): Antisemitism surfaces primarily through Deceit and Terror State framings — Israel cast as a conspiratorial expansionist actor pursuing “Greater Israel” (”War for greater Israel has begun again”; “they want the Nile to the Euphrates”). Power and Conspiracy tropes are present, including explicit Jewish control of US policy (”The Jews wanted it, so the USA must do it”; “Trump on his knees before his master”). Demonology appears in both religious and animalistic register. The “Khazarian” reference and Zelensky-as-Jewish-agent comment place this outlet at the intersection of classic and contemporary conspiratorial antisemitism.

Breaking Down the Predominant Concepts of Antisemitism in YouTube Responses

1. Power and Conspiracy Narratives

By a significant margin, Power and Conspiracy narratives constitute the dominant antisemitic register across all eight outlets in the February–March 2026 dataset. This represents a structural shift from the June 2025 case studies, where power tropes competed with demonization, Holocaust inversion, and schadenfreude for discursive primacy. The shift is directly attributable to the altered geopolitical context: with the United States as an active co-belligerent in a war explicitly framed by its own president as aimed at regime change, the conspiratorial claim that Israeli interests control American foreign policy has acquired a degree of apparent factual grounding that it lacked in June 2025.

The most common structural form is the subordination frame: the US is rendered as Israel’s instrument, its president as Netanyahu’s functionary, and its military as a mercenary force deployed in Jewish interests. This framing is expressed with varying degrees of explicitness:

“Trump is such a good boy for his master Bibi!” (Fox News)

“When daddy says attack, The US will attack” (Fox News)

“Israel puppet” (Fox News)

“Israel owns our government we always do what they say. MAGA is Israel first, America second” (Fox News)

“Netanyahu’s Fool = Trump 😮” (MS NOW)

“Not just an ally, they know how to work dumpty don to do their dirty work” (CNN)

“Israel’s bitch” (CNN)

“you spelled puppet wrong” [in reply to “US is ally of Israel”] (ABC)

"United States of Israel" (Forbes)

The final comment in this cluster is the most compressed: a renaming of the country itself. By substituting “Israel” for “America,” the commenter performs the subordination frame at the level of national identity — not merely asserting that the US acts in Israeli interests, but that the distinction between the two states has collapsed entirely.

— criticism of the US–Israel relationship and of Trump’s political alignment with Netanyahu falls within legitimate political discourse. What converts these comments into antisemitic statements is the structural logic they invoke: not that a specific government has pursued particular policies, but that Jewish power operates as an invisible commanding authority to which American political agency is categorically subordinate. The agency attributed to “Israel” or “Bibi” is not political but ethno-conspiratorial — it functions, within the comment’s logic, as a property of Jewishness rather than of a state.

Among the subordination frame statements in the dataset, two warrant particular analytical attention for the clarity and directness with which they express this logic.

“The Jews wanted it, so the USA must do it...” (Forbes)

This is arguably the most syntactically unmediated statement of Jewish command over American foreign policy in the entire corpus. There is no metaphor, no coding, no rhetorical indirection: Jewish collective will is the subject; American state action is the predicate; causation is absolute. The comment strips away the institutional scaffolding — AIPAC, the lobby, Netanyahu — that other comments rely on to perform the same assertion, and states the subordination frame in its purest form. Its very plainness is analytically significant: it signals a commenter for whom this logic requires no justification or elaboration, only statement.

“Trump on his knees before his master. 😮” (Forbes)

Where most subordination frame comments use the language of obedience (”good boy,” “puppet,” “daddy says”), this comment deploys a posture of bodily submission — kneeling — that introduces a dimension of sexual humiliation absent from the others. The master/slave dynamic is intensified by the physical image: Trump is not merely instructed but prostrated. The shocked emoji (😮) performs a performative incredulity that simultaneously distances the commenter from direct endorsement and amplifies the degradation on display.

Another comment performs the subordination logic in a visually distinct register — reading Jewish power onto the body of a political figure rather than expressing it through command:

“I see the imprint in his hair for his little round hat. I bet he’s happy.” (CBS)

This comment claims to detect the mark of a kippah beneath Trump’s hair — interpreting a physical trace as evidence of concealed Jewish identity or allegiance. The operative logic is that of the dog whistle in reverse: rather than a speaker encoding a hidden message for a knowing audience, here a reader claims to decode a hidden identity from a physical sign. Jewish power is not asserted but detected — inscribed on the subordinate’s body as proof of his master’s command. The verb “happy” implies that the conflict serves Jewish interests in ways invisible to ordinary observers, completing the subordination frame through somatized conspiracy.

This logic is made more explicit in comments that invoke institutional proxies — AIPAC in particular — as evidence of systemic Jewish capture of US democracy:

“That’s why A¡pac invested heavily on both sides of the election — now that investment is paying off.” (ABC)

“Because of AIPAC” (ABC)

“Free America from criminal AIPAC!” (CBS)

“billions and billions of taxpayer money spent on Israel First isn’t doing us any good either.” (MS NOW)

“This is Israel control the United States Republicans party... billionaires class support Israel” (MS NOW)

“Our kids will be living under intense Israeli surveillance forever. Israel lobby owns Trump and most of Congress. America is dead.” (MS NOW)

The AIPAC trope has a long history in American political discourse and exists on a spectrum that includes legitimate critique of lobbying influence. What distinguishes its antisemitic variant is the shift from describing political activity to asserting ethno-religious domination: “Israel lobby owns Trump and most of Congress” performs not a critique of campaign finance but an assertion of total Jewish capture of American democratic institutions.

The Epstein thread constitutes a distinct and analytically significant sub-pattern, appearing across all eight outlets — and entirely absent as a cross-platform formation from both June 2025 datasets. Its emergence as a cross-platform motif is specific to the 2026 context, in which direct US military involvement alongside Israel has given pre-existing conspiratorial narratives about Epstein’s alleged ties to Israeli intelligence a new geopolitical anchor. The thread functions as a shorthand for Jewish elite conspiracy:

“a friday night strike who would have guessed? Wag the dog. Release the Epstein Files. The President of Peace??” (NBC)

“All this is a distraction to hide the files, Why we should go in war for Israel??” (Fox News)

“Anything to distract from the Philes [= Files; alternate spelling used to evade content moderation] 🙄” (Fox News)

“It’s the so-called Epstein War.” (NBC)

“YEA YEA YEA WHAT ABOUT THE EPSTEIN FILES WITH ALL THOSE jewISH/ISRAEL TIES!!!!!!!!!!” (CBS)

“Because Mossad had dirt on Donnie” (ABC)

The Epstein thread operates on two levels simultaneously. On the surface, it invokes a legitimate political question about the release of files related to Jeffrey Epstein’s network. Beneath this surface, it activates a conspiratorial interpretive frame in which Jewish actors — indexed through the capitalized “jewISH/ISRAEL TIES” or the reference to Mossad — are alleged to have orchestrated both the current war and the suppression of evidence connecting Jewish elites to sexual exploitation and blackmail. The distraction narrative (”all this is a distraction to hide the files”) presents the war itself as an instrument of Jewish information control. This is among the most structurally complex antisemitic formations in the dataset, and one that has dramatically intensified and become cross-platform in the 2026 context: it links Jewish sexual deviance (a classical trope), Jewish financial power, Jewish media control, and Jewish manipulation of heads of state into a single conspiratorial edifice.

The most elaborate instance of this sub-pattern appears on CNN, where a commenter assembles a theological conspiracy narrative explicitly linking Jewish religious authority to demonic political power:

“The Orthodoxy Jews anointed the #Beast666 their Savior of Israel” / “made the THIRD TEMPLE COIN THAT SAYS TRIBUTE COIN THAT WILL BE REQUIRED” / “the false priesthood of the Jewish messiah will end” / “the THIRD GOLDEN TEMPLE IN JERUSALEM” (CNN)

This comment draws on Christian eschatological tradition to cast Orthodox Jewish religious leadership as collectively aligned with a demonic political figure, invoking the Third Temple as a symbol of secretly orchestrated world domination. The “false priesthood” construction echoes classical anti-Judaic formulations of Jewish spiritual illegitimacy. Combined with the #Beast666 reference — a direct invocation of the Antichrist — the comment exemplifies how conspiratorial and theological antisemitism converge in digital environments, producing a framework in which Jewish actors are simultaneously politically omnipotent and spiritually corrupt.

A further sub-pattern, not prominent in June 2025, concerns power over media institutions:

“What do you expect? Disney is owned by Israel.” (ABC)

The Disney comment instantiates the classical trope of Jewish media control in a contemporary corporate idiom — Jewish ownership of a major entertainment conglomerate presented as self-evident proof of broader informational control. A distinct but related formation appears at Forbes Breaking News:

“I guess Ukrainian bots are all here now. Jewish president Zelensky is paying with our money” (Forbes)

This comment operates through the logic of virtual orchestration: dissenting voices in comment sections are recast not as genuine public opinion but as coordinated Jewish-funded disinformation, with Zelensky — identified as Jewish — positioned as the financial conduit. The effect is double: it delegitimizes any pro-Israel or pro-Ukraine commentary as astroturfed, and it asserts Jewish control not merely over media institutions but over the comment space itself — the very arena in which this study’s corpus was collected.

The dataset’s most compressed and historically loaded power formulations appear at the Associated Press, where bare neo-Nazi code terms surface without elaboration:

“ZOG” (AP)

“KHAZARIAN MAFIA BIG BO$$ BATTLE” (AP)

“ZOG” — Zionist Occupation Government — is one of the oldest and most explicitly neo-Nazi acronyms in anglophone white supremacist discourse, asserting that Western governments are secretly controlled by Jewish forces. Its appearance as a standalone comment, without any surrounding elaboration, signals a commenter who expects the audience to be familiar with the term — a marker of community membership in extreme-right online spaces. “KHAZARIAN MAFIA BIG BOBATTLE”invokesthediscreditedKhazartheory—theclaimthatAshkenaziJewsarenotdescendedfromancientIsraelitesbutfromamedievalTurkicpeople,andthereforehavenolegitimateconnectiontothelandofIsrael—combinedwiththe“KhazarianMafia”conspiracythatportraystheseputativeKhazariandescendantsassecretlyrunningglobalfinanceandpolitics.Thedollarsignsubstitution(”BO”) reinforces the financial control dimension of the trope. Together, these two comments represent the most extreme end of the conspiracy spectrum in the dataset: not coded insinuation but bare ideological assertion in the language of organized antisemitic movements.

A further conspiratorial sub-pattern — less explicit but analytically significant — involves territorial expansionism as hidden Israeli motive: the claim that the current war is not a security operation but a pretext for the realization of a vast imperial project. This formation, which appeared as a standalone “Deceit” category in Part I of this series, surfaces here primarily at Forbes Breaking News:

“War for greater Israel has begun again it seems.” (Forbes)

“they want the Nile to the Euphrates” (Forbes)

The “Greater Israel” allegation invokes a long-standing antisemitic conspiracy theory — that Israeli leaders secretly pursue the establishment of a Jewish state stretching from the Nile to the Euphrates, encompassing large parts of Egypt, Jordan, Syria, Iraq, and Saudi Arabia. The claim derives from a tendentious reading of the biblical covenant in Genesis 15:18, selectively appropriated and decontextualized to assert a covert Jewish imperial agenda. “The Nile to the Euphrates” functions as a coded shorthand recognizable to those familiar with the conspiracy, operating as a dog whistle that simultaneously invokes territorial ambition, biblical mandate, and the concealment of true objectives behind the language of security. In this reading, Operation Epic Fury is not a preemptive military action but the opening move of a millennia-old land-grab — and the Trump administration is its unwitting or complicit instrument.

2. Demonology and Dehumanization

The second most prevalent antisemitic category in the dataset draws on a long tradition of representing Jewish actors as spiritually, morally, or ontologically corrupt — not merely politically objectionable, but evil in an absolute or metaphysical sense. As in the June 2025 case studies, these framings blend classical religious motifs with contemporary political language, producing a hybrid rhetoric in which theological vilification and geopolitical accusation reinforce each other.

The most compact expressions operate through symbolic condensation — single phrases that activate dense networks of antisemitic meaning:

“Small hats — big problems.” (Fox News)

This comment is structurally minimal but semiotically dense. The “small hat” (kippah) functions as a metonymic marker of Jewish identity, transforming an item of religious observance into a symbol of danger and malevolence. The rhyming structure performs a casual, mocking tone that intensifies rather than mitigates the hostility, casting Jewish identity itself as the source of global harm.

“Zionism is a okague [plague]” (Fox News)

The identification of Zionism — functioning here as a synonym for Jewish political self-determination — as a plague enacts a dehumanizing metaphor with specific historical resonance: pathological framings of Jews as disease, contamination, or pestilence form one of the oldest and most lethal strands of antisemitic rhetoric.

“God doesn’t bless Jewish demons” (Fox News) [in reply to “God bless us every one”]

This exchange uses the form of a benign greeting to introduce a theological accusation of demonic Jewish nature, drawing directly on Christian demonological tradition.

“Israhell” (CNN)

The “Israhell” portmanteau — combining Israel and Hell — recurs across platforms and represents one of the most widespread forms of symbolic demonization in digital antisemitic discourse. It condenses theological vilification with national delegitimization, suggesting that Israel’s very existence is infernal.

“Because Israel is an evil fascist state” (CNN)

“Are they really ur ally? U got friends that keep getting u in trouble?” (CNN)

This question — ostensibly rhetorical — enacts a dehumanizing move through the register of street vernacular: Israel is not cast as a dangerous state actor but as a bad friend, a troublemaker whose malevolent character is self-evident. The informality of the phrasing (”ur ally,” “getting u in trouble”) performs a knowing contempt, as though the speaker is incredulous that anyone could still be defending the alliance.

The most structurally complex demonizing statement in the dataset is a multi-turn exchange at CNN that fuses dehumanization with conspiracy and theological condemnation:

A: “WOMD [= WMD, Weapons of Mass Destruction] 2.0. What exactly are the American poor people about to die for?!”
B: “The Epstein Billionaire Class and the Synagogue of Satan. In other words, business as usual, and yes we are going to Hell if not already there...” (CNN)

The reply performs several antisemitic operations simultaneously. “The Epstein Billionaire Class” indexes Jewish elite conspiracy and sexual criminality as the beneficiaries of American military sacrifice. “The Synagogue of Satan” — drawn from Revelation 2:9 and 3:9, and one of the most historically consequential anti-Judaic formulations in Christian scripture — identifies the same elite as spiritually demonic. The phrase “business as usual” normalizes this framework as obvious, self-evident, and long-established. Together, the reply constructs a single conspiratorial-theological formation in which Jewish actors are simultaneously financially parasitic, sexually deviant, and satanically evil.

Two further exchanges illustrate how demonization operates through apparently neutral dialogue structures, with the antisemitic payload delivered in the final turn:

A: “Imagine how much peace the world would have without the Middle East”
B: “Israel to be precise.” (Fox News)

A: “War 🤬 what is it good for, absolutely nothing”
B: “Jews” (Forbes)

Both exchanges share the same rhetorical structure: an opening statement that appears to express general frustration is redirected by the reply to identify Jews or Israel as the specific cause. The Forbes exchange is maximally compressed — a single word answers a classic anti-war sentiment by attributing war itself to Jewish agency. Its brevity is not incidental; it performs the antisemitic attribution as self-evident, requiring no elaboration or argument.

“A world with Izrale [= Israel] is a world always filled with wars...” (AP)

“USA, Israel cancer of the world peace.” (AP)

The cancer metaphor — like the plague metaphor — carries a specific historical charge, having functioned as a central trope of exterminatory antisemitism. Its appearance in conjunction with calls for world peace renders Jewish existence as the obstacle to global harmony, echoing precisely the logic that historically preceded calls for elimination.

Several comments extend demonization of Israel as a state to the dehumanization of Jewish and Israeli people:

“they don’t act like allies. They act like parasites.” (CNN)

“I’m so tired of these Zionist rats and their wars...” (AP)

“Prayers for the Iranian people as they fight a joo [= Jewish] invasion 😢” (Forbes)

The parasite and rat metaphors carry a specific historical charge as central tropes of exterminatory antisemitism. “Joo” is a common phonetic misspelling used to perform both ethnic identification and contempt simultaneously.

The most extreme demonizing statements escalate from metaphor to explicit exterminatory rhetoric:

“No declaration of war is a war crime. Nuke the Zionist squatters now and all of the world’s problems will end instantaneously.” (Fox News)

“god’s chosen people you know” (AP) [sarcastic inversion]

The nuclear annihilation call frames genocide not as hatred but as a solution to a universal problem — precisely the rhetorical structure of historical exterminatory movements. The phrase “all of the world’s problems will end instantaneously” positions Jewish death as a therapeutic global act.

Blood Libel

A distinct and historically weighted sub-pattern within the Demonology category involves the transposition of the medieval blood libel — the accusation that Jews murder children and consume or exploit their blood for ritual or instrumental purposes — into a contemporary geopolitical register. The blood libel is one of the oldest and most lethal antisemitic formations in Western history, having directly precipitated pogroms, expulsions, and mass killings across centuries. Its appearance here is not in its classical ritual murder form, but its structural logic is intact: Jewish actors are cast as the consumers of children’s blood, and American families are cast as the involuntary sacrificial donors.

“America children blood is available free for spillage — long live Israel — First Israel needs protection — American parents should work 24/7 and provide aid to First Israel” (Fox News)

This comment frames American children’s lives as a resource freely extracted by Israel — blood as tribute, parental labor as subsidy. The sarcastic register (”long live Israel”) performs a bitter irony in which American sacrifice is inverted into Israeli glorification. Israel is not an ally receiving support but a sovereign demanding blood payment from its subjects.

The Forbes exchange enacts the same structure through dialogue:

A: “Gotta feed the war gods with the blood of our children.... 😢😢😢”
B: “Iran wouldn’t have it any other way”
C: “you misspelled Israel” (Forbes)

The opening comment frames the war in terms of sacrificial religion — “war gods” demanding child sacrifice — with genuine grief signaled by the crying emoji. The second reply deflects toward Iran. The third and final reply corrects the attribution with casual, devastating precision: it is Israel, not Iran, who is the true “war god” consuming American children. The dialogic structure performs an ideological sorting in real time, with the blood libel logic arriving as the authoritative final word. The brevity and confidence of “you misspelled Israel” — offered without elaboration — signals a shared conspiratorial understanding between commenter and assumed audience.

3. Schadenfreude, Death Wishes, and Glorification of Violence

In the June 2025 data, schadenfreude was distributed across multiple platforms. In the February–March 2026 dataset it is most sharply concentrated at Fox News and the Associated Press — a cross-ideological pairing that reflects the specific audience dynamics of those comment ecosystems rather than a uniform platform pattern.

At Fox News, Death Wishes take the form of gleeful anticipation of Iranian retaliation against Israel:

“Time for Iran to send their tens of thousands of drones after Israel along with hundreds of missiles.” (Fox News)

“I CANT WAIT FOR IRAN’S SPECIAL OPERATIONS 😅 I HOPE YOUR KIDS WEAR SUNGLASSES 😅” (Fox News)

“Iran should attack on Israel and America with 1000s missiles” (Fox News)

The second comment is analytically significant. “I hope your kids wear sunglasses” references the blinding flash of a missile detonation or nuclear strike, expressing the hope for mass casualties among Israeli children as a source of entertainment. The laughing emoji converts a death wish into a performance of gleeful anticipation, deploying humor as affective cover for genocidal fantasy.

The Associated Press comment section produces the most extreme Death Wish formulations in the dataset, including statements that move from conventional military framing into explicit annihilationary language:

“The world will benefit immensely from millions of d3d [= dead] israLIEs and Zi0nshytz [= Zionist sh*ts]” (AP)

“I wish Israel to turn to glass” (AP)

“Boom boom Tel Aviv is coming soon 🇮🇷” (AP)

“Iran destroy Israel very soon... Wate [= Wait] and See” (AP)

The “turn to glass” formulation warrants specific annotation. The phrase derives from nuclear weapons imagery: when a nuclear device detonates, the extreme heat melts sand and soil into glass-like material (historically called trinitite, after the first nuclear test at Trinity site). The expression “turn to glass” circulates in online discourse as a coded call for nuclear annihilation, functioning as a dog whistle that is legible to those familiar with the reference while maintaining plausible deniability. In this context, its meaning is unambiguous.

One comment from the AP dataset requires particular care in analysis:

“From Iran: The Iranian people stand with the Iranian government Ha Admaro Tel Aviv” (AP)

The phrase “Ha Admaro Tel Aviv” appears to be a phonetic approximation of the Arabic militant phrase هدم تل أبيب (hadm / haddam Tel Aviv) — “Destroy Tel Aviv.” If a commenter heard this phrase and attempted to transcribe it without knowledge of Arabic orthography, the result would plausibly approximate what appears in the comment. This reading is consistent with the comment’s framing as a statement of solidarity with the Iranian government and with the broader pattern of Arabic-language militant slogans circulating in anglophone comment sections. The comment thus functions as a translingual vehicle for a call to violence against Israeli civilians.

4. Nazi Analogy

The Nazi Analogy — the comparison of Israeli actors, policies, or military operations to Nazi Germany — was documented extensively in Parts I and II, particularly at CNN and ABC News. In the current dataset it appears with notable concentration at the Associated Press, where the most explicit instance in the corpus is found:

“Again U$ joins ZioNAZIs and there are no rules” (AP)

The portmanteau “ZioNAZIs” merges Zionist and Nazi into a single term, asserting an identity between Jewish political self-determination and National Socialist genocide. The prefix “Again” implies historical recurrence, casting this moment as a continuation of a pattern of Jewish-Nazi equivalence. The “U$” spelling — substituting the dollar sign for “S” — simultaneously invokes financial conspiracy, reinforcing the associative chain between Jewish power, American subordination, and Nazi ideology.

This formation is analytically distinct from the more diffuse Holocaust inversion documented in Parts I and II, where commenters typically reached for Nazi analogies through extended rhetorical elaboration. “ZioNAZIs” performs the inversion in a single compound, making it maximally portable as a hashtag, meme caption, or comment prefix — a characteristic of digital antisemitism’s adaptation to the compressed attention economy of social media.

The Nazi Analogy also surfaces implicitly in comments that invoke the Protocols of the Elders of Zion, the foundational text of modern antisemitic conspiracy theory, which was produced by the Tsarist secret police and subsequently adopted and promoted by the Nazi regime:

“BREAKING NEWS: The Protocols of Zion are real.” (CBS)

Presenting the Protocols as factual — framed here as a news announcement — performs a double move: it validates one of history’s most consequential antisemitic fabrications while deploying ironic news-headline form to suggest that the current war has finally confirmed what the Protocols alleged. In this reading, Operation Epic Fury is not a geopolitical event but documentary evidence of Jewish world domination.

5. Classic Financial and Resource Exploitation Tropes

A recurring sub-pattern across ABC and CNN, with additional instances at Forbes and Fox News, involves the invocation of financial and resource exploitation tropes that position Jews or Israel as extracting material wealth — blood, money, and labor — from American citizens. These tropes are among the oldest in the antisemitic repertoire, and their appearance in the context of US military expenditure and foreign aid gives them a contemporary political scaffold:

“Israel has no natural resources and the only thing they export is usury.” (CNN)

A multi-turn CNN exchange makes the financial exploitation logic explicit through a cumulative dialogue structure:

A: “Why is that wrong to have Israel as an ally?”
B: “Because they hate Jews”
C: “Because Israel is an evil fascist state”
D: “Because they bleed you of resources to the tune of billions annually and stir up forever wars for your children and not the politician’s children to fight in” (CNN)

The thread is analytically significant because it enacts a process of ideological sorting in public: multiple positions are offered in sequence, and the final reply — the one with the highest rhetorical investment — lands on financial exploitation and the blood of American children as the definitive answer. The phrase “bleed you of resources” activates the classical parasitic-financial trope; “forever wars for your children” invokes the blood toll framing; the contrast with “politician’s children” adds a class dimension that anchors the antisemitic claim in a populist register.

“Exactly and use their money, instead they use our kids[,] our money and then we give them billions just because our government sold us out for some silver.” (ABC)

“WE AMERICANS ARE TIRED OF YOUR PEOPLES NEVER ENDING THIRST!” (ABC)

“America children blood is available free for spillage — long live Israel” (Fox News)

“Wall Street > military industry > Israel >>>> US foreign policy.” (Forbes)

Although appearing only once in the dataset, the usury comment activates one of the most historically embedded formations in antisemitism: the identification of Jews with predatory lending as their primary economic and social contribution. The “sold us out for some silver” comment introduces a biblical register — the thirty pieces of silver paid to Judas for the betrayal of Jesus — repurposing a foundational narrative of Christian antisemitism to frame US support for Israel as an act of treason purchased with Jewish money. The Forbes comment constructs a causal hierarchy — Wall Street feeding the military-industrial complex feeding Israeli interests feeding US foreign policy — that maps the classical conspiracy of Jewish financial control onto a contemporary institutional infrastructure.

Conclusion

The February–March 2026 dataset — 1,600 comments across eight US news outlets, with a mean antisemitic content rate of 12.6% — confirms and significantly extends the patterns documented in Parts I and II, while revealing several structurally significant shifts.

Power and Conspiracy Narratives have consolidated as the dominant antisemitic register across all eight outlets, a development directly conditioned by the geopolitical context of US co-belligerence and publicly stated regime change objectives. The Epstein thread — documented in isolated form in the Part II Fox News dataset but absent as a cross-platform formation — has consolidated in the 2026 context into a near-ubiquitous organizing motif, appearing across all eight outlets, frequently under the shorthand ‘Trumpstein files,’ which fuses Trump’s political vulnerability to the Epstein network with the allegation that the current war serves as a deliberate cover. The Nazi Analogy — particularly the “ZioNAZIs” portmanteau at AP and the Protocols of Zion reference at CBS — now circulates in compressed, maximally portable forms adapted to the attention economy of social media. Death Wishes at AP and Fox News have escalated to explicit calls for nuclear annihilation and translingual militant slogans. Financial exploitation tropes are more prevalent as a standalone sub-category than in either June 2025 study.

CNN (21%) and Fox News (20%) again lead the dataset — as they did in both June 2025 studies — confirming what is now one of the most consistent empirical findings across the series: antisemitic comment culture is a cross-ideological phenomenon. The outlet at the top of the left-leaning spectrum and the outlet at the top of the right-leaning spectrum produce near-identical antisemitic response rates, separated in Part III by a single percentage point. The surface framings differ — CNN comment sections tend toward financial conspiracy and theological demonology; Fox News toward subordination, death wishes, and Epstein distraction narratives — but the underlying antisemitic logic is structurally homologous. This symmetry is analytically significant precisely because it resists reduction to a partisan explanation: antisemitism in these comment ecosystems is not a pathology of the right or the left but a cross-ideological formation that draws on a shared repertoire of tropes, adapted to the rhetorical register of each platform’s audience. The Associated Press produces the most diverse repertoire — spanning all five analytical categories, including the only unambiguous Nazi Analogy and the most extreme Death Wish formulations in the corpus. Forbes Breaking News, absent from Part II and appearing here with a complete sample for the first time, contributes a distinct Deceit/Terror State framing alongside Power and Conspiracy content, and introduces the “The Jews wanted it, so the USA must do it” formulation as one of the most explicit statements of Jewish command over American foreign policy in the dataset.

Together, these findings — across five analytical categories: Power and Conspiracy, Demonology and Dehumanization, Schadenfreude and Death Wishes, Nazi Analogy, and Financial Exploitation — confirm that the current escalation has not produced fundamentally new antisemitic formations, but has dramatically intensified and reorganized existing ones — providing factual scaffolding for conspiracy narratives, new focal events for demonizing projection, and a geopolitical context in which the rhetoric of Jewish omnipotence and American subordination can circulate with an appearance of empirical grounding. When political facts are themselves amenable to antisemitic interpretation, the work of identifying and contesting antisemitic rhetoric requires not only lexical precision but sustained contextual and historical analysis of the kind the Decoding Antisemitism Lexicon is designed to support.

This article is Part III of the Antisemitism in Real Time series. Parts I and II were conducted under the Decoding Antisemitism project. Part III marks the continuation of this work under Decoding Hate, the expanded research hub at NYU’s Center for the Study of Antisemitism. All three parts draw on the Decoding Antisemitism Lexicon.

AddressHate Research Scholar at NYU’s Center for the Study of Antisemitism | Lead, Decoding Antisemitism | Research Advisor, AddressHate

About This Essay

This essay is a condensed and publicly accessible version of a research report developed in the context of my Decoding Antisemitism project and co-authored with Gabrielle Beacken (University of Texas at Austin) and Liora Sabra (NYU’s Center for the Study of Antisemitism).

The full research report, titled “How Political Ambiguity Transforms Across Digital Platforms: Mechanisms of Discourse Radicalization in the Case of Zohran Mamdani,” is openly available via the NYU UltraViolet repository and can be accessed here.

While the complete report presents the empirical analysis, methodological design, and quantitative findings in full detail, the present essay translates its core arguments for a broader, non-academic readership. It focuses in particular on the dynamics of political discourse, influencer reframing, and participatory escalation in digital environments, while retaining the implications of the research for the study of antisemitism, online radicalization, and contemporary media ecosystems.

Introduction

Since October 7, 2023 — the Hamas attacks on Israel and the subsequent war in Gaza — public discourse about Israel and Palestine has intensified in volume, speed, and volatility. Political statements that once circulated primarily through institutional media now travel fluidly across platforms, influencer networks, and participatory audiences. Along the way, they are clipped, reframed, moralized, memefied, and reinterpreted — often producing outcomes that far exceed what any original speaker explicitly said or intended.

A recurring pattern has become increasingly difficult to ignore. Political statements framed as legitimate critique repeatedly reappear downstream as overtly antisemitic discourse. This has fueled a predictable but unsatisfying debate. Critics blame algorithms, platform toxicity, or radical audiences. Defenders respond that the original speech was not antisemitic and cannot be held responsible for how others interpret it. Both positions rest on an outdated model of communication in which meaning is transmitted linearly from speaker to listener.

The core problem is that political speech does not move through digital environments intact. It transforms. This article shows that when the same political event is reframed conspiratorially by influential opinion makers, the prevalence of antisemitic discourse within audience responses increases dramatically — even though the underlying news trigger remains identical.

This article examines how ambiguous political discourse changes as it circulates through layered digital ecosystems, and why ambiguity — often treated as a democratic safeguard — becomes structurally unstable in participatory media environments. The argument is not that political ambiguity is inherently malicious, nor that criticism of Israel is equivalent to antisemitism. It is that in circulation-driven platforms, ambiguity does not remain neutral. It invites completion — and the direction of that completion is shaped decisively by human intermediaries.

Ambiguity is not intent

Ambiguous political language is not evidence of hidden motives. Ambiguity can arise from coalition politics, institutional constraints, inherited activist vocabularies, genuine moral uncertainty, or attempts to speak across polarized constituencies. In democratic discourse, ambiguity is often understood as a virtue — a way to preserve pluralism, prevent premature closure, or keep dialogue open.

This analysis does not infer politicians’ private beliefs, nor does it assume covert dogwhistling. Intent is not the metric here. The object of analysis is what ambiguity does once it enters digital circulation. This is a study of discursive mechanics, not moral psychology.

From speech to circulation

Public debate still tends to assume a linear communication model: a speaker speaks, an audience interprets, and responsibility resides primarily with wording and intent. That model made sense when speech was ephemeral, audiences were bounded, and intermediaries were institutional gatekeepers.

Those conditions no longer apply.

Today, political statements move through layered circulation systems. They are excerpted by influencers, reframed into platform-specific narratives, embedded in affective and ideological contexts, and transformed into participatory raw material. Meaning is not transmitted; it is reconstructed. The central question is no longer whether a statement can be interpreted in problematic ways, but which interpretations are stabilized, amplified, and normalized as it travels.

In contemporary digital ecosystems, influencers function as de facto editors of political meaning, shaping not only how political speech is interpreted, but how it is stabilized over time within public discourse.

Why this case matters

To observe these dynamics empirically, the underlying study focuses on the digital circulation of statements by Zohran Mamdani, concerning Israel, Palestine, antisemitism, and political violence. This case is not chosen to single out an individual, nor to adjudicate political legitimacy. It is chosen because it offers an unusually clear analytical window: Mamdani’s rhetoric combines explicit opposition to antisemitism with formulations that remain open to divergent interpretations, generating sustained controversy and high digital circulation.

That combination — high ambiguity, high circulation, high engagement — allows for systematic analysis of how political meaning transforms as it moves from elite speech to influencer reframing to audience participation. The goal is not to judge Mamdani’s intentions, but to understand how ambiguous political discourse collapses into explicit antisemitic expression in participatory environments — and under what conditions it does not.

A three-phase model of digital transformation

The study treats political discourse as a multi-phase circulation process rather than a single communicative act.

At the first phase, elite political speech introduces ambiguity. At this level, ambiguity remains open. Competing interpretations coexist, but none is dominant, and discourse is still anchored to institutional norms such as interviews, journalistic mediation, and expectations of accountability. At this stage, antisemitic discourse remains rare.

The decisive shift occurs at the second phase: influencer reframing. Influencers do not merely repeat political statements; they translate them. They select excerpts, impose narrative structures, provide emotional cues, and align content with ideological expectations. In doing so, they reduce interpretive openness. Ambiguity begins to crystallize.

Two mechanisms matter most here. Influencers embed political speech within broader narratives — anti-imperialism, anti-elite critique, systemic oppression, or populist skepticism — and they supply affective signals such as outrage, certainty, moral urgency, or irony. These cues guide audiences not only on what to think, but how to feel.

The third phase unfolds in participatory spaces, such as comment sections. Here, ambiguity collapses. Participation rewards completion. Users escalate interpretations, compete for social validation, and push implicit meanings toward explicit claims. Over time, norms shift and what once seemed extreme becomes normalized.

Together, these phases form a structured sequence: ambiguity in elite speech, crystallization through influencer framing, and collapse in participatory discourse.

What we actually measured

To test whether ambiguous political speech itself produces antisemitism — or whether outcomes depend on how ambiguity is mediated — the study analyzed discourse across three layers of circulation.

At the primary level, four public appearances by Mamdani were examined, each addressing Israel–Palestine–related issues with language open to competing interpretations. At the secondary level, nine long-form influencer videos (approximately 91 minutes total) from four progressive political creators were analyzed; these videos reframed the same underlying political material rather than introducing new content. At the tertiary level, 3,404 audience comments were systematically sampled from 17 video postings across YouTube and TikTok, including cross-platform reposts.

These patterns held across both YouTube and TikTok, suggesting that framing effects outweighed platform-specific cultures or affordances.

This design allowed comparison of how the same political source material generated different audience outcomes depending on framing strategy, while controlling for platform, time period, and political context.

How antisemitism was identified — conservatively

Because the Israel–Palestine discourse is highly contested, the study used conservative coding principles developed in the Decoding Antisemitism project. Content was classified as antisemitic only when sufficient contextual and intertextual evidence supported that interpretation. When a comment permitted both a prejudiced and a non-prejudiced reading, it was coded as non-antisemitic — meaning the reported figures likely underestimate rather than exaggerate prevalence.

Five forms of antisemitic discourse were tracked: denial or relativization of antisemitism, victim–perpetrator reversal, delegitimization of Jewish institutions or Israel as such, conspiratorial attribution of political power, and Holocaust inversion. Criticism of Israeli policy, opposition to war, or support for Palestinian rights were not coded as antisemitic unless they activated these patterns.

Same source, radically different outcomes

The most striking empirical result is not that antisemitism appears online — that is, unfortunately, not new — but how dramatically its prevalence varies under controlled conditions.

Across the dataset, antisemitic comments ranged from low single-digit percentages in some videos to over thirty percent in others — an approximately eightfold difference. This variation cannot be explained by platform culture, algorithmic amplification alone, audience self-selection, or differences in the political source material. In several cases, the same influencer, on the same platform, discussing the same politician, produced radically different outcomes depending on framing choices.

In short: audiences did not change; framing did.

The decisive variable: empathy vs. conspiracy framing

The dominant explanatory distinction lies between empathy-oriented framing and conspiracy-oriented framing.

Empathy-oriented framing foregrounds humanitarian suffering, moral tragedy, and the limits of certainty. It preserves interpretive openness and discourages collective blame. Where this framing dominates, antisemitic discourse remains marginal.

Conspiracy-oriented framing, by contrast, emphasizes hidden power, elite manipulation, suppression of “real” public opinion, and structural control by vaguely specified or euphemistically named actors. This framing dramatically narrows interpretive space. Ambiguity is no longer open-ended; it is resolved through attribution of intent, coordination, and collective responsibility.

Crucially, conspiracy-oriented influencer framing generates nearly eight times more antisemitic audience responses than empathy-oriented framing of the same political speech. This is the single strongest quantitative finding of the study.

Importantly, conspiracy framing is not itself necessarily antisemitic. Rather, it supplies a causal grammar — politics as hidden coordination, disagreement as deception, outcomes as deliberate manipulation — that invites audiences to identify agents behind perceived power. In practice, this often meant that influencer references to vague “elite power” or “suppressed truths” were taken up in comment sections as increasingly specific claims about who controls media, politics, or public discourse — a shift from structural critique to personalized attribution. In historically saturated discourse environments, that completion reliably activates antisemitic tropes.

Where antisemitism actually enters the discourse

Another key finding concerns where antisemitism appears. It is rare in elite political speech, uneven in influencer commentary, and densest in participatory spaces. Comment sections do not invent antisemitic narratives from nothing; they extend and intensify interpretive cues that have already been made salient upstream.

This is why dismissing comment sections as mere noise is misleading. They are reactive environments. They reveal which interpretations have been normalized, authorized, and emotionally primed. When influencers frame politics as a struggle against concealed power, audiences respond by naming that power — often through historically familiar antisemitic patterns.

The uncomfortable implication

One of the most uncomfortable implications of this analysis is that radicalization does not require extremists. The strongest amplification effects occurred not in fringe spaces, but in mainstream progressive ecosystems that see themselves as morally serious and anti-racist.

Ambiguity plays a central role here. It allows morally charged narratives to circulate without triggering immediate resistance. It delays accountability while enabling alignment. Once participatory dynamics take over, the result is not fringe hate speech but normalized hostility — expressed as political insight, moral clarity, or courageous truth-telling.

What this means

Three conclusions follow.

First, ambiguous political speech does not automatically produce antisemitism, but it can enable it under specific circulation conditions. Second, influencer framing is the single most powerful determinant of downstream outcomes. Third, participatory spaces do not distort discourse randomly; they complete what upstream actors leave open.

The implication is not censorship, nor the end of political critique. It is a shift in responsibility. In digital environments, responsibility does not end with intent or wording. It extends to how speech is framed, circulated, moralized, and amplified by those who function as intermediaries.

When amplified by influential intermediaries, ambiguity becomes a powerful engine of discursive normalization — shaping not just momentary outrage, but enduring trends in how groups are discussed, blamed, and morally positioned.

AddressHate Research Scholar at NYU’s Center for the Study of Antisemitism | Lead, Decoding Antisemitism | Research Advisor, AddressHate

About This Essay

This essay is a condensed and publicly accessible version of the peer-reviewed research article “Decoding antisemitism online: linguistic and multimodal challenges in the age of AI,” which I co-authored with Jordan Blatter (BSA) and Oksana Stanevich (TAU), published a few days ago in Frontiers in Communication.

The original article appears in the special issue “Online Hate Speech: Linguistic Challenges in the Age of AI,” edited by Carmen Aguilera-Carnerero, and was reviewed by Elena Negrea-Busuioc and Ledia Kazazi.

The full academic version situates the analysis within discourse studies, media governance, and AI research; what follows translates its core arguments for a broader, non-academic readership—while retaining the implications for AI- and LLM-based systems tasked with analyzing or moderating online communication.

Introduction

Antisemitism has not disappeared from public life. It has adapted.

In digital spaces, it no longer relies primarily on explicit slurs or open incitement. Instead, it increasingly operates through irony, insinuation, cultural allusion, and multimodal signaling—often embedded in what appears to be political critique or moral commentary.

This shift poses a challenge not only for democratic societies but also for artificial intelligence systems, including large language models (LLMs), that are increasingly used to detect, classify, and moderate online hate. While such systems have become more effective at identifying overt abuse, they continue to struggle with antisemitism precisely where it now does most of its work: in implicit, coded, and deniable forms of communication.

This essay distills recent interdisciplinary research combining discourse analysis and computational modeling to explain why that is the case—and why the problem cannot be solved through better keywords, larger datasets, or more powerful language models alone.

Antisemitism in the Digitally Restructured Public Sphere

Social media has fundamentally altered how public meaning is produced. Unlike the era of traditional journalism, characterized by professional gatekeeping and top-down information flows, today’s digital public sphere is shaped by horizontal participation, emotional immediacy, and algorithmic amplification.

This transformation has two consequences that are especially relevant for antisemitic discourse—and for the AI systems trained on data generated within these environments.

First, platforms generate perceptual distortion. Through echo chambers and algorithmic reinforcement, users are led to believe their views are widely shared. This perceived consensus lowers inhibitions and encourages the testing—and transgression—of social taboos.

Second, platforms reward polarization. Engagement-driven algorithms privilege emotionally charged and transgressive content, amplifying precisely those utterances that provoke outrage, mockery, or moral certainty. Together, these dynamics create political-cultural opportunity structures in which antisemitic discourse can circulate while remaining partially obscured.

Antisemitism is particularly well suited to this environment—not because it is uniquely extreme, but because it is uniquely flexible. That flexibility poses a fundamental challenge for AI systems trained to recognize stable, surface-level indicators of hate.

Antisemitism as a Connective Ideology

Antisemitism does not operate like most other forms of hate speech. It is not merely hostility toward a group. It functions as a connective and enabling ideology—a discursive grammar that links conspiracy thinking, institutional distrust, misogyny, and anti-democratic resentment.

Historically, antisemitism has served as a reservoir of explanations for social complexity. In digital environments, it continues to play this role by offering simplified narratives of hidden power, orchestration, and malevolent control. For this reason, antisemitic tropes frequently appear at the intersection of otherwise distinct ideological camps.

This is what makes antisemitism analytically significant beyond Jewish communities. It also explains why AI systems trained on narrow definitions of abuse or extremism routinely underestimate its prevalence and misinterpret its function. What appears, at the surface level, as political critique or moral positioning often performs deeper ideological work that only becomes visible through contextual and discursive analysis.

Hate That Works Through Latency and Deniability

A central reason antisemitism is difficult to detect—both for humans and for AI systems—is that it often does not announce itself as hatred.

Long before the digital age, sociologists described this phenomenon as communication latency: antisemitic meaning persists in coded, indirect, or camouflaged forms that remain socially intelligible while preserving deniability. What has changed online is the scale, speed, and reward structure for such communication.

Latency is not an accident. It is a design feature.

Coded expressions allow speakers to signal alignment to in-groups, evade moderation systems, and retreat into plausible deniability when challenged. Antisemitic meaning emerges not from individual words, but from context, presupposition, and shared cultural and historical knowledge—and increasingly from multimodal cues that exceed the reach of text-based analysis.

This is why antisemitism so often appears not as abuse, but as irony, rhetorical questions, or “just asking.” Meaning does not reside in isolated lexical items, but in how utterances position themselves within a shared discursive horizon and activate presupposed frames of interpretation.

For LLMs and other language models, this creates a structural problem: they are asked to classify meaning where meaning is deliberately under-specified, context-dependent, and strategically ambiguous.

A Spectrum, Not a Slur

To capture this complexity, the research distinguishes four analytically relevant forms of antisemitic discourse:

1. Explicit antisemitism
Open hostility, calls for violence, or celebratory reactions to attacks on Jews.

2. Co-text-dependent antisemitism
Statements whose meaning becomes antisemitic only in relation to surrounding comments (e.g., pronouns or ellipses that rely on prior references).

3. Implicit or context-dependent antisemitism
Utterances that require historical, cultural, or intertextual knowledge to decode—Holocaust numerology, ironic Yiddish markers, coded allusions to Nazi violence.

4. Gray-zone cases
Ambiguous statements that plausibly support both antisemitic and non-antisemitic interpretations.

This spectrum is central not only to qualitative analysis but also to any responsible use of AI in this domain. Rather than forcing binary decisions, it enables AI systems to assist by identifying recurring patterns, clustering similar discursive moves, and flagging zones of ambiguity for expert interpretation.

Crucially, the analysis adopts a conservative ambiguity rule: when a statement plausibly allows a non-antisemitic reading, it is classified as non-antisemitic. This minimizes overreach, but it also means that measured prevalence represents a lower bound rather than the full extent of antisemitic meaning-making online—an important consideration for both human analysis and AI training.

Two Events, Two Discursive Modes

October 7, 2023: When the Masks Came Off

In the immediate aftermath of the Hamas terror attacks, antisemitic discourse surged across mainstream comment sections. Unlike earlier escalation phases, many users abandoned rhetorical camouflage altogether. Violence was openly justified, celebrated, and aestheticized.

This moment marked a discursive rupture. Antisemitism shifted from projection and insinuation to self-positioning: hatred asserted as spectacle. Multimodal signals—emojis, symbols, and shorthand references—reinforced alignment and celebration while evading keyword-based detection.

The result was not just more antisemitism, but a visible recalibration of what could be said in mainstream spaces.

Washington, D.C., May 2025: Antisemitism in the Gray Zones

By contrast, reactions to the antisemitic double murder at the Capital Jewish Museum rarely celebrated violence openly. Instead, antisemitism appeared through denial, mockery, and moral reframing.

Three patterns dominated:

• Conspiracy narratives suggesting Jewish or Israeli orchestration.

• Instrumentalization of antisemitism, portraying Jewish victimhood as exaggerated or manipulative.

• Moral relativization, implying that victims were complicit and therefore not innocent.

Here, antisemitism operated primarily through irony and insinuation. Hatred was not denied, but reframed as political reasoning—precisely the kind of discourse that remains most difficult for automated systems to classify consistently.

Why AI (and LLMs) Keep Missing the Point

Most content moderation systems, including contemporary LLM-based classifiers, are optimized for surface toxicity: insults, slurs, and threats. Antisemitism increasingly avoids all three.

As a result, AI systems fail in predictable ways:

• Explicit antisemitism is often detected.

• Implicit and coded forms are frequently missed.

• Gray-zone political speech produces instability and inconsistency.

• Counter-speech is often misclassified due to keyword bias.

This is not primarily a technical failure. It is a conceptual mismatch. Antisemitism is a discourse phenomenon, not a lexical one.

Even advanced LLMs tend to reproduce the limits of their training data. Where human annotators disagree—most notably in gray-zone cases—models struggle most. Scaling model size does not resolve this problem; it often masks uncertainty rather than addressing it.

Multimodality and Semantic Compression

A further complication is that antisemitic meaning increasingly migrates away from text altogether.

Emojis, memes, symbols, and visual shorthand function as semantic compressors: minimal surface form, maximal ideological load. They allow rapid diffusion, in-group recognition, and moderation evasion.

Text-only systems are structurally blind to this layer. Even multimodal systems struggle when meaning depends on timing, intertextual reference, or culturally specific symbolism that no standalone model can reliably infer.

Rarity, Harm, and the Democratic Stakes

These structural properties of antisemitic discourse have direct consequences for how machine learning systems perform—and fail—at scale.

Antisemitic content is statistically rare in large datasets. Yet its social impact is disproportionate. It shapes narratives, normalizes violence, and links disparate forms of resentment into coherent worldviews.

This creates a structural problem for machine learning: high harm, low frequency, high ambiguity.

From a technical perspective, these conditions undermine standard optimization strategies. From a democratic perspective, they generate a normative dilemma.

Over-blocking risks suppressing legitimate political speech, satire, or critical debate. Under-blocking normalizes antisemitic discourse and allows it to circulate as background noise in mainstream spaces. Both outcomes erode trust in the digital public sphere.

The challenge, then, is not to automate judgment away—but to design systems that support human interpretation, grounded in historical knowledge, contextual awareness, and transparent standards.

Conclusion: Why This Cannot Be Automated Away

AI can play an important role in identifying patterns, mapping discursive trajectories, and scaling analysis across large datasets. But it cannot replace human judgment grounded in linguistic, historical, and cultural expertise.

Antisemitism online is not a bug in the system. It exploits the system. It thrives on ambiguity, latency, and multimodality, adapting faster than static detection tools and benefiting from the affordances that define digital communication.

Confronting it, therefore, requires more than better models. It requires better conceptual foundations: fine-grained classifications, conservative annotation standards, and sustained interdisciplinary collaboration between social scientists, linguists, and AI practitioners.

This is not a rejection of AI—but a call to use it as part of an interpretive ecosystem rather than as an automated arbiter of meaning.

AddressHate Research Scholar at NYU’s Center for the Study of Antisemitism | Lead, Decoding Antisemitism | Founding Research Advisor, AddressHate

Yair Rosenberg’s recent Atlantic piece documents a striking pattern: antisemitism in the United States is no longer primarily a partisan phenomenon, but a generational one. Drawing on multiple large-scale surveys, Rosenberg demonstrates that twenty-five percent of Americans under 25 hold unfavorable opinions of Jewish people, with negligible differences between Trump and Harris supporters. Age predicts antisemitic attitudes better than political affiliation—a pattern that cannot be explained by partisan alignment alone.

This finding is empirically robust—and analytically incomplete. It contradicts the comfortable assumption that bigotry naturally fades with demographic change. But to understand why this generational divide exists requires examining the infrastructures of social media platforms and the communication conditions through which political meaning is now formed.

Rosenberg’s analysis identifies structural factors rather than partisan blame. He points to how social media platforms with weak moderation allow conspiracy theories to spread rapidly, algorithmically amplified because they generate engagement. He describes these platforms as an “escalator” that gradually funnels users from legitimate criticism toward explicitly antisemitic material. Crucially, he emphasizes that antisemitism remains a minority position—most young Americans reject it, but its visibility is amplified by outrage-driven media dynamics.

This framing validates what years of analyzing digital discourse have been demonstrating about how antisemitism actually spreads. The generational pattern isn’t about young people being uniquely susceptible to prejudice—it’s about the information infrastructure through which they form political understanding. What makes the convergence between Rosenberg’s work and ours particularly significant is the complementarity: his survey data provide population-level validation of patterns we’ve been documenting at the discourse level. Where our analysis shows how antisemitic content circulates through platform mechanisms, his surveys demonstrate that these mechanisms are having measurable effects on attitudes across entire demographic cohorts.

The statistics Rosenberg presents demonstrate why social media research is essential to understanding contemporary antisemitism. When survey data show dramatic age-based differences but no partisan gap, that reveals where political mindsets are actually being formed. For young people, social media platforms have become the primary arena for political formation—the space where frameworks for understanding complex issues develop. If we want to understand why 25% of Americans under 25 hold unfavorable views of Jews, we need to study the platforms where they encounter political information.

This response makes three interconnected arguments. First, it shows how the Decoding Antisemitism project’s discourse analysis contributes a mechanistic explanation for Rosenberg’s population-level findings—documenting how antisemitic content actually circulates across platforms and ideological contexts. Second, it translates Rosenberg’s “algorithmic escalator” metaphor into a concrete analysis of implicit coding, content moderation gaps, and cross-platform diffusion. Third, it proposes structural interventions that address how antisemitism spreads rather than attributing it to political camps or demographic groups. The goal is shifting focus from who holds antisemitic attitudes to the infrastructure that produces and amplifies them.

What the Research Shows About Cross-Spectrum Dynamics

The Decoding Antisemitism project, which has analyzed over 300,000 pieces of digital content across European and American contexts since 2020, offers a mechanistic account of what Rosenberg’s survey data reveal at the population level. Our work began by examining how antisemitic discourse manifests in mainstream comment sections across multiple countries and platforms. What emerged was a pattern that should fundamentally reshape how we think about contemporary antisemitism.

Different political contexts have different ways to communicate antisemitic ideas. This holds true of progressive, liberal, and conservative online spaces. The specific tropes vary—conspiracy theories about global elites in some contexts, accusations of colonial oppression and fascist crimes in others, claims about dual loyalty and classic tropes of malevolence, hypocrisy, and greed in still others. But the underlying mechanisms through which these ideas spread remain remarkably consistent across ideological boundaries.

Our most recent analyses examined comment sections following antisemitic attacks. In the Bondi Beach shooting case, we analyzed YouTube coverage from five major American outlets. Moving from left to right across the political spectrum, CNN’s coverage produced the highest rates of antisemitic discourse at 17%, ABC at 6%, NBC at 5%, Forbes at 14%, and Fox News at 10%. Notably, the Bondi case revealed something that challenges simplistic assumptions: Israel-related antisemitic tropes—false flag theories invoking Mossad, accusations that Jewish orchestration of immigration policies led to the attack, inversions blaming Israeli actions for antisemitic terror—appeared across the political spectrum, including in conservative-leaning comment sections. The patterns we documented included Great Replacement narratives explicitly attributing demographic change to “Jewish globalists,” conspiracy theories about Israeli intelligence staging attacks, and justifications framing the massacre as karmic consequence of Gaza.

For the Washington Embassy shooting in May 2025, where two Israeli diplomatic staff were murdered, we examined coverage from eight outlets spanning the Anglophone media landscape. Arranged from left to right: Al Jazeera English (66% antisemitic discourse), Guardian (40%), BBC (42%), CTV News (50%), The Times (54%), Forbes (28%), Sky News (40%), and Fox LiveNow (24%). The Washington case showed particularly strong cross-spectrum presence of conspiracy theories, with “false flag” narratives, mockery of Jewish grief (”kvetching intensifies”), and justifications invoking Gaza appearing across outlets regardless of ideological orientation.

These findings build on DA’s six Discourse Reports, which examine antisemitic responses across Germany, France, and the United Kingdom, utilizing different discourse events, languages, and online spaces. That larger corpus analyzed diverse trigger events, including antisemitic incidents, Holocaust commemorations, and Middle East escalation phases, through rigorous critical discourse analysis. The pattern held consistently: each type of event generated antisemitic outrage in social media comment sections of progressive, liberal, and conservative media outlets alike.

The pattern that emerges isn’t about left versus right. It’s about how discourse events trigger antisemitic responses across different media ecosystems, with variation correlating more strongly to platform characteristics, moderation standards, and event framing than to outlet ideology. Israel-related antisemitism—once primarily associated with left-progressive spaces—now circulates widely in right-wing contexts as well, often merging with older conspiracy frameworks about Jewish power and civilizational decline.

Documenting prevalence across aligned audiences is not a claim of moral symmetry but of infrastructural vulnerability. The mechanisms through which antisemitic content evades moderation, gets algorithmically amplified, and spreads across platforms operate regardless of the ideological framing that activates them. This is why the cross-spectrum presence matters: it redirects attention from political blame games toward the platform architectures that enable spread.

Of course, everything related to October 7, Israel, and the Middle East conflict has created asymmetric effects. There is substantial demonization of Israel and glorification of terror, primarily in left-progressive spaces—the inversions casting Israeli self-defense as genocide, the justifications treating Hamas’s massacre as resistance, the systematic denial of Jewish trauma. But we cannot ignore the staggering numbers of antisemitic discourses we document in conservative contexts as well. The cross-spectrum presence is an empirical reality, even as its specific manifestations and intensity vary across different political contexts. Different tropes, different triggers, different justification strategies—but the underlying platform mechanics that allow implicit content to evade moderation and reach mass audiences operate across ideological boundaries.

These findings describe the prevalence of antisemitic discourse within aligned outlet audiences during specific discourse events. They don’t imply that antisemitism is caused by political orientation. What they demonstrate instead aligns with Rosenberg’s generational finding: if age predicts antisemitism better than politics, this points toward platform infrastructure and event dynamics as explanatory factors rather than partisan identity. When we analyze comment sections under ideologically legible outlets, we’re examining where audiences aggregate, not why antisemitism appears—outlet ideology shapes which audiences gather, but platform affordances shape how antisemitic discourse circulates once they do.

The Algorithmic Escalator and Platform-Mediated Conditions

When Rosenberg describes social media as an “algorithmic escalator” that funnels users from legitimate criticism toward antisemitic radicalization, he’s identifying something our research has documented mechanistically. But the escalator metaphor, while apt, risks oversimplifying the problem as purely algorithmic. What actually operates is a fuller set of platform-mediated communication conditions.

Our DA taxonomy of over 170 classification units for hateful discourse reveals how contemporary antisemitism reaches young people through both explicit slogans and coded patterns. The survey data Rosenberg cites don’t distinguish between these forms. But what makes the generational pattern so pronounced is how implicit, coded discourse provides the scalable pathway through which antisemitic frameworks become thinkable, defensible, and politically legible over time.

Coded language activates established tropes without explicit slurs: “Certain people seem to have outsized influence in media and finance” doesn’t mention Jews but triggers the same conspiratorial framework; “dual loyalty” accusations function as apparently political criticism while drawing on centuries-old patterns of Jewish othering. This implicit discourse isn’t the only form antisemitism takes online—explicit hate remains visible and often sanctioned—but implicit content is scalable, deniable, and cumulative in ways explicit hatred cannot be.

The platform-mediated conditions enabling this normalization include but extend beyond algorithms: algorithmic recommendation systems surface progressively extreme content because engagement metrics reward conflict and conspiracy; anonymity structures reduce social accountability; participatory comment cultures create mutual reinforcement dynamics where extremity becomes normalized through repeated exposure; the omnipresence of hate speech renders it ambient rather than exceptional; collapsing contextual boundaries blur news, memes, and commentary; rapid, highly emotional interaction discourages reflective evaluation; strategic ambiguity and plausible deniability allow antisemitic tropes to circulate as apparently legitimate political analysis, with creators claiming “I didn’t say Jews, I said globalists” even as audiences receive the intended message.

These conditions create environments where minority hate becomes majority-visible discourse through platform mechanics rather than demographic preference. Tim Miller’s observation—the host of the Bulwark Podcast—about students “starting to think that the Jews killed Charlie Kirk” captures the endpoint: conspiracy thinking that seems organic but results from systematic exposure to implicit framings amplified through engagement-optimizing systems.

Why Cross-Spectrum Documentation Matters

When Guardian and Sky News show comparable prevalence during the same discourse event, that tells us outlet ideology isn’t the primary driver. When variation across American outlets doesn’t track with political orientation—NBC at 5%, ABC at 6%, Fox at 10%, Forbes at 14%, CNN at 17%—that contradicts simplistic partisan explanations. This cross-spectrum presence isn’t about claiming equivalence. The manifestations differ significantly: Israel-related events trigger different antisemitic tropes than conspiracy-focused events. But the underlying mechanisms—implicit coding, algorithmic amplification, event-driven activation—operate across ideological contexts, redirecting attention from political blame games toward structural intervention points.

Event-Driven Dynamics

Israel-related conflicts trigger massive spikes in antisemitic discourse across all outlets. Financial crises activate conspiracy theories about elites and globalists. Terrorist attacks produce false flag accusations and dual loyalty tropes. The content clusters around discourse events that provide legitimizing frames for antisemitic patterns.

Young people consume content through engagement-optimized platforms where each major news event becomes an opportunity for antisemitic discourse to spread. Rosenberg identifies the loss of Holocaust memory as one contributing factor. Our research suggests this historical amnesia combines catastrophically with platform dynamics: when traditional guardrails weaken simultaneously with platforms that algorithmically amplify extreme content, antisemitism finds new pathways providing “cognitively cheap answers to complex crises.”

The Structural Reality

Both Rosenberg’s survey data and our mixed-methods discourse analysis converge on a crucial insight: antisemitism in 2025 is fundamentally a structural problem. It travels through platform architecture, not demographic essence. It activates through discourse events, not partisan ideology. It persists through implicit coding that evades traditional moderation.

Rosenberg’s framing deliberately avoids partisan blame—he shows that age matters more than political affiliation, that the problem crosses ideological boundaries, and that solutions require consensus-building rather than political victory. Our research provides the mechanistic validation for this approach. The similar rates across Guardian and Sky News audiences during discourse events tell us where to look for solutions: not at which political camp to defeat, but at which platform mechanisms to reform.

Yet the consensus Rosenberg identifies—57% of young people rejecting antisemitism—fails to mobilize effectively. Understanding why requires recognizing how antisemitism is routinely externalized along three axes, each shifting attention away from the structures that actually enable its spread.

Externalization operates outward to foreign actors—blaming bots, hostile states, or external influence operations while avoiding examination of domestic platform architecture and homegrown participatory dynamics (see also my article on X’s location feature). It operates downward to fringe extremists—framing antisemitism as marginal, exceptional, and unrepresentative while obscuring normalization in mainstream comment sections and diffusion through influencer networks. And crucially, it operates sideways to the opposing political camp—turning antisemitism into evidence of the other side’s moral failure while immunizing one’s own camp against self-examination.

This sideways externalization is perhaps the most socially corrosive because it aligns perfectly with platform incentives. Demonizing the other political side generates engagement, rewards outrage over introspection, and hardens camps rather than addressing mechanisms (see also our study on the digital aftermath of the Charlie Kirk murder). Left spaces minimize antisemitism when it appears as “anti-Zionism,” right spaces minimize antisemitism when it appears as “anti-globalism,” and both focus obsessively on the other’s worst examples. Platforms amplify this dynamic because conflict between camps drives metrics.

Antisemitism, as a hyper-adaptable hate ideology, is effectively outpacing the oversimplified binary frameworks through which public discourse is organized. While political camps argue about whether antisemitism is fundamentally a left or right phenomenon, antisemitism itself morphs across ideological contexts, exploiting whatever rhetorical opportunities each political space provides. The dichotomic framework isn’t just inadequate—it actively enables antisemitism’s spread by preventing recognition of how it operates across boundaries.

The interpretation of antisemitism is itself polarized, and that polarization becomes a vehicle for antisemitism’s persistence. This explains why Rosenberg’s finding matters so profoundly. The generational gap with no partisan gap only makes sense if antisemitism is being reproduced through shared communication environments while political camps are busy externalizing blame sideways. If antisemitism were foreign manipulation, age wouldn’t predict it; if it were fringe extremism, prevalence would be low everywhere; if it were genuinely partisan, we’d see the left-right split that’s conspicuously absent. Rosenberg’s data pattern points directly toward platform-mediated conditions as the explanatory factor—and toward sideways externalization as the reason consensus fails to translate into action.

This is simultaneously more depressing and more hopeful than partisan narratives allow. More depressing because the problem is structural rather than confined to one political camp, which could theoretically be defeated. More hopeful because structural problems admit structural solutions—interventions that don’t require changing hearts and minds one person at a time, but rather changing the systems that shape how information flows. These are the mechanisms through which Rosenberg’s “escalator” operates, and these are the points where intervention becomes possible.

From Mechanism to Intervention

Understanding how antisemitism spreads reveals where intervention points exist. If implicit antisemitism evades keyword filters, detection systems need training on actual discourse patterns. If algorithmic amplification surfaces extreme content for engagement, transparency requirements and deamplification protocols become essential. If cross-platform migration spreads coded language faster than moderation adapts, industry-wide coordination matters more than individual platform policies.

Platform literacy education becomes crucial. We need to prepare young people to recognize explicit and implicit, verbal and multimodal tropes and narratives in their actual information environments—their social media feeds, algorithmic recommendations, and comment section exposures. This means developing critical consumption skills for platform-mediated content and recognizing when legitimate political discourse slides into antisemitic territory.

The Path Forward

Antisemitism remains a minority position even among young people—57% rejected antisemitic propositions. The challenge is preventing the 43% from growing. Rosenberg’s call for leadership that strengthens consensus rather than pandering to extremes requires understanding what threatens that consensus: infrastructure and persuasion working in tandem.

We need to address both dimensions. Yes, platforms must be reformed—when they surface conspiracy theories faster than education can counter them, when coded language evades moderation, when algorithms amplify conflict, consensus erodes through structural mechanisms. But we also need direct engagement with young people in primary schools, secondary schools, and universities. Educational interventions need to account for the fact that antisemitism functions as an unusually complex and adaptive belief system—one that undermines what young people typically strive for: understanding the world at its foundations (in its Grundfesten) and clearly distinguishing between good and evil. Antisemitism offers false clarity, simple explanations for complex crises, and the illusion of moral certainty while systematically distorting both understanding and judgment.

The generational gap with no partisan gap only makes sense if antisemitism is being reproduced through shared communication environments while political camps are busy externalizing blame sideways. This is the puzzle Rosenberg’s data present, and it’s the puzzle platform-mediated communication conditions solve.

The Decoding Antisemitism expansion currently underway at NYU, the numerous activities of AddressHate, and many allied initiatives here in the US represent efforts to build infrastructure for structural intervention and policy impact. Rosenberg’s piece calls attention to an urgent problem—our research provides tools for addressing it. The algorithmic escalator is real. Building off-ramps is no longer optional—it is a structural responsibility.

AddressHate Research Scholar at NYU’s Center for the Study of Antisemitism | Lead, Decoding Antisemitism | Research Advisor, AddressHate

On Monday, beachgoers’ belongings were seen stacked on the sand at Sydney’s Bondi Beach, following the mass shooting the previous day.

Introduction

On December 14, 2025—the first night of Hanukkah—two gunmen opened fire on a Jewish celebration at Sydney’s iconic Bondi Beach. Australian authorities later identified the suspects as Sajid Akram, 50, and his son Naveed Akram, 24. Police reported the discovery of Islamic State flags in the attackers’ vehicle, though investigators have emphasized that the perpetrators’ full ideological motivations remain under investigation. According to preliminary reports, witnesses described the attackers appearing to target Jewish attendees specifically. At least 15 people were killed, including a 10-year-old child, a Holocaust survivor named Alex Kleytman, and Rabbi Eli Schlanger, who had organized the community Hanukkah event featuring a giant menorah, donuts, and a petting zoo. More than 40 others were wounded. One of the shooters was killed by police; the other remains in critical condition after being disarmed by Ahmed Al Ahmed, a Lebanese-Australian bystander who tackled him from behind and wrestled away his rifle, sustaining gunshot wounds in the process.

Australian Prime Minister Anthony Albanese declared it “an act of evil, antisemitism, terrorism that has struck the heart of our nation”—the country’s deadliest mass shooting since the 1996 Port Arthur massacre that claimed 35 lives and fundamentally transformed Australia’s gun laws. The attack came at the culmination of a year in which Jewish community organizations documented a dramatic surge in antisemitic incidents: from several hundred annually before October 7, 2023, to more than 1,700 in the past year (an increase of nearly fivefold)—a pattern of escalation that included multiple arson attacks on synagogues, Jewish schools torched in Melbourne and Sydney, and threats so pervasive that Jewish community leaders had convened in Sydney just days earlier to demand government action. As one Jewish leader stated hours after the attack: “We are not surprised by last night’s attack. We are shocked, but we are not surprised.”

The patterns of online response were grimly familiar from our previous analyses. My team and I have documented nearly identical discursive repertoires following other episodes of antisemitic violence: the widespread celebration of October 7, 2023 as liberation rather than massacre (see also DA’s Discourse Report 6); the Washington Embassy shooting in May 2025, where two Israeli diplomats were murdered and online responses mocked Jewish grief as “kvetching intensifies”; and the Charlie Kirk assassination analysis in September 2025, where YouTube comment sections became sites not of political reflection but of dehumanization, conspiracy, schadenfreude, and death wishes. What these cases reveal is not merely that antisemitism persists online—this we already know—but that it has become structurally embedded in how digital publics process Jewish suffering: not with empathy, but with antipathy, mockery, and moral inversion.

Yet what unfolded in the digital sphere following the Bondi Beach massacre was complex and, in significant ways, divergent from these previous patterns. This article offers a preliminary snapshot of that digital afterlife. Drawing on a dataset of 1,000 comments collected from five major American news outlets’ YouTube channels within the first 24 hours of the attack—CNN, NBC News, ABC News, Forbes Breaking News, and Fox News—it documents how antisemitic tropes circulated even as families waited to identify bodies and a 10-year-old Jewish child lay dead on an Australian beach. But it also documents something unexpected: a notable shift in the ratio of empathy to hatred, suggesting that certain conditions—visual framing, victim profiles, attack settings—may influence how digital publics respond to antisemitic terror.

What strikes most forcefully about the response is its fundamental asymmetry disguised as balance. Commenters engaged in three dominant modes: first, genuine expressions of empathy and mourning for the victims; second, extensive debates about gun rights, firearms ownership, and comparisons between Australian and American gun laws—a discussion we also observed following the Washington Embassy shooting in May 2025 and the Charlie Kirk murder in September 2025; and third, antisemitic discourse that ranged from conspiracy myths and justification strategies to open schadenfreude and systematic dehumanization. This three-part structure—empathy, policy debate, and hate—creates an illusion of balanced discourse, as though antisemitic conspiracy theories and celebrations of Jewish death constitute legitimate “perspectives” equivalent to expressions of grief or policy debates about gun control.

Yet despite the persistence of antisemitic content, the comment sections also hosted anti-Muslim sentiment, racist invective, and anti-democratic rhetoric. Islamophobic responses weaponized the shooters’ identities to call for collective punishment of Muslim communities, demanding deportations, and conflating all Muslims with terrorism. Racist comments extended beyond anti-Muslim and antisemitic framings to include broader xenophobic calls for ethno-nationalist purges. Anti-democratic statements ranged from calls to suspend civil liberties to authoritarian political practices. This toxic confluence—antisemitism, Islamophobia, racism, and authoritarianism circulating in the same discursive space—illustrates how hate online operates not as isolated prejudice but as an interconnected ecosystem where different forms of dehumanization reinforce and amplify one another.

This article does not aim to provide comprehensive answers. A more extensive report—in preparation—will offer deeper linguistic, multimodal, and platform-specific analysis, including comparative examination of how these patterns manifested across X (formerly Twitter), TikTok, Instagram, and other social media ecosystems in the attack’s immediate aftermath. For now, this piece seeks to capture the immediate discursive responses, identify what differed from previous cases, and explore why certain conditions may generate more empathetic responses to antisemitic terror. The findings suggest both cause for cautious optimism—that framing and context matter—and continued alarm: even under the most sympathetic conditions, antisemitic discourse achieves significant visibility, and the discursive infrastructure that normalizes Jewish death remains firmly in place.

Preliminary Reflections: A Shift in the Digital Response?

Before examining the qualitative findings, it’s crucial to understand what made this case different.

The quantitative findings reveal something more complex than simple ideological polarization. The antisemitic responses, while present across all five mainstream outlets, appeared at notably lower rates than in our previous analyses. They comprised between 5% and 17% of analyzed comments, depending on the platform—a significant presence, but markedly lower than the proportions we documented following the Washington Embassy shooting or in responses to October 7. While NBC showed the lowest proportion of antisemitic discourse (5%), CNN featured the highest (17%), with ABC (6%), Fox (10%), and Forbes (14%) falling between. The ratio of non-antisemitic to antisemitic comments ranged from 19:1 (NBC) to 4.9:1 (CNN).

This relative decline in antisemitic content—and corresponding increase in empathetic responses—demands explanation. Several factors likely contributed to this shift:

First, the role of media framing cannot be understated. The Bondi Beach attack was visually documented in ways that made the violence undeniably clear and the victims undeniably sympathetic. News coverage prominently featured images and video of families at the beach, children present at a holiday celebration, and a Holocaust survivor murdered while celebrating Hanukkah with his wife. The world saw not diplomatic staff or political figures, but ordinary families—Jews and non-Jews alike—gathered for a public celebration on one of the world’s most recognizable beaches. This framing made it significantly harder to deploy the standard antisemitic inversions that cast Jewish victims as complicit in state violence or deserving of retribution.

Second, the visual and symbolic parallels to the Nova music festival massacre on October 7, 2023, were impossible to ignore—but this time, the attack occurred in Australia, not Israel. The setting—a beach celebration, families with children, a joyous public gathering transformed into a killing field—evoked immediate comparisons to Nova. But unlike October 7, where antisemitic discourse could hide behind geopolitical framings about “occupied territory” or “resistance,” Bondi Beach offered no such cover. This was unambiguously a terror attack on a Western democracy’s soil, targeting civilians at a religious celebration. The geographic displacement may have disrupted the standard rhetorical scripts that allow antisemitism to masquerade as anti-Zionism.

Third, the nature of the victims themselves shaped responses. By contrast to the Washington Embassy shooting in May 2025, where the focus centered on two murdered employees of the Israeli diplomatic mission—individuals who could be (falsely) construed as representatives of Israeli state policy—the Bondi victims were explicitly framed as community members: a 10-year-old child, a Holocaust survivor, a rabbi organizing a neighborhood Hanukkah celebration. The Washington case allowed antisemitic commenters to collapse the distinction between individuals and institutions, treating the diplomats as legitimate targets within an imagined geopolitical conflict. The Bondi framing made such rhetorical moves far more difficult. At least, this framing dominated the immediate media coverage and appears to have influenced the early comment section responses.

Fourth, platform and outlet dynamics likely played a role. YouTube’s moderation practices, the mainstream standing of the selected outlets, and the specific audiences these channels attract may have contributed to the lower rates of antisemitic content compared to platforms like X (formerly Twitter) or Telegram, where our preliminary monitoring suggests significantly higher concentrations of antisemitic responses. Additionally, the first 24-hour window may have captured a moment before more organized antisemitic campaigns mobilized, as we observed in later stages of the Washington and Charlie Kirk cases.

Yet the Bondi case also suggests that this normalization of antisemitic discourse is not absolute or inevitable. When visual framing emphasizes victim humanity, when geographic context disrupts standard geopolitical scripts, and when media coverage centers families rather than institutions, digital publics may respond with greater empathy—though antisemitic discourse never fully disappears.

Dataset

This preliminary analysis draws on a dataset of 1,000 user comments collected from the YouTube channels of five major American news outlets within the first 24 hours following the December 14, 2025, attack:

1. CNN – “Gunshots reported at Bondi Beach in Sydney.”

2. ABC News – “15 killed in Australia attack on Jewish community.”

3. NBC News – “Special Report: Targeted shooting at Australia’s Bondi Beach leaves multiple dead.”

4. Forbes Breaking News – “Mass Shooting At Hanukkah Celebration Near Sydney, Australia Leaves Several Dead And Injured.”

5. Fox News – “Eyewitness to Australia mass shooting recalls horror: ‘PANDEMONIUM broke out’.”

For each outlet, the first 200 comments appearing chronologically were collected, capturing immediate public reactions before narratives solidified or subsequent news cycles introduced new information.

Quantitative Breakdown:

• CNN: 200 comments → Antisemitic: 17.0% | Ratio 4.9:1

• ABC News: 200 comments → Antisemitic: 6.0% | Ratio 15.7:1

• NBC News: 200 comments → Antisemitic: 5.0% | Ratio 19:1

• Forbes: 200 comments → Antisemitic: 14.0% | Ratio 6.1:1

• Fox News: 200 comments → Antisemitic: 10.0% | Ratio 9:1

Total corpus: 1,000 comments analyzed across five outlets.

Method

The analysis employed a mixed-methods approach combining quantitative coding with qualitative multimodal discourse analysis using MAXQDA software. The analytical framework draws from the Decoding Antisemitism Lexicon (Becker et al., 2024), which systematizes contemporary antisemitic discourse, including explicit hate speech and implicit strategies such as coded language, sarcasm, and rhetorical inversions. Comments were categorized using the Decoding Antisemitism taxonomies, distinguishing between explicit antisemitism, contextual antisemitism requiring interpretive knowledge, and non-antisemitic content.

Key analytical categories include: conspiracy myths, denial of antisemitism, inversion of victimhood, affirmation of violence, instrumentalization of antisemitism, Holocaust distortion, schadenfreude, and Jewish control narratives.

An intercoder reliability test was conducted among Decoding Antisemitism research team members, with disagreements resolved through collective discussion. The goal is not merely to count antisemitism but to understand the discursive ecosystem in which antisemitic violence is processed, rationalized, and—too often—celebrated.

Empirical Findings: The Discursive Repertoires

The antisemitic repertoires themselves, though less prevalent than in previous cases, still displayed remarkable sophistication and ideological coherence. Commenters did not simply express hatred—they performed it, weaponized it, and embedded it within recognizable political narratives that transform antisemitism from crude prejudice into ostensibly justified critique. The discursive strategies identified include:

1. Conspiracy Narratives

Conspiracy myths constituted the most pervasive antisemitic strategy, encompassing both false flag theories and broader narratives of Jewish global control. These fell into several subcategories:

a) Mossad and Israeli Intelligence Orchestration

Many comments attributed the attack directly to Israeli intelligence operations designed to generate sympathy or manipulate public opinion:

“Mossad!” (Fox)
“Israeli mossad killing Jews as usual. Nothing special 😏” (CNN)
“Obvious Mossad False Flag Operation. Google Trends reveals hundreds of searches for Naveed and Sajid Akram hours before this tragedy actually occurred in both Israel and Iran.” (ABC)
“Sounds like an Israeli operation. Why do the Jews manage to screw up the false flags?” (ABC)

b) Jewish Control

The most elaborate conspiracy narrative invoked classic antisemitic tropes of Jewish financial control, listing prominent Jewish figures across finance, media, and technology:

“The International J’‘’‘ish Trillionaire Oligarchy that owns & controls Western Finance: Rothschild, Warburg, Oppenheim AND the billionaire CEOs: Laurence Fink (Blackrock), Frank Dimon (JP Morgan/Chase), Bourla (Pfizer), Bancel (Moderna) Zuckerberg (Facebook) or others A.G. Sulzberger (New York Times) Susan Wojecicki (former ceo of YouTube pushed the COVID Jab and unalived herself from it - it would seem -- high positive correlation to ‘turbo’ cancer) Larry Page & Sergie Brin (Google founders). Or histories genocide orchestrators USSR Kaganovich the Holodomor, Yagoda, Beria, or today’s Netanyhu, Gallant, Ben-Gvir. The list is endless...” (Forbes)

This comment exemplifies the networked structure of contemporary conspiracy thinking, weaving together financial elites, media moguls, pharmaceutical executives, historical Soviet perpetrators, and current Israeli political leaders into a single, transhistorical narrative of Jewish orchestration. The commenter’s list-making strategy—presenting an apparently endless catalogue of Jewish names—performs the antisemitic trope of Jewish omnipresence and omnipotence.

c) Great Replacement and Jewish Orchestration of Civilizational Decline

Great Replacement narratives positioned Jews as orchestrators of Western demographic transformation and civilizational decline:

“ironic that the Jews were the ones pushing Islam on the west with all their ngos” (ABC)

A: “The Authorities of Australia have failed in their duties. Projecting their Citizens.”
B: “But they haven’t failed their masters the Jw globalist” (Forbes)

These comments deploy the Great Replacement conspiracy theory—originally a far-right claim that elites are deliberately replacing white Western populations with non-white immigrants—but here explicitly attribute this alleged replacement to Jewish NGOs and “globalist” elites. The framing positions the Bondi Beach attack not as antisemitic terrorism but as a predictable consequence of Jewish-orchestrated immigration policies, thereby inverting responsibility from perpetrators to victims.

d) Generic False Flag Theories

Non-antisemitic conspiracy theories also circulated, framing the attack as a generic false flag operation without specifically invoking Jewish orchestration:

“False flag for sure” (Forbes)
“The false flags will continue until morale improves” (ABC)
“This smells like false flag/psy op. Expect the government to take away whatever illusion of freedom you have.” (CNN)

While these comments do not explicitly invoke antisemitic motifs, they contribute to a broader conspiratorial climate in which all expressions of Jewish victimhood are rendered suspect. The generic false flag narrative shares the same epistemic structure as antisemitic conspiracies—a refusal to accept official accounts, a presumption of hidden manipulation—and frequently serves as a gateway to more explicitly antisemitic framings.

2. Inversion of Victimhood and Justification

Inversions of victimhood reframed Jewish deaths as karmic consequences of Israeli state actions, denied the existence of antisemitism by recasting it as anti-Zionism, or explicitly justified the violence as a proportional response. This repertoire constitutes one of the most ideologically powerful antisemitic strategies: it simultaneously acknowledges Jewish suffering and denies its moral significance by attributing blame to Jews themselves.

“The genocide in Palestine continues to put Jewish lives at risk around the world. People see genocide committed under a flag with the star of David and easily become radicalized to use a level of violence usually reserved for the IDF.” (ABC)
“Genocide has global consequences” (NBC)
“[One of the perpetrators is] Probably a survivor of their genocide” (Fox)
“Glad at least you are shedding tears for these baby killers.” (Fox)
“😂😂 you guys killed a million kids in Gaza ?? What anti semitism you talking about about u mean anti satans 😂😂” (CNN)

These comments perform a double rhetorical move: they attribute collective guilt to all Jews for Israeli state actions (”you guys killed a million kids”), and they deny the category of antisemitism entirely by reframing it as justified political opposition (”anti satans”). The logic is totalizing—Jews are not victims of hatred but perpetrators facing consequences.

Users also invoked numerical comparisons to Gaza casualties to diminish the significance of Jewish deaths:

“1,000 dead in Gaza every day...” (CNN)
“100,000 palestinian children..” (CNN)

This comparative victimology establishes a moral hierarchy in which Jewish suffering is rendered negligible against Palestinian losses. The numbers themselves—wildly inflated—are less important than their rhetorical function: to delegitimize Jewish grief and establish that Jews, as a collective, have forfeited any claim to sympathy.

Others framed the attack through irony, implying that the shooters behaved like the Israeli army during the Gaza war when ceasefires were declared:

“Just an Israeli cease fire, nothing to see here...” (Forbes)
“Oh, so it’s an Israeli cease fire.” (NBC)

This dark humor weaponizes accusations of Israeli ceasefire violations to mock Jewish deaths. The implication is that what Israel allegedly does to Palestinians—killing during supposed ceasefires—is now being done to Jews, and this constitutes poetic justice rather than tragedy.

A particularly revealing exchange illustrates how justification discourse operates through layered affirmations:

A: “It’s not a mass shooting. It’s an ‘Israeli ceasefire.’”
B: “So killing Jews around the world is justified? What are you trying to say here bud.”
C: “hes saying...if you give no quarter, you shall recieve none. Pretty simply, actions have consequences for a society. Tiny hats are on borrowed time with humanity.”
D: “FREE PALESTINE!” (Forbes)

Here, Commenter B attempts to challenge the distortion and indirect demonization of Israel’s tactics implicit in A’s remark—the false equivalence comparing the IDF with terrorists who massacre civilians at a Hanukkah celebration. Commenter C justifies antisemitic violence by invoking the notion of karma: Jews, as a collective, will continue facing collective punishment for Israeli state actions. Commenter D’s interjection—”FREE PALESTINE!”—functions as an endorsement of this collective guilt framework, treating the murder of Australian Jews as part of Palestinian liberation.

One comment acknowledged the tension between mourning and justification, though even this concession is hedged:

“The killing of innocents can’t be justified, and this apply to the government of Israel too.” (Forbes)

By juxtaposing condemnation of the Bondi attack with immediate criticism of Israel, this comment suggests a moral equivalence between antisemitic terror and Israeli military actions, thereby subtly undermining the act of mourning it ostensibly performs.

3. Affirmations and Celebrations of Violence

Explicit affirmations and celebrations of violence ranged from mocking emojis and popcorn GIFs to direct expressions of joy and dehumanizing dismissals. This category is particularly significant because it moves beyond justification (violence as an understandable consequence) to endorsement (violence as a desirable outcome). The affective register shifts from reasoned critique to schadenfreude, glee, and eliminationist fantasizing.

A: “It was a HANUKKA menorah lighting.”
B: “😂” (NBC)

This exchange, cited in the article’s title, exemplifies the schadenfreude that pervaded responses. Commenter A deploys a macabre double meaning: “lighting” ostensibly refers to the Hanukkah menorah lighting ceremony, but functions as dark wordplay evoking “lighting up” victims with gunfire—transforming the sacred Jewish ritual into a punchline about mass murder. Commenter B’s laughing emoji affirms this sick joke, celebrating the twisted connection between religious observance and massacre.

“I’m eating popcorn right now waiting😂” (Forbes)

The popcorn emoji and its variations have become a standard signifier of schadenfreude in online culture—a performative declaration that one is entertained by others’ suffering. The act of “waiting” implies anticipation of further violence or revelations, treating the massacre as a spectacle.

A range of affirming, partly multimodal statements reinforced this celebratory tone:

“🎉 🍿” (Forbes)
“😂” (NBC)
“😂😂 i am very happy” (CNN)
“very nice” (CNN)
“i am too over joyed” (CNN)
“share with you the same sentiment” (CNN)
“yes.. time to rejoice” (CNN)

These responses demonstrate the role of emojis and brevity in digital hate speech. The laughing emoji, party popper, and popcorn become linguistic shorthand for celebration, allowing users to express pleasure at Jewish death without verbose justification. The exchange on CNN—where multiple users affirm one another’s joy—illustrates how antisemitic schadenfreude operates as a social bonding mechanism, creating in-group solidarity around shared hatred.

Sarcastic humor further weaponized the attack:

“any zionists among them?” (CNN)
“They had human shields” (Fox)
“Just some 4x2’s getting whacked, no biggie!” (Fox)

The first comment sarcastically inquires whether “Zionists” were among the dead, implying that killing Zionists (a term used here to dehumanize Jews who support Israel’s existence) would make the massacre worthwhile or justified. The “human shields” comment inverts accusations frequently leveled at Hamas—that it uses Palestinian civilians, especially children, as human shields—to sarcastically suggest that Australian Jews now find themselves in the same position as Palestinians in Gaza. This darkly ironic reversal mocks Jewish victimhood while implying moral equivalence between Hamas’s alleged tactics and the murder of Jews at a Hanukkah celebration. The final comment—”Just some 4x2’s getting whacked, no biggie!”—could be interpreted as a reference to Jews, and in any case represents schadenfreude and dismissive mockery given the tragic nature of the event, trivializing mass murder as inconsequential.

4. Demonization

Some comments deployed explicit demonization, equating Jews or Israel with satanic or evil forces. This strategy draws on longstanding Christian antisemitic traditions that cast Jews as agents of the devil, now secularized into generic accusations of “evil.”

A: “My God! These evil people, just want to take out lives like that”
B: “Israel? Yes they do” (ABC)

A: “Its the Religion of Peace! (referring to Islam)”
B: “Israel? No. They’re not into peace.” (Fox)

“Insane Devil Force” (Forbes) [referring to the IDF (Israeli Defense Forces)]

These exchanges illustrate a rhetorical tactic in which seemingly neutral condemnations of violence are immediately redirected toward Israel or Jews. Commenter A’s lament about “evil people” is co-opted by Commenter B to indict Israel, collapsing the distinction between the antisemitic perpetrators and their Jewish victims. Similarly, the sarcastic “Religion of Peace” jab aimed at Islam is deflected back onto Israel, positioning Jews as the true threat to peace. The “Insane Devil Force” comment explicitly demonizes the Israeli military, invoking satanic imagery to delegitimize Israel’s right to self-defense and casting Israeli Jews as demonic actors.

5. Affirmations of Violence Through Political and Religious Slogans

Several comments used the attack as an occasion to affirm pro-Palestinian slogans without additional context. While expressions of solidarity with Palestinians are not inherently antisemitic, their deployment immediately following a massacre of Jews—and in the absence of any acknowledgment of the victims—transforms them into implicit endorsements of the violence.

“FREE PALESTINE!” (ABC & NBC)
“Allahuakbhar” (Forbes)

The “FREE PALESTINE!” slogan, repeated verbatim by the Bondi Beach shooter during the attack, takes on a chilling resonance when echoed in comment sections celebrating or justifying the massacre. Its appearance functions not as a call for Palestinian liberation but as a signal of approval for antisemitic terror.

Similarly, “Allahu Akbar” (”God is greatest”)—a phrase with profound religious significance in Islamic practice, used in daily prayers and moments of spiritual reflection—becomes weaponized when deployed in comment sections responding to mass murder. While the phrase itself is not inherently violent, its use here in the immediate aftermath of an Islamist terror attack that killed 15 Jews transforms it into a celebratory affirmation of the violence. The commenter is not invoking the phrase in its devotional context but rather echoing the battle cry associated with jihadist attacks, signaling solidarity with the perpetrators rather than mourning for the victims.

One exchange acknowledged the tension between mourning Jewish victims and Palestinian casualties:

A: “My God! These evil people, just want to take out lives like that”
B: “Israel? Yes they do”
C: “Come on, bro. I’m against the war against Palestinians but please let us mourn these people.”
B: “I also mourn for innocent people dying… not cool at all… we all need to be better” (ABC)

While this exchange might initially appear as empathetic pushback, Commenter B’s response subtly performs the same moral inversion prevalent throughout the corpus. By responding to a request to mourn murdered Jews with “I also mourn for innocent people dying,” Commenter B implicitly equates Jewish families massacred at a beach celebration with Palestinian casualties in Gaza—a war zone where Israel is combating a terror organization following the October 7 massacre. The phrase “we all need to be better” further distributes moral responsibility equally, as though those mourning antisemitic terror bear an equivalent moral burden to those conducting it. So, without questioning that the loss of Palestinian lives is tragic, this framing treats the Bondi massacre and Gaza casualties as morally equivalent tragedies requiring equal mourning—collapsing the distinction between deliberate antisemitic terror and the complexities of armed conflict, thereby undermining the very request for Jewish-specific mourning it appears to accommodate.

6. Death Wishes and Eliminationist Rhetoric

Several comments explicitly articulated eliminationist intent, moving beyond celebration of past violence to anticipation or advocacy of future violence. One particularly revealing exchange exposed the genocidal logic underlying this discourse:

A: “Targeting non-Israeli Jewish civilians […] feeds the Zionist narrative that Jews around the world can only be safe if there’s a Jewish state […]. These anti-Jewish attacks only make Israel stronger”
B: “Fine. Let them all pile into one spot.” (NBC)

Here, Commenter A acknowledges that antisemitic attacks on diaspora Jews will backfire strategically—they will drive Jews to identify more strongly with Israel and validate the argument for a Jewish state as a necessary refuge. The commenter recognizes that such terror cannot succeed in weakening Jewish communities or the state of Israel. Yet Commenter B’s response reveals a far darker eliminationist fantasy: the desire for all Jews worldwide to concentrate in Israel (”pile into one spot”) so they can be extinguished en masse. This is not strategic anti-Zionism but genocidal antisemitism—the ultimate expression of the centuries-old fantasy of gathering Jews into one place for total annihilation.

Another comment invoked apocalyptic imagery:

“Tiny hats are on borrowed time with humanity.” (ABC)

This formulation suggests that Jews as a collective are facing imminent elimination—that their time among humanity is limited and contingent. The phrasing “borrowed time” implies both inevitability and deserved punishment, positioning Jewish existence itself as temporary and illegitimate.

Both comments exemplify how online antisemitic discourse can shift from reactive hatred (celebrating existing violence) to proactive incitement (fantasizing or advocating future attacks). They transform antisemitism from passive sentiment into active threat, creating a discursive environment in which violence against Jews is not only rationalized but anticipated and desired.

Synthesis

These patterns are not spontaneous outbursts but structured responses drawing on centuries-old antisemitic repertoires, now circulated at digital speed, amplified by algorithmic logics that reward engagement over accuracy, and normalized in spaces where the moral ground for rejecting antisemitism has eroded. The Bondi Beach case reveals both continuity and change: the same discursive strategies we have documented in previous analyses remain operative—conspiracy myths (including false flags, Jewish control narratives, and Great Replacement theories), victimhood inversions, schadenfreude, dehumanization—but they appear at lower overall rates and compete more visibly with expressions of genuine empathy and solidarity.

This variation likely reflects multiple factors: the mainstream standing of YouTube as a platform compared to more toxic spaces like X or Telegram; the moderation practices of these specific news outlets; the visual framing of the attack emphasizing families and children rather than diplomatic or political figures; and the symbolic resonance with the Nova massacre, transposed to a Western democratic context where geopolitical excuses for antisemitism become harder to deploy.

The findings suggest both cause for cautious optimism—that framing and context matter, that certain conditions can shift the balance toward empathy—and continued alarm: even under the most sympathetic conditions, antisemitic discourse achieves significant visibility, circulates alongside other forms of hate (Islamophobia, racism, anti-democratic rhetoric), and the infrastructure for normalizing Jewish death remains firmly in place. The digital afterlife of antisemitic terror is not an afterthought—it is a site of active antisemitic production that shapes, legitimates, and potentially incites future violence.

Looking Ahead

This preliminary analysis is part of a broader effort to understand how hate spreads across digital platforms. The full report—currently in preparation—will expand this analysis across X (formerly Twitter), TikTok, Instagram, and other ecosystems, revealing how different platform architectures and moderation practices shape the conditions under which antisemitic discourse spreads or is contained.

At NYU’s Center for the Study of Antisemitism, we are developing—in close collaboration with AddressHate—the methodological infrastructure necessary for this work: combining computational detection at scale with the qualitative depth required to decode implicit meanings, coded language, and cultural context. Without such tools, we cannot detect early warning signals of radicalization, map how elite ambiguity cascades into explicit hate, or develop interventions that distinguish legitimate critique from weaponized bigotry.

Social media offers unprecedented access to real-time attitude formation across millions of users. Understanding these patterns can illuminate not only the present but the future—revealing where harm is likely to emerge and how to intervene effectively.

Only through rigorous interdisciplinary research can we decode the discourses shaping our digital age and rebuild the moral grammar of democratic society—one that can gain greater immunity against the invisible advance of polarization, radicalization, and hate.

End of preliminary analysis. Full report forthcoming.

AddressHate Research Scholar at NYU’s Center for the Study of Antisemitism | Lead, Decoding Antisemitism | Founding Research Advisor, AddressHate

Image: Google DeepMind

A landmark study provides some of the clearest causal evidence to date that platform architecture drives political hostility. The policy response reveals how wide the gap remains between what research shows and what gets addressed.

Introduction

A breakthrough study published in Science this November provided some of the clearest causal evidence to date that social media algorithms shape political polarization. Researchers at Stanford built a browser extension that independently reranked users’ feeds in real time, demonstrating that algorithmic choices directly altered users’ political attitudes by amounts comparable to several years’ worth of polarization change in long-term US surveys.

For researchers who study online radicalization, the findings confirmed what extensive observational evidence had long suggested: platform architecture—what gets amplified, how content is ranked, which posts surface first—directly shapes democratic outcomes. The study was methodologically rigorous, causally clean, and measured effects that matter.

For the first time, independent researchers demonstrated algorithmic causation without requiring access to platform infrastructure—a milestone in platform accountability research.

Yet the policy conversation remains largely elsewhere.

Policy attention has focused instead on symbolic fixes, geopolitical narratives, and interventions that are politically legible, easier to implement, and do not require confronting platform business incentives. Around the same time the Stanford study appeared, lawmakers from both parties welcomed X’s new location feature, which displays the country from which an account operates, framing it as a step toward combating foreign influence and online antisemitism.

The Stanford study should reshape the conversation about platform accountability. Instead, the policy response that followed reflects a familiar pattern: research identifies mechanisms; policy focuses on symptoms.

This pattern repeats across digital governance. Breakthrough research isolates the forces shaping democratic outcomes. Policy attention gravitates toward solutions that fit existing narratives about external threats, require minimal structural change, and avoid examination of the architectures that enable radicalization.

The result is a widening gap between what we know works and what we’re willing to demand.

What the Research Actually Shows

The Stanford study, led by researchers Tiziano Piccardi, Martin Saveski, and Chenyan Jia, represents both a methodological and substantive breakthrough. Working without platform permission, they built infrastructure to intercept and rerank X feeds in real time, using large language models to identify content expressing “antidemocratic attitudes and partisan animosity” (AAPA)—including partisan hostility, support for undemocratic practices, opposition to bipartisanship, and biased evaluation of facts.

Over 1,256 participants received either increased or decreased exposure to AAPA content for one week during the 2024 presidential campaign. The effects were measured through surveys embedded directly in users’ feeds and post-experiment assessments.

Reducing AAPA exposure led participants to feel warmer toward the opposing party by 2.11 degrees on a 100-point scale. Increasing exposure caused a symmetrical cooling of 2.48 degrees. Critically, Democrats and Republicans responded identically to the interventions—the effects were bipartisan. These are not huge shifts for any single individual, but applied to millions of users over months or years, they translate into substantial structural tilts toward hostility.

Moreover, 74% of participants reported noticing no impact on their experience. Algorithmic effects operate below conscious awareness, shaping attitudes without users recognizing the intervention.

The study demonstrates what years of correlational research could not: that algorithmic ranking decisions directly shape political attitudes. The content remained the same. The users remained the same. Only the ranking changed—and polarization moved accordingly.

Three Drivers—And Why Only One Gets Policy Attention

To understand why this evidence matters—and why policy responses miss the mechanism—we need a framework for how antisemitism and antidemocratic hostility actually spread online.

My research, analyzing over 300,000 web user comments across major news channels using qualitative discourse analysis supported by computational methods, reveals three distinct but interconnected drivers:

• First: Malicious actors. Foreign and domestic campaigns deliberately inject polarizing content. State-backed influence operations and coordinated disinformation networks exist. This is the driver dominating political rhetoric and policy attention.

• Second: Algorithmic amplification. Platform ranking systems, optimized for engagement, disproportionately surface emotionally intense content. Outrage drives clicks. Hostility generates shares. Algorithms don’t evaluate content quality or social consequences—they optimize for metrics that happen to correlate with polarization. The Stanford study provides causal evidence for this mechanism.

• Third: Communication conditions. The socio-technical environment of online spaces—such as anonymity that removes social accountability, mutual reinforcement that validates extreme positions, and omnipresent hate speech that normalizes extremity—fundamentally shapes how users behave and what content they produce.

The Stanford study measured exposure effects but did not examine how these conditions radicalize the ordinary users who create the content algorithms then amplify.

These three drivers interact dynamically. Malicious actors inject content. Algorithms amplify it. Communication conditions radicalize ordinary users who then produce more extreme content—which algorithms amplify further, reaching new users who become radicalized under the same communication conditions.

The Stanford study isolates the second driver and proves it matters causally. But understanding all three is essential because policy interventions typically address only the first—and often in ways that miss even that mechanism.

Why Geographic Solutions Miss All Three Mechanisms

The X location feature exemplifies how policy attention focuses on one driver while ignoring the others entirely.

The feature addresses exclusively malicious actors, specifically foreign ones. It assumes that knowing where an account is located helps identify bad actors and discount their content accordingly.

This assumption is flawed on multiple levels. Geographic origin tells us nothing about intent, coordination, or impact. An account in Eastern Europe may be an ordinary person with opinions about American politics. An account in Iowa may be part of a domestic influence operation. Location and malicious intent are orthogonal categories.

More fundamentally, focusing on foreign accounts ignores where most high-engagement polarizing content actually originates. In our analysis, antisemitic discourse surged to 36-38% of comments on major UK news outlet YouTube channels following the October 7 attacks. After the antisemitic hate crime in Washington in May 2025, such content averaged 43% across major English-language news channels, reaching 66% on some outlets.

While we cannot completely exclude some automated activity, the idiom, references, and interaction patterns overwhelmingly indicate domestic participation. High-profile domestic actors—politicians, media figures, influencers—routinely traffic in strategically ambiguous statements that seed problematic interpretations. Digital intermediaries sharpen and amplify these messages. Comment sections collapse the ambiguity into explicit hate speech.

Foreign interference exists and deserves scrutiny—but in the datasets we examined, domestic dynamics overwhelmingly drove the volume and intensity of antisemitic expression.

The implication is clear: even if the location feature perfectly identified every foreign account, it would leave the dominant source of polarizing content completely untouched.

What the Stanford Study Reveals—And What It Doesn’t

The Stanford research provides strong causal evidence for algorithmic amplification. The experiment demonstrated that changing how content is ranked changes political attitudes, even when the content itself remains constant. This is precisely the kind of mechanistic evidence that should drive policy.

But the study also reveals, by design, what remains unexamined: communication conditions.

The researchers showed that exposure to AAPA content increases polarization. They measured what happens when algorithms amplify certain material. What they could not measure was how and why that content emerged in the first place.

The AAPA posts that participants encountered were produced by users operating under specific communication conditions. Some content likely reflected what users would express in any context. But much was probably shaped by anonymity (enabling transgressive expression), mutual reinforcement (validating extreme positions), and normalized hostility (making extremity seem ordinary).

Communication conditions produce radicalized content. Algorithms then amplify it. This reaches users who are further radicalized by those same conditions, producing more extreme content in a self-reinforcing cycle.

The Science study isolates one mechanism in a much larger system. It shows, causally, that algorithmic amplification shapes political attitudes even when the underlying content and users remain constant. But amplification does not operate in isolation: it interacts with malicious actors who inject polarizing material and with communication conditions that shape what ordinary users express.

Location labels are framed as a response to malicious actors, yet they capture only the purported geography of an account—a poor proxy for the coordinated behaviors that actually matter. And crucially, they do nothing to address what the Science study demonstrates: how ranking systems elevate emotionally intense content and thereby shift attitudes at scale. In short, the policy response targets the most symbolically convenient driver while leaving the empirically important one untouched.

What Evidence-Based Accountability Would Require

If policymakers took the Stanford findings seriously, they would demand fundamentally different interventions.

For algorithmic amplification:

• Mandatory transparency into recommendation systems and ranking algorithms

• Independent researcher access without platform gatekeeping

• Regulation of engagement-optimization that profits from polarization

• Required testing of alternative ranking approaches, prioritizing democratic values

The Stanford researchers had to build their own infrastructure to study algorithmic effects. This shouldn’t be necessary. Platforms should be required to enable independent auditing of core ranking systems.

For communication conditions:

• Foundational: fine-grained multimodal discourse analysis.
  Before diagnosing communication conditions, we need a detailed empirical understanding of the discourse itself. This requires systematic, multimodal analysis of online hostility—its lexicon, metaphors, frames, narrative arcs, escalation patterns, and audience uptake over time. Without understanding how hate is communicated, we cannot meaningfully intervene in the conditions that enable it.

• Link communicative patterns to platform affordances.
  These patterns must then be connected to specific design features—ranking systems, recommendation engines, visibility mechanics, reply structures, quote-tweets, stitching, duets, and comment threading—that amplify conflict and facilitate in-group reinforcement. Affordances are not neutral; they shape discursive behavior.

• Design interventions targeting mutual-reinforcement loops.
  Once escalation points are empirically mapped, platforms can introduce interventions that disrupt the loops through which hostility is validated and intensified—dogpiling dynamics, outrage cascades, quote-wars, and recursive comment spirals.

• Address how anonymity shapes hostile discourse.
  Debates about anonymity should be grounded in observed communicative practices, not assumptions. Research consistently shows divergent effects: pseudonymity can enable performative transgression, whereas stable identity cues introduce reputational constraints.

• Introduce friction mechanisms along radicalization pathways.
  Deploy targeted friction—delays, prompts, contextual warnings, confirmation steps—at moments where discourse transitions from disagreement to dehumanization or annihilationist rhetoric. The aim is not censorship but interruption of momentum.

• Develop systems that surface normalization of extremity.
  Create user-facing indicators that reveal when conversations drift into patterns associated with dehumanization, conspiracism, or incitement. These signals translate discourse analysis into real-time literacy tools for users.

Without intervening at the level of communicative dynamics, platforms will continue to produce the very content that amplification systems learn to elevate.

For malicious actors:

• Detection focused on behavioral patterns, not geography

• Recognition that domestic actors often matter more than foreign ones

This requires rethinking fundamental platform architecture. But without addressing communication conditions, we leave intact the environment that continuously produces content for algorithms to amplify.

Why the Gap Persists

Why does policy gravitate toward interventions like the location feature rather than what evidence suggests would work?

The answer reveals the political economy of platform governance. The location feature requires almost nothing from platforms: add a label, done. No change to business models. No transparency into proprietary systems. No examination of how design enables radicalization.

Real accountability would require changes that platforms resist because:

• Transparency into recommendation systems threatens competitive advantage and reveals engagement-optimization mechanics

• Independent researcher access creates legal and PR risks

• Regulating engagement optimization could reduce user time and advertising revenue

• Redesigning communication affordances requires significant engineering resources

The contrast between what is easy for platforms and what is necessary for democracy explains the persistent gulf between research and regulation.

Some lawmakers may genuinely believe location labels increase transparency. But in practice, these measures function as accountability theater—the appearance of action without institutional change.

This dynamic is bipartisan. Republicans can condemn foreign influence without examining how right-wing media ecosystems mainstream conspiracy theories. Democrats can criticize online hate without confronting how some progressive rhetoric traffics in problematic tropes. Both parties position themselves as defenders against external threats while avoiding harder questions about domestic discourse, platform business models, and communication architecture itself.

Closing the Gap

The celebration of X’s location feature, in the same week that breakthrough research demonstrated what actually drives polarization, encapsulates the persistent gap between what research shows and what policy addresses.

We now have causal evidence that algorithmic amplification matters. We have extensive empirical evidence that communication conditions—anonymity, mutual reinforcement, omnipresent hate speech—radicalize ordinary users who produce the content algorithms amplify. And we have clear data showing that domestic actors and participatory dynamics drive most high-engagement polarizing content.

Against this evidence, focusing on account geography represents more than a missed opportunity. It represents a fundamental misunderstanding of how online radicalization operates.

Online antisemitism and antidemocratic hostility are not primarily imported threats. They are produced through the interaction of algorithms that reward polarization, communication conditions that enable radicalization, and malicious actors—often domestic—who exploit these dynamics.

If we’re serious about addressing online radicalization, we need interventions aimed at mechanisms, not geography. We need transparency into algorithmic systems. We need research into how platform affordances shape radicalization. We need policies addressing all three drivers, not just the politically convenient one.

The location feature tells us where an account is. The research tells us what actually matters: what content gets amplified, and how communication conditions shape radicalization. One is trivial to implement and largely ineffective. The other is difficult to address and essential to democracy.

The Stanford study demonstrated what becomes possible when research operates independently of platform gatekeeping. The policy response demonstrates how far we remain from demanding the accountability that evidence requires.

The evidence is clear. The question is whether our policies will finally catch up.

AddressHate Research Scholar at NYU’s Center for the Study of Antisemitism | Lead, Decoding Antisemitism | Founding Research Advisor, AddressHate

Image: Chen Te

We focus obsessively on where accounts are located—ignoring the platform structures and online conditions shaping what we see.

Introduction

A few days ago, as Jewish Insider reported, lawmakers from both parties celebrated what they characterized as a breakthrough in combating online antisemitism. X’s new location feature—which displays the country from which an account operates—had begun revealing that some accounts engaging in U.S. political discourse and spreading antisemitic content were based abroad. For Republican and Democratic legislators alike, this represented a kind of digital unmasking. “The evidence is insightful,” Rep. Don Bacon (R-NE) told Jewish Insider, noting how the feature exposed “foreign interests trying to spread antisemitic poison” while “masquerading as Americans.”

The bipartisan response was swift and enthusiastic. Sen. James Lankford (R-OK) argued that Americans “deserve to know which accounts are run from abroad so we know the true source of these narratives.” Rep. Tom Suozzi (D-NY) framed the issue in geopolitical terms: “Beijing, Moscow and Tehran know they cannot defeat us economically or militarily, so they exploit controversial issues, like Israel and antisemitism, and try to divide.” Former Ambassador Nikki Haley called the location feature “a huge win for transparency and American security,” urging other platforms to follow suit.

The narrative is compelling: foreign adversaries weaponize antisemitism to divide Americans, and transparency about account origins helps us identify and counter these external threats. And there is truth to this concern—malicious actors do exploit divisive issues. But the bipartisan celebration of X’s location feature reveals a dangerous pattern of misdiagnosis.

As the lead of the Decoding Antisemitism project—which has analyzed more than 300,000 items of digital content across multiple crises—I have observed three distinct but interconnected drivers of antisemitic radicalization online: coordinated malicious actors (both foreign and domestic), algorithmic amplification by platform architectures, and homegrown participatory dynamics enabled by the conditions of online communication itself—anonymity, mutual reinforcement, and the normalization of extremity through constant exposure.

The problem with the current discourse is not that it identifies foreign influence—that factor is real. The problem is that it treats foreign influence as the primary or sole driver while ignoring the structural and domestic conditions that allow antisemitic narratives to take root and spread. The way we diagnose the problem shapes how we design solutions for it—and misdiagnosis doesn’t just limit our response, it actively misdirects resources, attention, and political will away from the factors we can actually control.

Understanding Antisemitism as a Digital Pathogen

Understanding online antisemitism requires a virological approach: examining not just the geographic origin of accounts, but how hate narratives mutate, which platform features enable their transmission, and what environmental conditions allow them to thrive and spread. Yet social media platforms remain essentially a black box—we lack systematic tools to understand the dynamics unfolding within these digital spaces. Lawmakers celebrate a feature that reveals account geography while leaving the actual black box—algorithmic recommendation systems, engagement optimization, and content amplification—completely unexamined.

The recurring discursive pattern in the lawmakers’ response is one of externalization of moral responsibility. What is striking is not only what is said, but what is absent: there is no mention of algorithmic amplification, platform incentives, the broader attention economy, or the deep roots of homegrown polarization and hate. Instead, the rhetoric follows a predictable sequence—Externalization → Deflection → Moral Simplification.

The framing positions antisemitism primarily as something imported from abroad. This narrative allows lawmakers to shift responsibility outward and to portray American society as a largely innocent recipient of “poison,” rather than as an active site where antisemitic ideas are produced, circulated, and normalized. Rep. Bacon made this logic explicit: “On one hand, I’m glad much of the antisemitism poison is not coming from the U.S.”

But this relief is misplaced. While foreign malicious actors do contribute to the problem, focusing exclusively on them obscures how American political culture, media ecosystems, platform architectures, and the participatory dynamics of online communication create the conditions for antisemitism to flourish domestically.

The Conflation Problem

The Jewish Insider article—and the lawmakers quoted within it—collapse “foreign,” “adversarial,” and “antisemitic” into a single analytic category. This conflation obscures more than it reveals.

It treats geographic origin as dispositive of intent and impact. An account based in South Asia or Eastern Europe engaging with U.S. politics is not necessarily a state-directed influence operation. It may simply be a person with opinions about American affairs—something that has characterized global engagement with U.S. politics long before social media existed. The location of an account tells us nothing about whether the content is coordinated, conspiratorial, state-driven, or simply the opinion of an individual posting from another country.

More importantly, the emphasis on foreign accounts distracts from what we know empirically about the production and amplification of antisemitic content in the United States. Recent empirical analysis reveals the scale of domestic antisemitic expression: following the October 7 attacks, antisemitic discourse surged to 36-38% of comments on major UK news outlet YouTube channels—nearly double the pre-crisis baseline. After the Washington museum shooting in May 2025, antisemitic content averaged 43% across major English-language news channels, with some outlets reaching 66%. These are not fringe platforms infiltrated by foreign bots—these are mainstream digital spaces where domestic audiences actively produce and amplify antisemitic narratives.

Research on how antisemitic discourse actually spreads reveals a three-phase domestic process: elite figures articulate strategically ambiguous statements, digital intermediaries (podcasters, YouTubers, influencers) reframe and sharpen this messaging, and comment sections collapse ambiguity into explicit hate speech. This “cascading radicalization” is primarily a homegrown phenomenon, driven by domestic actors and platform dynamics—not solely by foreign infiltration.

A methodological note: Our analysis cannot definitively establish the geographic origin of every anonymous commenter. What we can observe is linguistic and cultural markers—idiom, references, political framings—that indicate domestic participation, combined with the absence of coordination patterns typical of bot networks or influence operations. The antisemitic discourse we documented emerges through what we call “dialogical warfare”: organic, seemingly spontaneous exchanges between users who present as ordinary Americans and deploy antisemitism as a convenient explanatory framework for complex political and social issues. This is not to claim foreign actors are absent—only that the dominant pattern we observe is grassroots participation rather than orchestrated infiltration.

The vast majority of high-engagement antisemitic narratives circulating on major platforms display linguistic and cultural markers of domestic origin and spread through participatory dynamics rather than coordinated bot activity—suggesting they are either generated or significantly amplified through U.S.-based networks. Domestic influencers, partisan media ecosystems, conspiracy-driven communities, and the broader attention economy play a decisive role in producing the narratives that drive real-world antisemitic sentiment and action.

When a U.S. congressmember amplifies antisemitic tropes, when a popular podcast platforms guests who traffic in conspiracy theories about Jewish power, when partisan media figures deploy dual loyalty accusations—these are not foreign operations. They are homegrown productions, embedded in American political discourse and amplified through domestic networks.

The Missing Architecture

What’s most striking about the celebration of X’s location feature is what it leaves unexamined: the platform itself.

There is virtually no discussion in the lawmakers’ responses about platform design, algorithmic amplification, recommendation systems, the broader attention economy, or the structural dynamics that allow hateful content to scale. The new feature is treated as inherently truth-revealing, a tool that exposes hidden foreign manipulation. But this framing evades the much more important questions: why do certain narratives spread, how do platform architectures enable their amplification, and how do the conditions of online communication—anonymity, mutual reinforcement among users, constant exposure to extremity—create environments where antisemitic ideas mutate and take hold?

The answer has little to do with account location and everything to do with how platforms are built and how online communication functions. Engagement-based algorithms reward content that provokes strong emotional reactions—outrage, fear, tribal solidarity. Recommendation systems create filter bubbles and radicalization pathways. The architecture of virality privileges simplification, moral clarity, and the identification of villains. The attention economy—structured around the commodification of user engagement—systematically rewards content that polarizes and enrages. These are design choices, not inevitable features of digital communication.

Meanwhile, the conditions of online communication itself—the anonymity that removes social accountability, the mutual confirmation that occurs when users encounter only like-minded voices, the omnipresence of hate speech that normalizes extremity—create a participatory environment where ordinary users become active contributors to radicalization.

Antisemitic narratives thrive in this environment not primarily because foreign adversaries plant them, but because:

· Platform algorithms amplify extremity

· Participatory comment cultures normalize hostility

· Domestic influencers accelerate radicalization

· Ambiguous political rhetoric creates interpretive openings

· Narratives mutate across digital formats

· The line between critique and conspiracy is steadily eroded

· Anonymity removes social constraints on expression

· Mutual reinforcement among users validates extreme positions

None of these dynamics can be addressed by knowing where an account is located.

Moreover, contemporary antisemitism increasingly operates through coded expressions, memes, and multimodal signals that evade simple keyword detection. The watermelon emoji, the paraglider symbol—these function as in-group markers regardless of the account’s geographic location. Strategic ambiguity, not foreign origin, is the mechanism that allows antisemitic narratives to spread while maintaining plausible deniability.

Focusing on the geographic location of accounts allows us to avoid confronting this reality. It suggests that the problem is infiltration rather than infrastructure, bad actors rather than bad design and bad communication conditions. It enables lawmakers to demand transparency about account origins while remaining silent about the recommendation algorithms that determine what billions of users see and the participatory dynamics that turn ordinary users into amplifiers of hate.

Even if location data were perfectly accurate—and it is not—it would tell us nothing meaningful about how antisemitic narratives arise, how they mutate across communities, how ambiguity invites radical reinterpretation, how multimodal forms evade detection, how algorithms intensify harmful speech, how participatory dynamics normalize extremity, or why millions of Americans find these narratives compelling.

The Political Convenience

The “foreign adversaries spreading antisemitism” narrative aligns neatly with a long-standing bipartisan preference: attributing social breakdown to hostile external actors. This framing is politically convenient across the spectrum.

For Republicans, it allows condemnation of antisemitism without confronting how right-wing media figures and influencers have mainstreamed antisemitic conspiracy theories, from “great replacement” narratives to accusations about George Soros to rhetoric about “globalists.” For Democrats, it enables criticism of online hate without reckoning with how segments of progressive activism have normalized anti-Zionist rhetoric that often slides into antisemitic tropes about Jewish power and loyalty.

What this framing obscures is how strategic ambiguity functions in domestic political discourse. Elite political actors craft statements that permit multiple interpretations—appealing to extremist audiences while maintaining plausible deniability. This ambiguity then gets amplified and radicalized as it circulates through domestic digital networks, aided by algorithmic systems that reward engagement and participatory cultures that normalize extremity. Focusing on foreign account locations allows us to ignore how domestic political rhetoric itself seeds and legitimizes antisemitic narratives, how platform architectures amplify them, and how the conditions of online communication enable their spread.

The foreign influence frame permits what we might call symbolic accountability—the appearance of action without the friction of actual institutional change. Lawmakers can call for transparency about account locations, celebrate when platforms implement such features, and position themselves as defenders against external threats. What they don’t have to do is examine how American political rhetoric itself contributes to the normalization of antisemitism, push for the kinds of regulatory interventions that would actually alter platform incentives, or confront how the basic conditions of online communication create radicalization pathways.

This is not simply analytical sloppiness. It is a moral abdication.

What Accountability Would Actually Require

Genuine accountability for online antisemitism would require confronting all three drivers of the problem—not just one.

First, it would mean acknowledging that while malicious actors (both foreign and domestic) do exploit divisive issues, they operate within a much larger ecosystem. We must recognize the United States as an active site where antisemitic ideas are produced, circulated, and normalized through domestic political culture, media ecosystems, and participatory online dynamics—not merely as an innocent target of external manipulation.

Second, it would mean confronting how platform architecture shapes what spreads, and demanding transparency not just about account locations but about algorithmic recommendation systems, content moderation decisions, the attention economy’s structural incentives, and the metrics that drive platform design. Like virologists who must understand viral mechanisms at the molecular level before developing vaccines, we must decode how hate narratives mutate, spread, and exploit platform vulnerabilities.

Third, it would mean recognizing that high-profile domestic actors—politicians, media figures, influencers with millions of followers—bear far more responsibility for mainstreaming antisemitic narratives than anonymous accounts, foreign or domestic. It would mean understanding the three-phase process through which elite ambiguity cascades into radicalized discourse through digital intermediaries and participatory comment cultures.

Fourth, it would mean examining how partisan political discourse itself traffics in antisemitic tropes through strategic ambiguity, and being willing to call this out even when it’s politically inconvenient.

Fifth, it would mean acknowledging the role of online communication conditions themselves—anonymity, mutual reinforcement, constant exposure to extremity—in creating environments where ordinary users become active participants in radicalization.

X’s location feature may provide some useful information about one factor among many. Knowing that an account engaging in U.S. political discourse is based abroad can be relevant context. But treating this as a revelatory tool that exposes the “true source” of antisemitic narratives is a dangerous misdiagnosis. It mistakes geography for causation, one contributing factor for the whole picture, and symbolic transparency for substantive accountability.

The architecture of online antisemitism is not primarily foreign. It is embedded in platform design, domestic political cultures, the attention economy, the conditions of online communication itself, and the structural incentives that reward certain kinds of content regardless of where it originates. The resurgence of antisemitism is inseparable from the broader architecture of the digital age: collapsing contextual boundaries, algorithmic acceleration, weakened epistemic authorities, participatory radicalization, and the mutual reinforcement enabled by anonymous online spaces.

If democracies want to confront antisemitism seriously, they must address all three drivers: yes, monitor and counter malicious actors where they exist, but also examine the platform architectures that amplify hate, and confront the participatory dynamics and communication conditions that enable antisemitic narratives to flourish in mainstream spaces. The way we diagnose the problem shapes how we conceptualize solutions for it. A comprehensive understanding without distortion isn’t just academically desirable—it’s operationally essential. Focusing exclusively on foreign actors may be politically convenient. It will not protect the public sphere.

AddressHate Research Scholar at NYU’s Center for the Study of Antisemitism | Lead, Decoding Antisemitism | Founding Research Advisor, AddressHate

Abstract

Social media platforms have become the central marketplace of ideas in contemporary democracies—yet they remain a black box. Without systematic tools to deconstruct the dynamics unfolding within these digital spaces or track their evolution in real-time, we lack the fundamental intelligence necessary to understand the societal and political trends shaping our societies. This article argues for the urgent integration of social media studies—combining discourse analysis with actor-focused research—into every attempt to comprehend polarization, radicalization, and the erosion of democratic norms.

Adopting a virological approach, this article treats hate ideologies—antisemitism, racism, misogyny, conspiracy theories, and anti-democratic resentment—as adaptive pathogens that mutate, spread, and exploit vulnerabilities within the digitally restructured public sphere. Just as virologists must decode viral mechanisms at the molecular level before developing vaccines, researchers must systematically map how hate discourse manifests, transforms, and circulates across platforms through strategic ambiguity, algorithmic amplification, and participatory radicalization. This requires a transdisciplinary synthesis: computational social science to detect patterns at scale, qualitative discourse analysis to interpret implicit meanings and cultural codes, and AI-enhanced methodologies to track real-time evolution.

The stakes are existential for democratic societies. Social media platforms have fundamentally restructured political communication, creating favorable “political-cultural opportunity structures” for extremist narratives while simultaneously dismantling traditional gatekeepers and arbiters of truth. New intermediaries—influencers, algorithms, and billionaire platform owners—now shape public discourse without institutional accountability, amplifying transgressive content that generates engagement regardless of the social harm it causes. The result is accelerating polarization, normalization of previously marginal ideologies, and erosion of the shared epistemic foundations necessary for democratic deliberation.

Yet policymakers have largely neglected this quasi-unregulated “town square”—as Cal Newport aptly describes it—allowing it to influence societal consciousness in uncontrolled and increasingly dangerous ways. Without rigorous social media studies that can penetrate the black box of digital discourse—revealing who says what to whom, how messages transform as they circulate, which actors and algorithms amplify which narratives, and what offline consequences follow—we operate blind. This article presents a roadmap from research to action: building advanced mixed-method infrastructures, understanding phenomena and trends through systematic monitoring, translating knowledge into policy and education, and developing context-sensitive interventions. Only through sustained investment in social media studies as a central pillar of democratic governance can we hope to counter the viral spread of hatred and preserve the integrity of the public sphere.

Introduction

In recent years, we have witnessed an alarming resurgence of hate-based ideologies across Western democracies—antisemitism, racism, misogyny, conspiracy thinking, and anti-democratic resentment—accompanied by a broader revival of authoritarian movements and systematic disinformation campaigns. These phenomena represent not merely temporary spikes in prejudice but rather the reemergence and transformation of age-old hatreds, now amplified by digital communication technologies. What makes this particularly concerning is the accelerating societal polarization and radicalization these dynamics produce, fundamentally threatening the cohesion of democratic societies.

This transformation reflects what scholars have termed a fundamental shift in “political-cultural opportunity structures”—the conditions under which hate speech, conspiracy theories, and anti-democratic ideologies can spread and gain legitimacy.¹ The digitally restructured public sphere, coupled with what has been described as an “epistemic crisis” stemming from eroded trust in traditional gatekeepers and truth-arbiters such as mainstream media and democratic institutions, has created highly favorable conditions for the dissemination of extremist narratives across the ideological spectrum.² These dynamics are further exacerbated by what some characterize as a “global wave of autocratization,” wherein democratic norms and institutions face systematic erosion worldwide.³

What makes this revival particularly concerning is the widespread moral disorientation it reveals: segments of civil society that traditionally champion progressive values have, often unwittingly, aligned themselves with movements that embody the very authoritarianism, sexism, racism, and homophobia they claim to oppose. Until recently, many believed that higher education and critical thinking would serve as bulwarks against extremist narratives. Yet the activities observed at universities across North America and Europe since October 2023 challenge this assumption. These demonstrations, praised by authoritarian regimes hostile to liberal democracy, raise fundamental questions about how educated individuals come to embrace ideologies fundamentally at odds with democratic principles.

The phenomenon emerges from multiple reinforcing factors: genuine lack of information, collective anxiety prompting people to embrace seemingly comprehensive solutions despite their better judgment, and relativist thinking creating confusion about fundamental values. However, these factors alone cannot fully account for the current erosion of moral and intellectual clarity. Social media platforms have fostered an environment defined by materialism, hyper-individualism, and the relentless pursuit of self-validation, bringing together all the essential elements for a perfect storm characterized by extreme feelings of aggression, fear, and isolation.

The recent surge in antisemitism, racism, misogyny, and conspiracy-driven disinformation stems from interconnected factors. These phenomena can be partially attributed to malicious actors who deliberately manipulate public discourse—numerous reports have documented the orchestration of public opinion by state and non-state actors seeking to destabilize democratic societies.⁴

Furthermore, the nature of digital communication involves both bottom-up and top-down dynamics that amplify hate and disinformation. New gatekeepers—influencers, billionaire media moguls, and platform algorithms—operate without the legal and institutional oversight standards that circumscribe established media. Algorithms privilege attention-generating, transgressive, polarizing, and emotional content, amplifying voices that promote disinformation and hatred in what has been termed a “post-factual age.”⁵

Yet the most significant factor remains the inherent nature of online communication itself—characterized by anonymity, mutual reinforcement among users, and the rapid dissemination of extremist ideologies. What has emerged is a complex and far-reaching toxicity unparalleled in human history. Operating under its own set of rules, the interactive web is shaping the consciousness of society at large. Both bottom-up and top-down dynamics work together to expand and legitimize digital hate, fostering what has been characterized as a communicative authoritarian “politics of transgression.”⁶ This politics of transgression operates across all forms of hatred and anti-democratic discourse, gradually shifting the boundaries of socially acceptable speech and normalizing extremist framings as legitimate opinion.

Largely neglected by policymakers, this quasi-unregulated digital public sphere is influencing societal thought in an uncontrolled manner.⁷ Anything short of a serious assessment of this concatenation of events virtually guarantees harmful consequences for democratic stability.

Against this backdrop, linguistic and computational analyses become indispensable tools. As with any social contagion, combating hate-based ideologies requires knowledge about their contemporary expressions and transmission mechanisms. While the dynamics described in this article apply across multiple forms of hatred and disinformation, we focus primarily on antisemitism as a paradigmatic case for understanding broader patterns of digital hate and radicalization.

Antisemitism provides a particularly instructive lens because it combines historical persistence, discursive adaptability, and cross-ideological circulation in ways few other hate ideologies do. Its discursive versatility and its function as a connective ideology link otherwise distinct forms of hatred and illiberal resentment. Moreover, the resurgence of antisemitism since October 2023 has been characterized as part of a “global antisemitic wave” unfolding both online and offline, connected to broader crises in democratic governance and the rise of authoritarian movements worldwide.⁸ This wave exemplifies how hate ideologies operate within favorable political-cultural opportunity structures created by the digitally restructured public sphere.

Research on antisemitism has identified two partly competing and partly complementary dynamics through which hate speech expands in digital environments. First, social media creates favorable conditions for transgressive dynamics, gradually shifting the boundaries of socially acceptable speech—allowing unfiltered hate, conspiracy myths, and extremist narratives to gain space and legitimacy. Second, hate ideologies broaden their reach by transforming over time and adapting to social codes of acceptability, circulating in camouflaged forms that maintain plausible deniability.⁹ Both dynamics—transgression and transformation—operate simultaneously across different forms of hate speech, from explicit racism to coded antisemitism to conspiracy-driven anti-democratic rhetoric.

As Rob Williams from the University of Southern California (USC) Shoah Foundation emphasizes regarding antisemitism, “we have to name it.” Collecting knowledge about its various forms constitutes “the first steps we must take in the fight against antisemitism.”¹⁰ Yet the methodological and analytical approaches developed for understanding antisemitism offer broader applicability: the same techniques for decoding implicit bias, tracking algorithmic amplification, and identifying radicalization pathways prove essential for confronting racism, misogyny, conspiracy theories, and anti-democratic disinformation campaigns. If we seek to understand and combat the most prevalent manifestations of hatred, we must focus on what has become by far the most important venue for political opinion formation and debate culture: social media.

We still know far too little about the current state of hate discourse in the online world. This applies to more conventional platforms like YouTube and Facebook as well as to Instagram, TikTok, Reddit, and various fringe platforms like 4chan. These digital spaces play a pivotal role in shaping the political perspectives of young individuals, yet they remain largely unexplored, functioning essentially as a black box.

Recent empirical work has begun to document the scale and complexity of this challenge. Analysis of YouTube comment sections following the October 7, 2023 attacks revealed that antisemitic discourse surged to 36-38% of comments on major UK news outlets—nearly double the pre-crisis baseline of 15-25%. Similarly, reactions to violent attacks on Jewish targets throughout 2024 and 2025 have shown sustained elevation in antisemitic content across major English-language news channels, with some outlets reaching peaks of 66%.¹² These figures illustrate not merely isolated incidents but a systematic normalization of antisemitic expression in mainstream digital spaces, contributing to broader patterns of societal polarization and radicalization. Parallel patterns emerge in racist discourse following immigration-related events, misogynistic content during feminist advocacy campaigns, and conspiracy theories during public health crises—each following similar amplification and normalization dynamics.

The multitude of expressions—denial, trivialization, justification, or celebration of violence against targeted groups, as well as stereotypes and conspiracy narratives across diverse platforms and online environments—are not adequately understood. Such understanding constitutes a fundamental requirement for any effective counterstrategy, whether initiated by educational or political actors.

The Spectrum of Hate Speech

Understanding hate speech in digital spaces requires recognizing its discursive versatility. Hate communication operates along a continuum from explicit incitement to implicit and coded expressions whose interpretation depends on contextual and cultural knowledge. This spectrum exists because hate speech thrives on what scholars have termed “communication latency”—the persistence of hateful meaning in coded or camouflaged forms that remain socially intelligible while retaining deniability.¹⁶ Such latent forms rely on irony, ellipsis, and intertextual cues rather than openly hostile statements, making meaning context-dependent and discursively mediated. Digital platforms magnify this dynamic through what can be described as a “politics of transgression”: implicit expressions test and expand the boundaries of acceptable speech through repetition and algorithmic reinforcement while maintaining plausible deniability.¹⁷

These questions become even more urgent when considering cases of physical violence. Historically, incidents such as those in Pittsburgh, Christchurch, and Halle have been linked to the online environments of perpetrators, where hate speech, conspiracy theories, and calls to violence were continuously propagated. More recently, violent attacks on Jewish institutions and individuals throughout 2024 and 2025 have demonstrated how online radicalization pathways precede offline violence. When individuals are consistently exposed to reality-distorting discourse, the likelihood of their beliefs being radicalized increases substantially. This underscores how language and discourse play pivotal roles in shaping attitudes, which ultimately can manifest in violent actions. Similar patterns emerge in racist mass shootings (Buffalo 2022, Jacksonville 2023), anti-LGBTQ+ violence, and misogynistic attacks—each preceded by immersion in relevant hate discourse communities.

The Multimodal Challenge

Contemporary hate speech does not confine itself to text alone. It increasingly manifests through multimodal assemblages where images, memes, emojis, and videos carry as much semantic weight as words. Following the October 7 attacks, emojis such as paragliders (🪂), Palestinian flags (🇵🇸), and watermelons (🍉) functioned as proxies for solidarity with Hamas or approval of violence. The watermelon emoji performs a particularly complex semiotic function: it recontextualizes the colors of the Palestinian flag while simultaneously implying that expressions of solidarity require coded forms due to supposed censorship—an insinuation that aligns with conspiratorial notions of Jewish control over public discourse.¹⁸

Similarly, racist discourse deploys the 👌 hand gesture (reappropriated by white supremacists), monkey emojis in reference to Black individuals, and Pepe the Frog memes. Misogynistic communities use specific emoji combinations and visual memes to signal affiliation with incel or manosphere ideologies.

These multimodal signals evade keyword detection systems while remaining readily legible to in-group audiences. Memes repurpose popular formats to insert hateful analogies: antisemitic memes portray Jews or Israel as deceptive, greedy, or manipulative; racist memes deploy dehumanizing imagery; misogynistic memes spread narratives about female manipulation or inferiority. Profile images featuring swastikas, Hitler portraits, or other extremist symbols signal ideological alignment without requiring textual content. This complexity introduces higher-order challenges for computational systems: models must align visual, textual, and symbolic cues while integrating cultural literacy to interpret meaning accurately.

Comprehending these manifestations and the interconnections between stimuli, discourse events (both offline incidents and online occurrences with the potential to inflame hatred), and the subsequent waves of distortion, conspiracy narratives, and disinformation can enable us to reconstruct user motivations and, by extension, understand offline behaviors. This illuminates what appeals to particular audiences and what their underlying needs and concerns entail. This kind of social media research represents the most effective—and likely the only—means of generating actionable insights into societal sentiments.

A Virological Approach: Roadmap from Research to Action

Understanding hate ideologies in the digital age requires us to adopt a fundamentally new perspective. Like virologists studying pathogens, social media and hate researchers must examine how antisemitism, racism, misogyny, conspiracy thinking, and anti-democratic resentment exist, morph, and spread through contemporary communication networks. Just as virologists develop vaccines by understanding viral mechanisms at the molecular level, we must decode the structure, transmission patterns, and mutation dynamics of hate discourse to create effective remedies.

Recent research has revealed a critical transmission mechanism: strategic ambiguity functions as a catalyst for what can be termed “cascading radicalization” across digital platforms.¹⁹ This process follows a three-phase model whereby ambiguous elite discourse is reframed by digital intermediaries and then collapses into explicit hate speech in audience participation—a sequence summarized as “ambiguity → reframing → collapse.”²⁰ Understanding these transmission dynamics is essential for combating all forms of hate speech, from antisemitism and racism to misogyny and conspiracy-driven disinformation.

This virological approach demands sophisticated analytical tools that can only emerge from the systematic integration of humanities, social sciences, and data science. The challenge before us is not merely interdisciplinary—it requires a transdisciplinary synthesis where computer science, quantitative social research, and qualitative interpretive methods work in concert throughout every phase of investigation. While this article draws primarily on antisemitism research to illustrate methodological principles, the framework applies equally to understanding and combating other forms of hate speech and disinformation.

Step 1: Advanced Research Infrastructure—Mixed Methods and LLM-Enhanced Methodologies

The foundation of effective intervention begins with robust research capabilities that transcend traditional disciplinary boundaries. Contemporary computational social science has typically combined data science with quantitative methods, but this framework overlooks the indispensable role that qualitative, interpretive research plays throughout the research process—an oversight that produces significant blind spots and biases.²¹

A comprehensive approach integrates three complementary domains: computer science (technical infrastructure for large-scale data collection, algorithmic processing, and machine learning), quantitative social science (theory-driven frameworks, statistical modeling, and rigorous hypothesis testing), and qualitative social science (exploratory approaches for understanding emerging phenomena, developing nuanced theoretical frameworks, and providing contextual understanding necessary for meaningful interpretation). This integration produces a computational social science capable of both statistical analysis and interpretive depth—one that can explore new phenomena while testing hypotheses at scale, interpret complex cultural meanings while processing millions of data points, and establish evaluation criteria that reflect genuine understanding rather than mere pattern matching.

Recent developments demonstrate the necessity of systematic integration across five research stages:²¹

(1) Concept definition and operationalization translate theoretical constructs—such as the IHRA working definition of antisemitism, the UN’s racism framework, or feminist theories of misogyny—into linguistically testable indicators that account for co-text, context, and multimodal expression.²²

(2) Data collection and sampling balances breadth of coverage with depth of contextual understanding. The Decoding Antisemitism project has annotated over 300,000 comments,²³ yet this remains modest compared to billions circulating online. Similar challenges face researchers studying racist discourse, misogynistic communities, and conspiracy theory networks—each requiring sustained data collection across multiple platforms and temporal periods.

(3) Annotation and model training represent the critical juncture where human interpretation guides machine learning. Expert annotation integrates three knowledge domains: language knowledge (linguistic structures, pragmatic meanings), co-text knowledge (immediate discourse context), and world knowledge (historical understanding of hate tropes and cultural codes).²⁴ Critically, annotation applies a conservative attribution principle: when statements permit multiple interpretations—one prejudiced and one non-prejudiced—the statement is coded as non-prejudiced unless substantial evidence establishes the prejudiced interpretation as more probable, protecting against over-interpretation while acknowledging potential underestimation of actual hate prevalence.²⁵

(4) Computational analysis leverages recent advances in large language models through prompt engineering and chain-of-thought reasoning. Emerging retrieval-augmented generation (RAG) architectures couple LLMs with external knowledge bases—annotated corpora, lexicons, stereotype taxonomies—reducing hallucination by integrating relevant evidence. Looking ahead, context-engineered architectures could provide models with conversational co-text, historical event context, named-entity resolution, and situational metadata, approximating human expert reasoning.²⁶

(5) Model evaluation and validation ensure automated systems produce meaningful insights rather than reproducing biases. Preliminary experiments with recent LLMs (Gemini-2.5-Flash, Llama-3.3-70B, MoonshotAI Kimi-K2) achieved F1-scores above 0.88 on antisemitism detection tasks.²⁷ Yet, significant limitations remain: fine-tuned BERT models miss approximately one-third of cases,²⁸ antisemitism remains statistically rare (1-2% of datasets), and data scarcity is exacerbated by platform restrictions and resource-intensive expert annotation.²⁹

Step 2: Understanding Phenomena and Trends

Armed with these sophisticated mixed-method tools, researchers can move beyond static description to dynamic understanding. This phase focuses on identifying how hate narratives—antisemitic, racist, misogynistic, conspiratorial, and anti-democratic—emerge, evolve, and proliferate across digital networks. Key objectives include mapping the mutation of hate tropes as they adapt to contemporary discourse, tracking transmission pathways across platforms and communities, identifying superspreader accounts and amplification mechanisms, detecting early warning signals of escalating radicalization and societal polarization, and understanding the interaction between offline events and online discourse intensification.

Recent research has identified a systematic three-phase pattern in how hate speech and disinformation spread through digital ecosystems:³⁰

(1) Primary discourse involves elite figures articulating statements characterized by strategic ambiguity—rhetoric permitting multiple interpretations that appeal to divergent audiences while maintaining plausible deniability.³¹

(2) Secondary discourse occurs when digital intermediaries (YouTubers, TikTokers, podcasters, influencers) reframe elite messaging, typically reducing ambiguity by sharpening language and positioning it within explicit ideological frameworks. Unlike traditional journalists, these creators operate within attention economies that reward provocation and emotional intensity, often shaping downstream interpretation more powerfully than original statements.

(3) Tertiary discourse unfolds in comment sections where audiences actively co-create meaning through participatory engagement. Here, ambiguity collapses: implicit meanings become explicit, political critique blurs into conspiracy theory, and extreme positions gain legitimacy through social proof rather than authoritative endorsement.

This three-phase circulation enables what researchers term “cascading radicalization”—ambiguous elite rhetoric becoming progressively more explicit and extreme through semantic crystallization (formulations becoming endorsements), emotional intensification (measured rhetoric becoming charged), and normalization of extremity (shocking statements becoming everyday discourse).³²

Hate ideologies function as living, adaptive systems that respond to interventions, exploit platform vulnerabilities, and continuously evolve new expressions to evade detection through orthographic manipulation, compound neologisms, and rhetorical indirection.³³ The aftermath of October 7, 2023, exemplified this dynamic through open celebration of violence, while subsequent attacks throughout 2024 relied more heavily on conspiracy narratives, denial, and coded dismissal—illustrating how different triggering events activate different patterns within the broader spectrum of hate communication.³⁴ Parallel evolution occurs in racist discourse (from explicit slurs to “race realism” and “human biodiversity”), misogynistic content (from overt hostility to “men’s rights” framing), and conspiracy theories (from specific allegations to meta-narratives about “the elites”).

Understanding these dynamics in real time provides the intelligence necessary for effective response across all categories of hate speech and disinformation.

Step 3: Knowledge Transfer into Policy, Law, and Education

Research insights remain academic exercises unless translated into actionable frameworks for societal actors. This critical phase involves multiple dimensions of knowledge transfer:

Policy Development requires providing policymakers with evidence-based recommendations for platform regulation, content moderation standards, and legislative frameworks that balance free expression with protection from incitement. The European Union’s Digital Services Act represents a first attempt to mandate greater transparency in content moderation and algorithmic processes, yet effective implementation requires collaboration with academic experts and access to high-quality datasets.³⁵

Legal Application means equipping law enforcement and judicial systems with the knowledge to identify, investigate, and prosecute hate crimes while understanding the online radicalization pathways that precede offline violence. Legal frameworks must account for cross-jurisdictional differences: what constitutes protected speech in the United States may qualify as group defamation or hate speech under EU or German jurisprudence.

Educational Intervention involves developing curricula and pedagogical approaches informed by actual understanding of how young people encounter and internalize hate narratives online. This includes creating media literacy programs that specifically address the rhetorical strategies and emotional appeals that make extremist content compelling—teaching students to recognize communication latency, coded language, and multimodal signaling. Such educational efforts are essential for countering radicalization and reducing societal polarization among younger generations.

Platform Accountability entails engaging technology companies with specific, evidence-based insights about how their algorithmic systems amplify harmful content and providing concrete recommendations for architectural changes that would reduce this amplification without requiring impossible levels of content moderation. Current moderation approaches often focus on slurs or direct threats while ignoring implicit or multimodal forms of hate speech, and enforcement remains inconsistent across languages and regions.³⁶

Platform dynamics introduce additional complexity through what has been termed “context collapse”—the flattening of diverse audiences into a single public sphere.³⁷ Statements crafted for one interpretive community may be decoded very differently by others when they circulate across platforms. Social media users engage in “impression management” across collapsed contexts, potentially increasing reliance on ambiguous formulations that permit multiple readings. Algorithms further shape circulation patterns: content that generates engagement (including controversy) receives amplification regardless of whether that engagement reflects accurate interpretation. Understanding these dynamics is essential for developing effective content moderation and platform governance strategies that address not merely individual harmful posts but the systemic amplification mechanisms that enable hate speech to spread.

A persistent challenge lies in the tension between over-blocking and under-blocking. When content is removed too aggressively, legitimate political critique or neutral references risk being censored. Research has shown that widely used moderation services systematically inflate toxicity scores for comments containing identity-related terms regardless of stance.³⁸ Conversely, when hateful content remains online, it risks normalization and causes emotional harm to targeted communities. Automated moderation systems must therefore achieve context-sensitive classification that distinguishes between legitimate political critique and hate speech employing political framing.

Step 4: Building Prerequisites for Effective Prevention and Intervention

The ultimate goal is not merely to understand hate speech but to prevent its spread and intervene effectively when it emerges. This requires several interrelated capabilities:

Early Warning Systems deploy the computational tools developed in Step 1 to monitor online environments continuously, detecting concerning trends before they escalate into offline harm. Whatever unfolds during the next pandemic, global economic crisis, or climate-related resource shortage often simmers beforehand in the depths of the anonymous web. With appropriate analytical tools, we can detect early warning signs.

Rapid Response Capabilities create institutional structures that can quickly mobilize counter-narratives, fact-checking, and strategic communications when hate disinformation campaigns emerge. This requires understanding not only what narratives circulate but also which superspreader accounts and amplification mechanisms drive their dissemination.

Community Resilience uses research insights to strengthen the psychological and social resources of communities targeted by hate, helping them recognize manipulation tactics and maintain cohesion in the face of attacks. Targeted communities expect protection from online hatred, yet poorly designed interventions risk re-traumatization or further marginalization.

Rehabilitation Pathways develops intervention strategies for individuals showing signs of radicalization, informed by an understanding of what draws people into extremist ideologies and what might draw them back out. This requires mapping not merely explicit expressions but the gradual radicalization process through which distorted ideas solidify into fixed belief systems. Effective de-radicalization programs must address the societal polarization that creates fertile ground for extremist recruitment.

Conclusion

This roadmap represents a comprehensive response to hate ideologies that matches the sophistication of the threat. Just as modern medicine relies on molecular biology to develop targeted therapies, modern democracy requires computational social science to develop targeted interventions against hate in all its forms—antisemitism, racism, misogyny, conspiracy thinking, and anti-democratic disinformation. These viruses of hatred have found new vectors of transmission in social media; we must become the virologists who decode their mechanisms and develop the remedies our societies desperately need.

While social media has emerged as one of the most significant destabilizers of our time, it also presents a tremendous opportunity. Over the past century, our societies have made varying degrees of effort to eradicate the deeply ingrained hatred that has permeated cultures for millennia. This enduring challenge prompts many to doubt whether these ideologies of hate can ever be completely eradicated.

Traditionally, surveys were our primary means of gauging the prevalence of hatred in society. However, they pale in comparison to the insights gleaned from a thorough analysis of social media discourses. Social media interactions capture voluntary expressions, dialogues among thousands of individuals, and the entire spectrum of human interaction. Understanding online discourse equates to understanding the formation of attitudes within our society—a level of insight we have never before possessed in human history.

The thorough identification of predominant communication patterns online extends beyond documenting the present; it can illuminate the future. Understanding the emergence of trends within our society and deciphering how hatred manifests, through which channels, and employing what messaging strategies can reveal where future harm is likely to occur.

To grasp the forthcoming direction of our society—encompassing both online trends and their offline ramifications—rigorous interdisciplinary research into antisemitism, racism, misogyny, conspiracy thinking, and other hate-based ideologies threatening our free societies, democratic political systems, and social cohesion is imperative. This research must utilize all available methodological approaches. Social scientists and data science experts must engage in ongoing dialogue and mutual learning. While political interventions or platform regulations alone cannot solve these threats to democracy, neither can artificial intelligence address them adequately in isolation. Only through collaborative efforts can we discern, in real time, the evolution of online debates, the resonance and dissemination of specific extremist narratives and disinformation campaigns, and the individuals or groups who wield influence in shaping the interpretation of critical issues.

This text serves not only as an assessment of hate-related social media studies but also as a plea. As a society, we must acknowledge that the traditional methods of upholding social peace, embodied in our political, legal, educational, and security institutions, are inadequate for navigating this new digital domain. We must dedicate our efforts not merely to exploring these dynamics with a few spotlights but rather to comprehensively understanding the phenomenon to identify both the symptoms, their underlying causes, and the potential consequences. This will enable us to initiate a truly well-informed dialogue with the drifting segments of society. For this, we need robust research and fruitful collaborations between academia, the political arena, and civil society—developing methodologies that work across all forms of hate speech, from antisemitism and racism to misogyny and anti-democratic disinformation.

Only by decoding the discourses that shape our digital age can we begin to rebuild the moral grammar of democratic society.

Footnotes

1. Lars Rensmann, „Politischer Antisemitismus im postfaktischen Zeitalter: Formen und Ursachen in Demokratien des 21. Jahrhunderts“ (Baden-Baden: Nomos, 2025); Heiko Beyer et al., „Antisemitismus in der Gesamtgesellschaft von Nordrhein-Westfalen im Jahr 2024,“ Antisemitismusbeauftragte des Landes Nordrhein-Westfalen (2024).
   

2. Gabriele Cosentino, “Social Media and the Post-Truth World Order: The Global Dynamics of Disinformation” (Cham: Palgrave Macmillan, 2020).
   

3. Anna Lührmann and Staffan I. Lindberg, “A Third Wave of Autocratization is Here: What is New About it?” Democratization 26, no. 7 (2019): 1095–1113.
   

4. Institute for Strategic Dialogue, “Information Laundromat Detects Banned Russian Propaganda Across Hundreds of Websites,” accessed October 2024, https://www.isdglobal.org/isd-in-the-news/information-laundromat-detects-banned-russian-propaganda-across-hundreds-of-websites; Institute for Strategic Dialogue, “Russia-Ukraine War,” accessed October 2024, https://www.isdglobal.org/tag/russia-ukraine-war; Institute for Strategic Dialogue, “China, Russia and Iran Are Exploiting the Israel-Hamas Conflict for Their Advantage,” accessed October 2024, https://www.isdglobal.org/isd-in-the-news/china-russia-and-iran-are-exploiting-the-israel-hamas-conflict-for-their-advantage; see also Deen Freelon and Chris Wells, “Disinformation as Political Communication,” Political Communication 37, no. 2 (2020): 145-156, https://doi.org/10.1080/10584609.2020.1723755.
   

5. “Facebook Whistleblower Frances Haugen Testified That the Company’s Algorithms Are Dangerous—Here’s How They Can Manipulate You,” The Conversation, October 5, 2021, https://theconversation.com/facebook-whistleblower-frances-haugen-testified-that-the-companys-algorithms-are-dangerous-heres-how-they-can-manipulate-you-169420.
   

6. Ibid.
   

7. Lars Rensmann, “The Noisy Counter-Revolution: Understanding the Cultural Conditions and Dynamics of Populist Politics in Europe in the Digital Age,” Politics and Governance 5, no. 4 (2017): 123–135.
   

8. Cal Newport, “The Real Problem with Twitter,” accessed October 2024, https://calnewport.com/the-real-problem-with-twitter.
   

9. Matthias J. Becker and Lars Rensmann, “The Dynamics of Digital Antisemitism: Anti-Jewish Narratives in the Aftermath of October 7th,” in Antisemitism Online: An Ancient Hatred in the Modern World, ed. Todd Pittinsky (Oxford: Oxford University Press, in press).
   

10. Ibid.
    

11. USC Shoah Foundation, “Facing Antisemitism,” accessed October 2024, https://sfi.usc.edu/video/facing-antisemitism.
    

12. Decoding Antisemitism, https://decoding-antisemitism.eu.
    

13. Matthias J. Becker et al., eds., Decoding Antisemitism: A Guide to Identifying Antisemitism Online (Palgrave Macmillan/Springer Nature, 2024), https://link.springer.com/book/10.1007/978-3-031-49238-9.
    

14. Matthias J. Becker et al., Celebrating Terror: Antisemitism Online After the Hamas Attacks on Israel. Preliminary Results I (Berlin: Technische Universität Berlin, Center for Research on Antisemitism, 2023), https://doi.org/10.14279/depositonce-19143.
    

15. Matthias J. Becker, Jordan Blatter, and Oksana Stanevich, “Decoding Antisemitism Online: Linguistic and Multimodal Challenges in the Age of AI,” In Frontiers in Communication: Media Governance and the Public Sphere (under review).
    

16. Werner Bergmann and Rainer Erb, “Kommunikationslatenz, Moral und öffentliche Meinung. Theoretische Überlegungen zum Antisemitismus in der Bundesrepublik Deutschland,” Kölner Zeitschrift für Soziologie und Sozialpsychologie 38 (1986): 223–246.
    

17. Becker and Rensmann, “The Dynamics of Digital Antisemitism.”

18. Becker, Blatter, and Stanevich, “Decoding Antisemitism Online.”
    

19. Parker Bach, Carolyn E. Schmitt, and Shannon C. McGregor, “Let Me Be Perfectly Unclear: Strategic Ambiguity in Political Communication,” Communication Theory 35, no. 2 (2025): 96–106; Matthias J. Becker, Gabrielle Beacken, and Liora Sabra, “How Political Ambiguity Transforms Across Digital Platforms: Mechanisms of Discourse Radicalization” (working paper, 2025).
    

20. Ibid.

21. Katharina Soemer, Daniela Gurschow, and Steffen Egert, “Social Science and AI Joining Forces: Towards New Approaches for Computational Social Science” (paper presented at the interdisciplinary conference on computational methods, 2025).
    

22. International Holocaust Remembrance Alliance, “Working Definition of Antisemitism” (2016), https://holocaustremembrance.com/resources/working-definition-antisemitism; Becker et al., Decoding Antisemitism.
    

23. Becker, Blatter, and Stanevich, “Decoding Antisemitism Online.”
    

24. Ibid.
    

25. Matthias J. Becker and Hagen Troschke, “Decoding Implicit Hate Speech: The Example of Antisemitism,” in Challenges and Perspectives of Hate Speech Research, ed. C. Strippel et al., Digital Communication Research 12 (2023): 335–352, https://doi.org/10.48541/dcr.v12.20.
    

26. Becker, Blatter, and Stanevich, “Decoding Antisemitism Online.”
    

27. Jordan Blatter and Blue Square Alliance Against Hate, “Testing Next-Generation AI to Better Detect Antisemitism,” Blue Square Alliance Command Center Insights (2025), https://www.bluesquarealliance.org/command-center-insights/testing-next-generation-ai-to-better-detect-antisemitism.
    

28. Milena Pustet and Helena Mihaljević, “Automated Detection of Antisemitic Texts: Is Context All We Need?” In Decoding Antisemitism: An AI-driven Study on Hate Speech and Imagery Online. Discourse Report 6, ed. Matthias J. Becker et al. (Berlin: Technische Universität Berlin, Center for Research on Antisemitism, 2024), https://decoding-antisemitism.eu/publications/sixth-discourse-report.
    

29. Becker, Blatter, and Stanevich, “Decoding Antisemitism Online”; Elisabeth Steffen, Milena Pustet, and Helena Mihaljević, “Algorithms Against Antisemitism? Towards the Automated Detection of Antisemitic Content Online,” in Antisemitism in Online Communication: Transdisciplinary Approaches to Hate Speech in the Twenty-First Century, ed. Matthias J. Becker et al. (London: Open Book Publishers, 2024), https://doi.org/10.11647/OBP.0406.08.
    

30. Becker, Beacken, and Sabra, “How Political Ambiguity Transforms Across Digital Platforms.”
    

31. Bach, Schmitt, and McGregor, “Let Me Be Perfectly Unclear.”
    

32. Becker, Blatter, and Stanevich, “Decoding Antisemitism Online.”
    

33. Ibid.
    

34. Becker and Rensmann, “The Dynamics of Digital Antisemitism.”
    

35. Jash Patel, Heer Mehta, and Jeremy Blackburn, “Evaluating Large Language Models for Detecting Antisemitism,” arXiv preprint arXiv:2509.18293 (2025), accepted to EMNLP 2025 Main Conference.
    

36. Ibid.
    

37. danah boyd, “Social Network Sites as Networked Publics: Affordances, Dynamics, and Implications,” in A Networked Self: Identity, Community, and Culture on Social Network Sites, ed. Zizi Papacharissi (New York: Routledge, 2010), 39–58.
    

38. Alexis Chapelan et al., Decoding Antisemitism: An AI-driven Study on Hate Speech and Imagery Online. Discourse Report 5 (Berlin: Technische Universität Berlin, Center for Research on Antisemitism, 2023), https://doi.org/10.14279/depositonce-17105; Lucas Dixon et al., “Measuring and Mitigating Unintended Bias in Text Classification,” in Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society (AIES) (2018), 67–73.

AddressHate Research Scholar at NYU’s Center for the Study of Antisemitism | Lead, Decoding Antisemitism | Founding Research Advisor, AddressHate

© Dylan Costa

Acknowledgement

This analysis was made possible in part through the generous data support of the Bright Initiative.

Introduction

The assassination of Charlie Kirk at Utah Valley University on September 10, 2025, was a political shock and cultural rupture alike. In the hours that followed, reactions across digital and physical arenas revealed how deeply polarized, conspiratorial, accusatory, and corrosively ironic public discourse around violence has become. Celebrations of Kirk’s death appeared not only in ad-hoc surveys, street demonstrations, vigils, and university spaces but also in the comment sections of mainstream media platforms — evidence of how hatred, mockery, and conspiracy narratives now circulate seamlessly between embodied protest and digital publics.

These responses illustrate a broader shift: political debate or reflection on the causes of violence is increasingly displaced by the demonization and delegitimization of opponents, a discursive move that normalizes destructive attitudes. Notably, the projections of blame sparked by Kirk’s murder extended beyond partisan divides, fusing with anti-elitist and antisemitic worldviews and underscoring how entrenched discursive patterns structure the digital afterlife of political violence.

This study sets out to look directly into the face of hate as it emerges in the immediate aftermath of a hate crime in our time. By capturing digital reactions in unfiltered form, it seeks to document how online publics respond to violence not with mourning or restraint but with Schadenfreude, projection, and conspiracy claims — exposing the rhetorical repertoires that fuel the digital afterlife of political violence.

Although political polarization and online hate have been widely studied, far less attention has been paid to how acts of political violence are discursively processed in real time on platforms such as YouTube. The immediacy and visibility of these reactions make comment sections uniquely revealing sites for examining how violence is rationalized, celebrated, or denied, and how existing ideological repertoires are mobilized in the wake of shocking events.

This article does not seek to assess whether Kirk’s rhetoric amounted to hateful or exclusionary speech. Instead, it examines the discursive consequences of his assassination: how the event was framed, contested, and appropriated across digital publics. At its center is a case study of YouTube comment sections on twelve widely circulated clips from leading U.S. media outlets, spanning the political spectrum from left to right. By analyzing 2,400 comments collected within the first 24 hours after the attack, the study captures the immediate responses before narratives were reframed by subsequent news cycles. Combining quantitative coding with qualitative multimodal analysis, it highlights how discursive schemata — from Schadenfreude and empathy refusal to antisemitic tropes and conspiracy narratives — shaped the digital afterlife of Kirk’s murder.

Kirk as a Polarizing Figure

Kirk, the 32-year-old co-founder of Turning Point USA, had become one of the most visible figures of the U.S. conservative movement. He built a powerful youth organization, cultivated links to the Republican establishment, and drew equal measures of admiration and condemnation for his combative style. To his supporters, he embodied the defense of free speech, Christian conservatism, and an unapologetic defense of Israel. To his critics, he symbolized the mainstreaming of conspiratorial thinking and culture-war rhetoric that fueled U.S. polarization.

Any account of Kirk’s legacy must also confront the ambiguity of his positioning on antisemitism. Critics pointed to his references to the Great Replacement theory and his claims that Jewish donors disproportionately financed “anti-white” or “cultural Marxist” causes (FactCheck, September 12, 2025). At the same time, Kirk often issued some of the sharpest denunciations of antisemitism in contemporary politics. In July 2023, he posted on X: “Jew hate has no place in public discourse, period, end of story.” At other events, he rebuked questioners who attempted to bait him with conspiratorial attacks on Judaism by calling antisemitism “demonic” and “in the pit of Hell” (Washington Free Beacon, September 12, 2025).

After his death, these ambiguities gave rise to public controversy. The New York Times initially attributed an antisemitic statement to Kirk himself before issuing a correction clarifying that he had been quoting and then critiquing it (NYT, September 11, 2025; Washington Free Beacon, September 12, 2025). Supporters accused mainstream outlets of smearing him, while detractors pointed to his alignment with Elon Musk in claiming that “some of the largest financiers of left-wing anti-white causes have been Jewish Americans” (Media Matters, November 16, 2023). Others highlighted how his rhetoric drew on older antisemitic tropes of Jewish power and influence, including claims that Jewish donors were behind “cultural Marxism” and anti-whiteness in Hollywood and academia (TRT, September 11, 2025). Kirk’s memory thus became a contested site: for some, a dog-whistler drawing on classic antisemitic tropes of Jewish influence; for others, one of the most forceful conservative voices rejecting antisemitism.

Kirk’s polarizing reputation extended beyond antisemitism. He disparaged diversity, equity, and inclusion (DEI) programs as “anti-White,” expressed skepticism about professional diversification — remarking of a Black pilot, “boy, I hope he’s qualified” (Washington Post, September 13, 2025) — and urged women to “submit to your husband” while rejecting feminism (Atlantic, September 16, 2025). He also opposed men working as preschool teachers and resisted women in combat roles (Media Matters, August 18, 2023). As a whole, these positions aligned his rhetoric with conservative gender norms and positions critics viewed as racist. Supporters, by contrast, framed them as opposition to race-based policies, further intensifying his polarizing image.

Shooter’s Trajectory in Hyper-Online Subcultures

The primary suspect, Tyler Robinson (22), followed a markedly different trajectory from the figure he targeted. Once described as a high-performing student, Robinson graduated from Pine View High School in St. George in 2021 and briefly attended Utah State University on a scholarship before dropping out after his first semester. For several years afterward, he appeared disengaged from politics. Family members and acquaintances later recalled that his political interest grew only in 2024–25, when he expressed strong disdain for Kirk, calling him “full of hate and spreading hate” at a family dinner shortly before the Utah event (CNN, September 13, 2025).

Robinson’s background complicated easy explanations. He came from a middle-class family in Washington, Utah, where his father — a 27-year veteran of the county sheriff’s department — ultimately turned him in after the shooting. His mother worked in disability support services, and family photographs depicted him with two younger brothers on vacations and at family celebrations. The household lived in a six-bedroom home valued at around $600,000, far removed from stereotypical profiles of marginalization or deprivation.

Material evidence pointed less to a coherent ideology than to Robinson’s immersion in digital subcultures. Investigators recovered bullet casings with engraved inscriptions ranging from antifascist slogans (“Hey Fascist! Catch!”, “Bella Ciao”) to ironic provocations (“If you read this, you are gay lmao”) and meme-based humor from gaming and furry role-play (“Notices bulges / OwO what’s this?”). One casing bore the sequence “↑ → ↓↓↓” — the exact Helldivers 2 code for triggering the Eagle 500kg Bomb — next to the taunt “Hey Fascist! Catch!” Released in 2024, Helldivers 2 is a satirical co-op shooter in which players fight for “Super Earth,” a parody of fascist militarism explicitly modeled on Starship Troopers (see Wired; Cosmic Book; Polygon; KSL-TV, September 12, 2025). The game itself quickly became a flashpoint as fans moved to distance it from real-world violence, with moderators on Discord and Reddit locking threads overwhelmed by posts condemning the association. Subsequent reporting noted that Utah Gov. Spencer Cox told investigators about alleged Discord messages, including what appeared to be a confession sent hours before Robinson’s arrest; Discord, however, denied that the platform was used to plan or promote the attack, and some of the circulating messages may have been retrospective rather than contemporaneous (Washington Post; CBS; Guardian, September 15–16, 2025). As extremism researcher Alex Newhouse cautioned, it is less the face-value meaning of any inscription than the very act of engraving meme references on bullets that reflects the irony-posting, edgelord aesthetics of deep internet culture.

In this interpretive vacuum, observers stressed that the markings did not amount to a legible ideology but exemplified what Charlie Warzel called “shitposting in real life” (Atlantic, September 12, 2025): a hyper-online style of provocation designed to confuse audiences, bait media attention, and inscribe oneself into the genealogy of infamous shooters. From this perspective, Robinson’s act can be read not only as a political attack but also as a performance within a transnational digital fandom of violence.

Other aspects fueled speculation. Commentators reported that Robinson’s romantic partner is transgender, though authorities have not confirmed whether this detail is central to his motive (Guardian, September 19, 2025). Meanwhile, conspiracy-leaning voices on the right speculated about Robinson’s supposed ties to the Groypers — a claim lacking credible evidence and rejected by both Groypers’ leaders and researchers — but it reflects how rapidly extremist ecosystems move to place new incidents into their existing narratives (Atlantic, September 18, 2025).

These details suggest a profile of a young man politicized in the digital age, who developed deeply negative views of figures like Kirk and sought to set an exemplum. This phenomenon is not rare today, as the norms of democratic debate and the capacity to endure conflict are displaced by:

• a lack of empathy, emotional intelligence, and conceptual flexibility,

• aggressive self-positioning and the demonization of the other, and

• desires for elimination and destruction.

Such attitudes are cultivated across the political spectrum and amplified by the dynamics of online culture. Just as Robinson’s online aesthetic spilled over into real-world violence, so too did public reactions spill outward — from anonymous comment threads into street protests, vigils, and university spaces — where Schadenfreude and mockery of Kirk’s death became highly visible. It is to these public, embodied responses that the analysis now turns.

Street Celebrations and Student Reactions

The aftermath made clear that the celebration of Kirk’s death was not confined to anonymous online spaces. Activists in several U.S. cities held street gatherings where participants mocked his killing. Videos circulated on TikTok and LinkedIn showed young Bob Vylan chanting “We got Charlie in the neck.” At a Seattle vigil, one participant shouted, “I would have killed him myself”; at the University of Texas, another declared, “Someone had to do it.” In Idaho, a man disrupted a memorial shouting “Fuck Charlie Kirk!” (Free Press, September 12, 2025).

At Oxford University, George Abaraonye, the incoming president of the Oxford Union, privately messaged friends: “Charlie Kirk got shot, let’s fucking go.” He later walked back his remarks as “impulsive,” insisting that “nobody deserves to be the victim of political violence.” Still, the case underscored how celebratory rhetoric appeared even within elite institutions.

The Free Press (September 12, 2025) noted how such responses — calling Kirk a “Nazi” or a “Zionist,” dismissing his death as “meaningless” — proliferated in the hours after the shooting. These public displays illustrate how Schadenfreude moved from digital comment threads into embodied and highly visible forms of protest.

Digital Afterlife

Against this backdrop, Kirk’s assassination quickly became a discursive flashpoint. If street gatherings illustrated the embodied dimension of Schadenfreude, the digital aftermath unfolded across multiple platforms, with YouTube providing the most immediate and large-scale record of public reactions. Within hours of the first news reports, tens of thousands of comments accumulated across videos from mainstream U.S. outlets. As the Foundation to Combat Antisemitism (FCAS) observed in its bulletin, Kirk’s killing generated more than 26 million posts globally in just two days (FCAS Bulletin, September 12, 2025).

Beyond YouTube, celebratory rhetoric and conspiracy theories converged on platforms like X, TikTok, and Bluesky. Far-right accounts circulated a doctored Mossad logo with the apocryphal motto “By way of deception thou shalt do war,” insinuating Israeli responsibility. Influencers such as Jake Shields, Jackson Hinkle, and others claimed Kirk had been “asking the uncomfortable questions about October 7,” suggesting his killing was retribution for criticizing Israel. Commentary accounts resurfaced earlier warnings that Kirk might “turn on Israel” and be silenced, while conspiracy broadcaster Harrison H. Smith alleged that Kirk himself feared assassination if he abandoned pro-Israel positions. Adding to this conspiratorial framing, Candace Owens claimed—without evidence—that Kirk had been threatened at a Hamptons gathering by figures such as Bill Ackman, and suggested his assassination was the result of a broader pro-Israel plot (NY Post, September 16, 2025). Tucker Carlson amplified aspects of this narrative, speculating that Kirk had grown disillusioned with Israeli leadership and warning that officials were “hijacking” his death for political ends (MSN, September 18, 2025). These projections exemplify how antisemitic tropes—of Jewish power, Mossad orchestration, or Jewish financiers pulling strings—were woven seamlessly into the celebratory framing of his death, even when based on unverified or demonstrably false allegations.

The YouTube corpus analyzed for this study mirrors these dynamics. Much of the speculation unfolded in the hours before the primary suspect, Tyler Robinson, was identified, when the absence of facts created a vacuum quickly filled by competing projections and conspiracy claims. Users debated who was responsible, expressed satisfaction or even joy, and circulated conspiracy theories. What stands out is not only the polarization of blame but also a striking refusal of empathy — a readiness to see violence against a political opponent as deserved or even welcome. While some comments expressed grief or condemned political violence, a substantial share amplified speculation, mockery, and conspiracy narratives. Many insisted that Kirk had “brought it on himself,” accused mainstream media of manipulation, or framed the assassination as the inevitable outcome of partisan conflict. Schadenfreude was pervasive, often couched in the language of irony or partisan loyalty.

These comment spaces quickly became discursive battlegrounds, where overt hate, Schadenfreude, and dehumanizing jokes interwove with conspiratorial claims. The “blame game” was not random noise but followed recognizable schemata:

• Deterministic belief systems — the conviction that discursive conventions, political slogans, or legal frameworks must inevitably lead to violent outcomes.

• Blame-game schemata — rapid simplifications of reality that recycled conspiracy narratives and relentlessly framed the killing through the binary lens of the U.S. two-party system, with Democrats and Republicans alternately depicted as culprits or beneficiaries.

The pattern closely parallels my earlier analysis of the Washington Embassy shooting in May 2025 (Becker, Kvetching Intensifies, May 23, 2025). There, too, digital publics refused empathy, mocked victims, and inverted responsibility — often weaponizing antisemitic tropes to reframe Jewish victimhood as manipulation. In sum, the Washington and Utah cases illustrate how the digital afterlife of political violence operates: not as open-ended mourning or reflection, but as an immediate re-inscription of violence into entrenched repertoires of polarization, conspiracy, and antisemitic projection.

Dataset and Method

This study draws on twelve strongly circulated YouTube clips from leading U.S. media outlets, covering a broad spectrum from left to right. The corpus includes:

• Left / Progressive: MSNBC, CNN, The Daily Show, The Young Turks (TYT)

• Center / Broad Audience: ABC News, NBC News, CBS News, Associated Press (AP), USA Today

• Populist / Heterodox: Breaking Points (Krystal Ball & Saagar Enjeti)

• Right / Conservative: Fox News, Forbes Breaking News

From each video, the first 200 comments appearing chronologically were collected within the first 24 hours after the event (September 10–11, 2025), resulting in a total corpus of 2,400 comments. This timeframe captures the most immediate reactions, before narratives had settled or been reframed by subsequent news cycles.

The dataset was analyzed using a mixed-methods approach that combined quantitative coding (to determine the frequency of rhetorical strategies) with qualitative multimodal content analysis. Coding and annotation were carried out in MAXQDA, a software package designed for qualitative and mixed-methods research. To ensure consistency, an intercoder reliability test was conducted among team members, with disagreements discussed and collectively resolved. The main categories of analysis included:

• Projections of guilt (left, right, media, Kirk himself)

• Affirmation and glorification of violence

• Non-antisemitic conspiracy narratives

• Antisemitic tropes and conspiracy narratives

• Threats and Calls for Retaliation

These categories were derived from the Decoding Antisemitism Lexicon (Becker et al., 2024), which systematizes rhetorical strategies and discursive patterns in online antisemitic and adjacent hate speech. This framework also ensures comparability with earlier case studies, such as reactions to the murder of two Israeli diplomats in Washington in May 2025, which revealed strikingly similar repertoires even in ostensibly moderate online milieus. The distribution of repertoires across outlets—for instance, stronger Kirk-blame on progressive platforms, denser antisemitic conspiracies on NBC/ABC/TYT, and predominance of anti-Left blame on Fox/Forbes—is analyzed in detail in the Results and Synthesis sections below.

3. Results

a) Quantitative Overview

• Comment volume (200 per video, 2,400 total, distribution across outlets).

• Category frequencies (Left’s fault, Right’s fault, Media, Kirk himself, Schadenfreude, Lack of empathy, Glorification of violence, Conspiracy [AS vs non-AS], other antisemitic concepts).

© Oksana Stanevich

2. Qualitative Findings: Discursive Repertoires

Note upfront: A striking feature of the YouTube discourse is the way commenters collapsed broad political categories into singular, guilty collectives. Just as the conservative camp was frequently equated with the “radical right” or with MAGA extremists, so too were liberals treated as interchangeable with “the Left,” the Democratic Party, or progressive activists. Nuances between institutional actors, grassroots movements, and media figures were flattened, producing a polarized blame game in which entire political segments were cast as responsible for Kirk’s death. This reduction of complex political landscapes into binary camps underscores how online publics process violence through entrenched repertoires of projection, rather than through differentiated analysis of actors or causes.

2.1. Discursive Blame Projections

Public reactions to Kirk’s murder were saturated with blame projections. Commenters sought immediate culprits, rarely engaging in reflection or mourning. Four distinct targets dominated these projections: the political Left, the political Right, mainstream media, and Kirk himself.

a) The Left’s Fault

On MSNBC and CNN, many commenters projected blame almost reflexively onto Democrats and the broader Left. They framed liberals as violent, hypocritical, and intolerant, insisting that this was simply “terrorism from the left AGAIN” or that “Democrats shot an American hero.” Others turned the “tolerance” claim against liberals themselves, arguing that the “peaceful and tolerant left” was in fact dangerous and warning that “things won’t be safe until they are eliminated.”

This projection often intensified through identity politics and conspiratorial cues. On MSNBC, users claimed that “the shooter is a trans Antifa member” or that “Woke and LGBT+ culture laugh and celebrate this, and then ask for understanding and play the victim.” On AP threads, the rhetoric escalated to eliminationism: “the left needs to be stricken from existence.” Forbes threads added a systemic angle, asserting that “It is the MSM and Democrat Party that have all the dangerous rhetoric” or that “Dems are unhinged. Claim to be on the side of love and acceptance until you disagree with one of their views, then they try to take you out.”

CNN comments followed a simpler formula but reinforced the same narrative: “All the violence, every bit of it, is coming from the left … conservatives push for safe gun use.” Across outlets, the underlying logic was consistent: Democrats and liberals were not only responsible for Kirk’s death but represented the principal source of political violence in contemporary America.

Key Examples

• “It is the MSM and Democrat Party that have all the dangerous rhetoric.” (Forbes)

• “Dems are unhinged. Claim to be on the side of love and acceptance until you disagree with one of their views, then they try to take you out.” (Forbes)

• “Terrorism from the left AGAIN.” (MSNBC)

• “Democrats shot an American hero.” (MSNBC)

• “The left needs to be stricken from existence.” (AP)

These comments portray the Left as hypocritical—claiming to defend tolerance but, in practice, enabling or celebrating violence. In Fox and Forbes threads, this trope dominated, while MSNBC and AP threads escalated the projection into eliminationist or conspiratorial rhetoric.

b) The Right’s Fault

On MSNBC and CNN, many commenters inverted the blame frame and argued that responsibility lay with conservatives themselves. Some insisted that the Right had normalized political violence, claiming that “nearly every single violent crime in the past decade has been from a RIGHT WINGER!!” or that “MAGA is eating its own. A MAGA tried to unalive Trump and another one got Chucky.” Others directed the blame squarely at Trump, writing that “we only have violence when Trump is in office. Blood is on MAGA’s hands.”

This projection often fused accusations against Kirk with broader indictments of the conservative movement. On MSNBC threads, one commenter declared bluntly: “No Trump and Kirk caused this with their hateful rhetoric.” Others personalized responsibility to Trump and his allies: “Trump is president, that’s karma,” or “JD Vance, Donald Trump and Mike Johnson? Yeah they sure need to be held accountable.”

On CNN and Daily Show threads, commenters emphasized pattern and precedent. One detailed catalogue listed: “The man who targeted and killed Rep Melissa Hortman … was a Trump supporter … The man who tried to kidnap Nancy Pelosi and assaulted her husband was a Trump supporter … There is no equivalent list of Democrats committing similar attacks.” Another mocked conservative attempts to externalize blame: “Trump would blame the left if he got a hangnail.”

Elsewhere, the projection focused more directly on Kirk himself as part of the right-wing problem. On ABC threads, one user claimed: “Charlie Kirk was shot as a direct result of Trump’s intentionally hateful rhetoric to ramp up division 😡👀.” CBS threads echoed this framing, with one remarking that “It may be sickening but it is a totally predictable consequence of the Trumpidian coup.” On TYT, a similar causal logic was made explicit: “Call it what it is: if Donald Trump was not president this would not have happened.”

By contrast, on Fox and Forbes threads, such blame-attribution was rare and often countered immediately. Still, even there, isolated voices turned the accusation back onto the right, e.g., “Floyd didn’t make a living with racist and hate speeches … Live by the sword, guess what?”

Key Examples

• “Nearly every single violent crime in the past decade has been from a RIGHT WINGER!!” (MSNBC)

• “We only have violence when Trump is in office. Blood is on MAGA’s hands.” (MSNBC)

• “No Trump and Kirk caused this with their hateful rhetoric.” (MSNBC/Forbes)

• “The man who targeted and killed Rep Melissa Hortman … was a Trump supporter … There is no equivalent list of Democrats committing similar attacks.” (CNN/Daily Show)

• “Charlie Kirk was shot as a direct result of Trump’s intentionally hateful rhetoric to ramp up division 😡👀.” (ABC)

• “It may be sickening but it is a totally predictable consequence of the Trumpidian coup.” (CBS)

• “Call it what it is: if Donald Trump was not president this would not have happened.” (TYT)

These comments frame the Right not only as complicit in Kirk’s death but as the structural source of political violence in the U.S. They reflect a mirror-image projection to that seen on Fox or Forbes, where the Left was accused of orchestrating violence.

c) The Media’s Fault

Another recurrent blame frame targeted mainstream media. Commenters accused outlets such as MSNBC, CNN, and Fox of fomenting division, spreading hate, or even inciting violence directly. On MSNBC threads, some declared that “MSNBC and all its HATE FILLED staff have blood on its hands! They encourage a hate filled ideology giving evil and/or unstable people a reason to do terrible atrocities.” This trope of “blood on the media’s hands” condensed widespread resentment into a moralized accusation: the press was not reporting on violence but actively producing it.

On CNN threads, the suspicion was redirected toward conservative networks. Users retorted sarcastically: “I wonder where the shooter gets his news? Hmm. Fox trash news.” Such reversals illustrate how partisan publics refracted the same media-blame schema through their own ideological lens: liberals saw Fox as the corrupting agent, while conservatives saw MSNBC or CNN.

On Forbes and MSNBC threads, media-blame was fused with partisan accusations against Democrats, crystallizing in formulas such as “The Democrats and the mainstream media are evil and contributed to this Charlie Kirk murder” (Forbes) or “CNN and MSNBC must be held accountable for this horrific day 😡” (MSNBC). On NBC and ABC, media-blame comments took a sharper conspiratorial edge. One commenter insisted, “Hey NBC News your hate finally paid off.” Another generalized: “All MSM is complicit in his X’ing!” Such formulations positioned media not only as biased but as participants in a coordinated elimination of political opponents.

Although less frequent on Fox and Forbes, isolated cases of anti-media blame still appeared, often folded into anti-left accusations. The overall picture is one of symmetrical projection: left-leaning publics accused Fox, right-leaning publics accused MSNBC or CNN, and centrist outlets attracted blanket condemnation of “mainstream media” as a hostile force.

Key Examples

• “MSNBC and all its HATE FILLED staff have blood on its hands!” (MSNBC)

• “The Democrats and the mainstream media are evil and contributed to this Charlie Kirk murder.” (Forbes)

• “CNN and MSNBC must be held accountable for this horrific day 😡.” (MSNBC)

• “I wonder where the shooter gets his news? Hmm. Fox trash news.” (CNN)

• “All MSM is complicit in his X’ing!” (ABC)

These examples show how distrust of the media cuts across the spectrum, with outlets selectively blamed depending on political orientation. In each case, the press was constructed as an active agent of violence rather than a neutral observer.

d) Kirk’s Own Fault

A final projection of blame targeted Charlie Kirk himself. Unlike accusations against political camps or media outlets, this frame personalized responsibility: Kirk’s rhetoric, worldview, and persona were said to have invited the violence. Within this discourse, four interwoven strands emerged: attributions of fault, refusals of empathy, claims that he deserved it (often framed through depictions of evil), and expressions of Schadenfreude.

a) Attributions of Fault

Many commenters argued that Kirk’s own words had normalized violence and therefore rebounded upon him. On MSNBC and CNN threads, users stressed causal reciprocity: “Your words do have consequences” and “If he wasn’t such a hater someone would not have shot him.” On CNN, others recycled his past statements, reminding readers that “Charlie Kirk himself called for violence and defended school shootings.” This “live by the sword, die by the sword” logic framed his death as the predictable outcome of his politics.

b) Refusals of Empathy

Other reactions stripped Kirk of sympathy altogether. Rather than framing his death as deserved, commenters simply refused to mourn. Irony and cold detachment dominated: “Thoughts and prayers. That’s what we do right? Am I doing it right? … WOOO… NAILED IT!” Another wrote: “O well … world keeps rolling … in Palestine they kill 100k people lets talk about that.” By denying Kirk empathy, commenters inverted his own statements — especially his dismissal of “empathy” as a damaging idea — and turned them back against him.

c) Attributions of Deservedness and Evil

A sharper variant cast his death not just as the result of his politics, but as morally justified. Some framed his death as karmic justice: “Karma” (MSNBC), “Racism is very expensive. It’s cost a lot of people their lives” (CBS), or “Blood is on your hands, son. Reap what you sow” (MSNBC). Others were more blunt: “He got what he deserved” (ABC), “this is on him. Full stop” (AP), “Flying too close to the sun has consequences” (ABC), or “Zero prayers. The world is already a better place” (CBS).