We scored 21 camp-versus-camp narratives across 14 fault lines for the window of September 14 to 20, 2026, which closes at about 19:00 UTC on September 20, with a median score of 3.6. Four sit at the top watch tier: the El-Sayed association frame, the reckless-war frame and the AI doomer-op frame at 4.4, and the China-psyop backlash frame at 4.1. Scores are published to one decimal. The four are listed by score for reference, and a gap of a tenth or two is inside the noise of five judged factors. The vice president’s recorded instruction to House Republicans to run against one Michigan Senate nominee everywhere, and the president’s posts calling AI fear a hoax and the data-center backlash a conspiracy, are the triggers behind three of the four. Eighteen verdicts came back mixed, two came back contradicted by our checks, and one reads core unsubstantiated by our checks; no frame was fully supported. A card reads mixed only where a check on its narrow reading found independent support, contradicted where a probative check on the claim’s core proposition failed, and core unsubstantiated where no probative check reaches that core proposition either way. Twenty-one checks on twelve cards are kept as context and carry no weight in a verdict. Two narratives read as ignited. One reached ignited for the first time on its chain this week, the Kraft billionaires-loyalty frame, ignited in its first window. One continues: the El-Sayed association frame, in its fourth ignited week. Six sit at near ignition: the Woke Right cabal frame, the China-psyop backlash frame, the British two-tier frame, the Clancy sympathy-as-endorsement frame, the Sánchez frame and the Kirk death-cult frame. Rulings dated September 20 shape those counts. A cleared item ignites a card only when it names the card’s trigger or the card’s target in its own words, and under that test the Clancy and Sánchez cards, ignited a week earlier, read near ignition on their markers, and the Kirk death-cult card holds at near ignition for a fourth week. The demonization-defense card, which carried the same claim as the El-Sayed association card on the same trigger with the same six carriers, is folded into it, so the week has 21 cards. No trajectory reading was applied to any card this week. Three measured readings met the standing test for an applied multiplier, the five-post floor in both windows on a card past its first appearance, and are published beside their scores as unapplied diagnostics, because our collectors read nothing between 06:33 UTC on September 15 and 22:55 UTC on September 19 and the window was read back from account timelines, so the two windows were collected differently. Every reach figure here is a floor. Nothing measured here cleared our harm bar, with one exception this week: items inside two narratives cleared the harm bar, were routed to the harm instrument, and are reported only there. What remains on these cards did not clear our harm gate.
Everything on this dashboard is ordinary political expression under this instrument, with one exception this week: items inside two narratives cleared the harm bar, were routed to the harm instrument, and are reported only there, and what remains on these cards did not clear our harm gate. A narrative appearing here describes a framing pattern and accuses no one of hate, extremism, or bad conduct. Our separate harm methodology, with its own much higher bar, is reported in The Weather Report.
Collection note for this week. Our collectors read nothing between 06:33 UTC on September 15, the last successful read before the outage, and 22:55 UTC on September 19, the first after it. X was backfilled after that hour by reading the timelines of all 152 tracked accounts, which returns original posts only and stops at 200 posts per account; Bluesky was backfilled at 100 posts per account; 4chan holds /pol/ threads for September 14, 15, 19 and 20, and nothing created on /pol/ between about 06:29 UTC on September 15 and 09:11 UTC on September 19, about four days; the 45 threads dated September 16 to 18 are all from /g/; news for September 16 to 18 is a partial backfill; and collection ran live again from 22:55 UTC on September 19 until the record closed at about 19:00 UTC on September 20, so the window is September 14 to 20 with its last day partial. Every reach figure on this page is a floor, a low count or an unanswered mirror may partly reflect the gap, and no trajectory reading is applied, because the measured readings compare this corpus with one collected mostly live a week earlier. The Limitations block under Method sets out the detail.
Our weekly harm report measures the narrow band of content that collectivizes, dehumanizes, or incites against groups of people. Most of the culture war never gets near that bar. This report measures that larger remainder: same week, same verification standard, different instrument.
We track framing contests. An event happens, and within hours commentators on each side stop arguing about the event and start arguing about what it reveals about the other camp's motives and character. We state each narrative's claim in a testable form, check how the predictions it generates fare against what is observable, look for the mirror narrative on the other side, and score the whole contest for reach, intensity, penetration, elite adoption, and mirror lock. We also watch each narrative for seven specific markers that it is drifting toward the territory our harm methodology covers.
Four narratives reached the top tier this week among the cards we scored, listed here by score for reference. Each is given with last week’s score beside it, and with the two weeks collected differently, the difference between the two scores carries no directional reading. Every score on this page is published to one decimal. Three sit at 4.4. The El-Sayed association frame was scored at 4.4 this week and at 4.4 last week, and reads as ignited for a fourth week, now holding the demonization-defense card folded into it under a ruling dated the evening of September 20: Vice President Vance told a closed House Republican conference, in a recording CNN obtained, that a member facing a Democrat who is ‘not that crazy’ should ‘just talk about Abdul El-Sayed’, and within three days the NRSC, AIPAC, the RJC and Mark Levin were carrying Hasan Piker’s answer to Axios, that America is worse than al-Qaeda ‘pound for pound as far as the number of deaths’, as proof of what the Michigan Senate nominee and the Democrats funding him are. The reckless-war frame was scored at 4.4 this week against 4.1 last week; its record this week holds Rep. Massie’s eight articles of impeachment against Defense Secretary Hegseth, Speaker Johnson’s canceled final session day, a Congressional Budget Office estimate of about $38 billion in direct war costs, and a Washington Post report that six officials put US military deaths above the Pentagon’s 18, with five officials each carrying a part of it: Sen. Warren wrote that the administration is covering up the number of troops killed, Sen. Sanders attacked the president’s remark that $4.44 a gallon is ‘a very inexpensive price’, Rep. Jeffries wrote of a ‘reckless war of choice’ that is ‘a failure’, Rep. Khanna wrote that Speaker Johnson ‘doesn’t want a vote on stopping the war’, and Rep. Massie filed the impeachment articles. The cover-up charge is Warren’s alone. The AI doomer-op frame was scored at 4.4 this week against 4.0 last week; the president wrote that AI fear ‘is a HOAX’ and PCAST co-chair David Sacks called the labs’ inspector proposal ‘Trust & Safety 2.0’; on September 19 the president wrote on Truth Social that the concerns belong to a long line of hoaxes ‘generated by the Radical Left’ and announced an ‘AI Force’, and the card records both. The China-psyop backlash frame was scored at 4.1 this week against 3.5 last week, on the president’s post that a ‘SICK conspiracy’ against data centers pleases only China, with its warning to ‘Treasonists, Traitors, and Leakers’; his September 19 post that ‘the crazed Data Center attack has largely failed’ blames Democrats, leaves China out, and is recorded on that card. Four cards sit just under the tier line at 3.9: the serve-Israel frame, the British two-tier frame, the recess frame, and a new card on the Kraft venue story, in which Glenn Greenwald, Max Blumenthal, Cenk Uygur and Candace Owens read Robert Kraft’s ban on Macklemore as pro-Israel billionaires showing where their loyalties lie. The Woke Right cabal frame sits in the active tier at 3.7, against a published 2.1 a week earlier: Mark Levin, Laura Loomer and James Lindsay reading Tucker Carlson’s report on a planned transfer of 40,000 bombs to Israel, and his live show with Ana Kasparian of The Young Turks, as the work of an organized faction allied with the left against the country, in Lindsay’s words ‘two hands of one Woke beast’. Its five-factor score was 3.5 last week and is 3.7 this week, and the rest of the difference is a 0.6 multiplier applied a week earlier and none applied now; the trajectory paragraph below sets out the arithmetic. Twelve sit in the active tier and five at W3, with none at W4, and a median of 3.6 across all 21 against 3.7 across twenty-three a week earlier. Seventeen play out in the US arena, two in the UK, one in Spain and one in Canada. Officials or major outlets carried nine of the cards as their own, the top elite-adoption anchor: the vice president on the El-Sayed association card, the president on the two AI cards, two senators and the House minority leader on the war, three senators on the AI industry’s money, two House members of opposite parties on the canceled voting day, the Senate Republican Conference’s open-borders frame in its seventh window, a sitting member of the Commons on the Portsmouth mask charge, and Rep. Massie on the FDA. That count describes the narratives we scored, and the whole discourse is outside its reach. Six cards had no card a week earlier. Five open new lineages: the Kraft billionaires-loyalty frame and its mirror, the pro-Israel reading of the backlash against Ed Sheeran and Kraft as a Free Palestine hate mob; the Piddington frame, in which Steve Laws, Jayda Fransen and Paul Golding recast an Oxfordshire village’s symbolic secession vote over an asylum site as the hypocrisy of Liberal Democrat voters; the reading of Meryl Streep’s casting as Aslan as proof that Netflix and Hollywood hate Christian viewers; and Massie’s recasting of the FDA’s 2021 vaccine approval as Peter Marks running off the career experts so the mandates could issue. The sixth resumes a lineage last carded in W36: the reading of Johnson’s canceled voting day as flight from the Hegseth impeachment vote and the Epstein petition.
The predictions are our operationalizations of each frame. Eighteen of the 21 verdicts came back mixed this week, two came back contradicted by our checks, and one reads core unsubstantiated by our checks; no frame had all of its predictions resolve in its favor. A week earlier the split was eighteen mixed and five contradicted across twenty-three cards, on a scale that carried no third verdict. Nine verdicts are held at low confidence and twelve at moderate. Four dated rulings govern the verdict line, and a reader comparing this week with last should read every verdict change through them, because a change that follows from a rule is no new evidence about a frame. The first, dated the evening of September 19, supersedes the rule recorded earlier that day: a card reads mixed only when at least one check on its narrow reading came back true or mixed on independent evidence. A narrow reading that is untested or unresolved, or that rests only on the carrier’s own material, or on political adoption of the claim such as a resolution, a vote or an endorsement, adds no positive evidence, and a card whose widest reading fails with no independently supported narrow reading reads contradicted. The second, dated September 20, sets how much each check weighs. Twenty-one checks on twelve cards are marked as context only: each records a fact that belongs to the story and cannot tell the claim’s readings apart, so it carries no weight in the verdict and is labeled as context in the drill. Where no probative check bears on a wide reading, the card says that reading is unsubstantiated on what we found, which counts neither for the claim nor against it. The third, dated the evening of September 20, fixes what a narrow check is judged against: the narrow reading as the card states it. A narrow check that is true or mixed on independent evidence, and that could have come out the other way, is probative and makes the card mixed, even where it lends nothing to the wide reading. No scope exception applies to one card alone, and the same fact carries the same weight on both cards of a pair. The fourth, dated the night of September 20, opens a third verdict beside mixed and contradicted. Where no probative check reaches the claim’s core proposition in either direction, the card reads core unsubstantiated by our checks, and contradicted is held for a card whose core proposition a probative check tested and failed. One card reads core unsubstantiated this week, the Aslan casting frame, which read contradicted before the third verdict existed. One verdict differs from a week earlier on its chain, and that difference follows from these rules. The Clancy sympathy-as-endorsement frame reads mixed at low confidence, as it did in each of its three earlier weeks, so its verdict is no longer a change on its chain. The governor reading fails on the public record, a proposal that funds screening and home visits and does nothing for the defendant. The camp-line reading, women and the left siding with the killer, is unsubstantiated on what we found, the jury’s eleven-to-one split being context that describes a court panel and cannot show who the public sympathizers are. The check that holds the card at mixed is the one on the account of deliberations, probative and back at mixed on independent reporting: the holdout juror says he had no doubts and was cut off, at least five jurors describe him acknowledging doubt, and nothing outside the jury room settles the two accounts. The harassment reading rests on the word of the holdout juror’s lawyer, with no police confirmation and no source identified for the threats, so it carries no weight in either direction. The Senate Republican Conference’s open-borders card, in its seventh window, read contradicted a week earlier and reads mixed at low confidence. Its narrow check holds on independent records: Rep. Ro Khanna’s in-window ‘Abolish ICE’ post, Rep. Thanedar’s bill and an Arkansas nominee’s recorded remark put named Democrats on the record for abolishing or defunding ICE. Under the ruling of the evening of September 20 that check is true, probative and independent, which is mixed under the rule as written; the scope exception that had held this card alone at contradicted, that a finding about a wing lends nothing to a claim about a party, was applied to no other card and is withdrawn. The party-wide claim gains nothing from it: the conference cited no bill, plank or pledge for noncitizen voting or a halt to all deportations, and the step from a wing to the whole party stays unsupported. This change, too, follows from a rule. The Woke Right cabal frame reads mixed at low confidence, as it did in each of the two weeks before this one. Its narrow check documents Carlson’s live show with Ana Kasparian, Massie’s promotion of her from his congressional account, his joint work with Ro Khanna, and a staffing project its builders advertise, and judged against the card’s own narrow reading, that members openly cooperate with left figures and have built their own personnel institutions, it is true on independent records. The widest reading, a coordinated operation, fails on a public trigger that Newsweek reported the same day, with no funding link, directive or foreign principal offered, and a search that finds no link cannot exclude an undisclosed one. An organized faction working with Islamist interests against the country is unsubstantiated on what we found. Two cards read contradicted. On the China-psyop backlash card, at moderate confidence, the widest reading fails on two probative checks. The first is the independent audit record carried from a week earlier: X and OpenAI disclosed the state-linked account sets, so those accounts are independently documented, and audits by Clemson’s Darren Linvill and by Graphika found little or no engagement and no breakout from them. The second is the frame’s own witness, who told a House Foreign Affairs subcommittee that ‘Two things can be true at once’ and that Americans have ‘genuine concerns about the data center buildout’, a concession against his own side where the claim holds that opponents have no genuine grievance. The House vote of 417 to 3 for the Ratepayer Protection Act and the replies under the president’s own Truth Social posts show a grievance that is widely held and cannot show where the sentiment came from, so both are context. That opponents serve a hostile power is unsubstantiated on what we found. On the narrow reading, the state-linked accounts exist and reached almost no one, and the presence of Singham-funded groups in specific campaigns is alleged by the frame’s carrier alone: the campaign count comes from an institute whose head of research testified for the frame, and no independent body has replicated the count or verified the evidence under it, so that check is context. On the Sánchez card the verdict rests on one probative check, the one on the week’s operation inside Ceuta and the Red Cross. The widest reading has the state forwarding people to the continent with the Red Cross as its accomplice. The week’s one large movement of people ran from Ceuta’s beaches to a tented site inside Ceuta’s port under police escort: France 24 reported hundreds moved, Euronews reported more than 4,000 people arriving at a site with capacity for about 1,700 and more than 2,000 going back to the beaches when it filled, and neither reported a transfer to the peninsula. The counts of who remains in Ceuta are context. Both were published on September 18: Reuters reported that more than 72,000 people entered the enclave at the end of July and that Spain estimates 12,000 remain, and France 24 reported the Ceuta authorities’ estimate of 13,000. They do not show where the others went, and leaving the enclave is a different fact from leaving Spain or Europe. Deliberate flooding and criminal conduct are unsubstantiated on what we found, and the card reads contradicted at low confidence, because a probative check reached its core proposition and came back against it. The narrow reading, that regularisation works as a pull, gained a European Parliament resolution adopted 339 to 225; a resolution is evidence that the claim has entered mainstream politics and no evidence of a pull effect, so it leaves the verdict where it stood. One card reads core unsubstantiated by our checks, a new card, the Aslan casting frame. One check on its widest reading fails on an independent record, the released image, which a fan site with no stake in defending Netflix describes with no rainbow colouring, drag or sex change in it. The other two checks on that reading, the Lewis company’s producing role and the costly theatrical release, are context. The card’s core proposition is a motive, hatred of Christian viewers, and no probative check reaches it in either direction, which is what the third verdict records. The narrower reading of a knowing provocation rests on the director’s own account, which gives an acting reason, and the card reads core unsubstantiated by our checks at low confidence. The other two cards that came back contradicted a week earlier, the Mamdani ceremony-clip frame and the Netflix caption frame, took no card this week. A mixed verdict means at least one check on the narrow reading found independent support while other checks failed or stayed unresolved, so a mixed verdict on a compound claim substantiates none of its clauses on its own. Several mixed cards split along the same seam: a documented kernel and an imputed motive that no independent check reaches. Kraft’s own statement confirms the ban and the promoter reported six other stadium owners refusing the act, while the claim of a power that banishes critics across every major power center failed in a week when every remaining opening act quit the tour and eight outlets covered the removal within 72 hours. On the mirror card, the Free Palestine hate-mob frame, the fact that Kraft announced the ban explains why he was the person named and is the same under both readings, so it is context there; that card’s mixed verdict rests on the harm-side record of an abusive edge at the margin of the backlash. The war’s cost and the gap between internal and published casualty counts are independently documented, while a deliberate cover-up rests on denials and unnamed sources. Donor money reaching AI-safety journalism and a super PAC is on the record, and on the mirror card so is industry money aimed at regulation-minded candidates; each money check is probative on its card’s narrow reading alone and carries no weight for the wide one. The breadth of concern about AI, in two polls and in replies under the president’s own posts, is context on the doomer-op card as it is on the China-psyop card, so a Democratic product manufactured for the midterms is unsubstantiated on what we found, as is the reading that the labs’ leaders do not believe their own warning. On the Trump dividend card, Republican senators declining to endorse the checks, among them Sen. Barrasso giving no commitment on Meet the Press on September 20 and Sen. Cassidy calling the checks ‘absolutely inflationary’ on Face the Nation the same day, show fiscal objection inside the party; senators oppose sincere proposals on the same grounds, so the check is context, the card stays mixed at low confidence on the party’s documented standing, and an offer with no policy behind it is unsubstantiated on what we found. Confidence is low on two further mixed cards: the serve-Israel card, whose one narrow check rests on a poll from August 2025, fielded before the war, and the Kirk death-cult card, whose one narrow check is held at mixed with the memorial vandalism unattributed. Each card carries the prediction, what we observed, and the source, and each of the fifteen persisted narratives and the resumed one ran its predictions against this window’s record.
Four narratives record a locked mirror this week, two pairs, both inside the card set. One comes from the Michigan Senate fight: the El-Sayed association frame against the reading, carried by Piker, Uygur and Blumenthal, that the Axios headline and the disavowal questions are a ‘uniparty coordinated media effort’ against the left. One comes from the president’s AI posts: the doomer-op frame against the industry-money card, on which Sen. Van Hollen calls the administration’s refusal to act ‘part of an implicit deal’, Sen. Sanders faults AI CEOs for spending through their super PACs against candidates who want regulation, Sen. Warren calls the labs’ antitrust ask ‘transparently self-serving’, and Rep. Khanna writes that ‘The tech lords came after me’. Two pairs that locked a week earlier loosened. The frame that the right exploits 9/11 did not recur, its five volume terms matching no in-window post, and the demonization-defense card it locked with is folded into the El-Sayed association card; the frame that the right is the violent side did not recur either, and the Kirk death-cult card moved from locked to answered with its lock factor lowered. The Kraft pair is carded on both sides so that each carries a verdict, and it reads as answered and short of a lock: the two sides cite each other often, and the evidence on each side is mostly the event itself. Fifteen of last week’s narratives persisted under ruled lineage joins, five opened new lineages, and one resumed across a gap week; the ledger below counts the resumed card among its six new because it had no card a week earlier. Eight lineages took no card this week. Seven are the Netflix caption frame, the WNBA panic-as-pretext frame, the Mamdani ceremony-clip frame, the right-exploits-9/11 frame, the right-is-the-violent-side frame, the Bernier frame and Kennedy’s silenced-mothers frame, and for those the record says only that they did not clear the bar or were not surfaced, which it cannot separate, and in a week collected by backfill the second possibility weighs more than usual. The eighth is the demonization-defense lineage, carded since W32, and its absence has a known cause: its W38 card was folded into the El-Sayed association card under a ruling dated the evening of September 20, and its material is carried there. Four moved tier on their own chains: the Woke Right cabal frame from W3 to W2, where a week earlier an applied 0.6 multiplier held it at W3, the China-psyop backlash frame and the doomer-op frame from W2 to W1 on their factors, and the Trump dividend frame from W2 to W3 on its factors. One changed verdict on its chain, the open-borders frame from contradicted to mixed, and the change follows from the verdict rules adopted this week, with no new evidence behind it. No card crossed into the harm instrument this week. Rulings on the record, each dated September 19 except the last six: the Kraft venue story was carded in two lanes and folded into one, the foreign-policy card keeping its factors, its mirror link and its handoff set and taking in the folded card’s carrier links and one marker instance; a card reads mixed only when a check on its narrow reading found independent support, a rule set in the evening that supersedes one recorded earlier in the day, and under it the China-psyop and Sánchez cards stay contradicted; no trajectory reading is applied to any card this week, because the two windows were collected differently; a card at near ignition or ignited takes the W2 floor, since it cannot sit below a warming card on the watch list, and the Kirk death-cult card shows W2 over a computed 2.9; a lineage may resume across a gap week, which is how the recess card continues its chain from W36; the El-Sayed association card reads no coordination identified, with the vice president’s recorded instruction to the House conference stated on it as a party openly directing its own message; dated September 20, a check marked as context carries no weight in a verdict, and where no probative check bears on a wide reading the card says unsubstantiated on what we found, a ruling under which twenty-one checks on twelve cards are context; and, dated the evening and the night of September 20, five more. A narrow check is judged against the narrow reading as the card states it, under which the open-borders and Woke Right cards read mixed at low confidence. The demonization-defense card is folded into the El-Sayed association card on the ground of the Kraft fold: the same six carriers, near-identical claims and three shared handoff items, with the surviving card keeping its factors, mirror and lineage and taking in one carrier, three marker instances, three handoff items and one check. A harm-side handoff ignites a card only when the cleared item’s hostile content carries the card’s own claim or falls on its own target, under which the Clancy and Sánchez cards read near ignition and four items leave the El-Sayed and Kraft handoff sets. Dated the night of September 20 and tightening that test, a cleared item ignites a card only when it names the card’s trigger or its target in its own words, under which the Kirk death-cult card holds at near ignition and the Posobiec post leaves its handoff set. A third verdict, core unsubstantiated by our checks, records a core proposition that no probative check reaches either way, and the Aslan casting card reads there. Private individuals are described by role and never named, whether or not the press names them. Confidence stands at low on nine cards: the Aslan, Sánchez, Clancy, Piddington, Woke Right, open-borders, serve-Israel, Kirk death-cult and Trump dividend cards. The war card and the recess card stand with their shared Khanna and Occupy Democrats carriage noted on each. The lineage record, including the joins we refused, is at hobocode.net/lineage.
One event inside the window produced no card. Mecklenburg-Vorpommern and Berlin voted on September 20, and exit polls for the public broadcaster ARD put the AfD first in Mecklenburg-Vorpommern on about 37 percent, one point ahead of the SPD, and Die Linke first in Berlin on about 24.5 percent. We read the day’s posts on both results against the clustering bar, and nothing cleared it. The posts on the firewall are coalition-strategy argument from a single commentator. The material casting AfD voters as extremists is four single fringe accounts making four different claims, so no two of them share a canonical claim. The two Berlin board threads are anonymous, and one of them is already in the harm record.
The trajectory instrument applied no reading this week. Ten readings were measured, and all ten are published beside their scores as unapplied diagnostics with their confidence marked low. Its measure is a corpus-share rate of change. Volume is matches per thousand collected posts in each window, and the two corpora differ in size, 14,312 posts in the prior window and 11,835 in this one as it stood on September 19, when the ten readings were measured, so a share moves differently from the raw count behind it. The 1,693 posts dated inside the window that were collected on September 20 sit outside those denominators, and with no reading applied, no published score depends on them. Those denominators count every collected post dated inside the window on X, Bluesky and 4chan, where a 4chan row is a thread’s opening post: 2,031 of them in the prior window and 820 in this one. The screening ledger under Method counts 11,015 posts for this window on September 19 because lane screening leaves 4chan out, an anonymous board having no named carrier to offer, and 11,015 plus 820 is 11,835. A rate is a card’s matches divided by the trajectory denominator, times one thousand, so 35 matches over 11,835 posts is 2.96. The change is a percentage change: this window’s rate divided by the prior window’s rate, minus one, so the Woke Right card’s 2.96 per thousand over 0.91 per thousand is plus 226 percent. Each reading recorded on a card sets that change directly against the bands below, which are centered on zero, and no field median is subtracted from it. The instrument can also centre the bands on the median change across the cards with a measurable change, by dividing one plus the card’s change by one plus the median and subtracting one; across this week’s twelve such cards the median is plus 41 percent, and centring on it would leave all ten readings in the same band. The bands are set in advance and recorded as tunable: a change above plus 15 percent reads accelerating and carries 1.3, a change below minus 15 percent reads decaying and carries 0.6, and anything between reads stable and carries 1.0. A multiplier is normally applied where both windows clear the five-post floor and the card is past its first appearance. Three cards met that test, and under a ruling dated September 19 none of the three was applied: our collectors read nothing between 06:33 UTC on September 15 and 22:55 UTC on September 19 and this window was read back from account timelines, so the two windows were collected differently, and a change in share cannot be separated from a change in collection. The Woke Right cabal frame matched 13 posts in the prior window, 0.91 per thousand, and 35 in this one, 2.96 per thousand, a change of plus 226 percent and a reading of 1.3. Applied, it would have taken a computed 3.7 to 4.8 and the card from W2 to W1. A week earlier the same rule read the same card as decaying, and its 0.6 was applied and took a computed 3.5 to a published 2.1. The card’s published score was 2.1 a week earlier and is 3.7 this week, a difference of 1.6, of which 1.4 is the multiplier applied a week earlier and 0.2 is the factors. The Trump dividend frame went from 18 posts to 6, 1.26 to 0.51 per thousand, a change of minus 60 percent and a reading of 0.6 that would have taken a computed 2.4 to 1.4. The Allison inquiry frame went from 46 posts to 8, 3.21 to 0.68 per thousand, a change of minus 79 percent and a reading of 0.6 that would have taken a computed 2.2 to 1.3. Seven further readings are held at 1.0 under the standing rule, and each of those cards shows the reading it holds unapplied with its confidence marked low. Three are first appearances, where the earlier count is chatter from before the card existed, and all three read accelerating at 1.3: the Kraft billionaires-loyalty frame from 7 posts to 107, the Free Palestine hate-mob frame from 5 to 10, and the Piddington frame from 1 to 6. Four have a window under the five-post floor: the recess frame from 2 posts to 9, reading 1.3, and the Kirk death-cult frame from 12 to 2, the doomer-op frame from 6 to 1 and the AI-industry frame from 8 to 2, each reading 0.6. Applied, the Kraft and recess readings would each have carried a W2 card to the 5.0 ceiling, and the doomer-op reading would have taken a top-tier card from 4.4 to 2.6; the rule declines to let a handful of posts do either. The remaining eleven carry 1.0 as a default, with no prior-window base or too few posts in both windows to read. Every share in this paragraph compares two corpora collected in different ways. The prior window was collected mostly live, night by night, and 843 of its 6,238 X posts arrived through the same backfill. This window was rebuilt after 22:55 UTC on September 19 by reading account timelines, which return original posts only, stop at 200 posts per account on X and 100 on Bluesky, and sat beside three days of /pol/ when the readings were measured. A change in share therefore mixes a change in the discourse with a change in the collection, and we cannot separate the two. Every published score this week is therefore the five-factor score alone. The five-post floor is a chosen threshold below which a count is treated as noise, a standing rule under review that has not been validated, recorded as tunable with a date.
Our harm methodology’s gates run first on every item, and this week items inside two carded narratives cleared them and set those cards’ status. Two narratives read as ignited. One reached ignited for the first time on its chain this week, the Kraft billionaires-loyalty frame, ignited in its first window. One continues: the El-Sayed association frame, in its fourth ignited week. Two rulings dated September 20 set the test for a handoff. The first, dated that evening, holds that a cleared item ignites a card only when its hostile content sits inside that card’s own arena: the content that cleared the bar carries the card’s own claim or falls on the card’s own target. A cleared item that mentions a card’s target in one clause while its hostile content falls on someone else stays on the harm side, and the card reads on its markers. The second, dated that night, tightens the first: the cleared item has to name the card’s trigger or its target in its own words, so an item carrying the card’s claim in collective form while naming neither stays on the harm side as well. Every handoff item on every card was read against both tests. On the El-Sayed association card, five Laura Loomer posts put a jihadist label on the nominee himself, and two of them extend it to the party funding him. They are scored on the harm side under the handoff references w38-x-unassigned-066 and 094 and w38-gap-11, 12 and 13 in the Muslim-official-disloyalty family; the last three came with the folded demonization-defense card. Two further Loomer posts sit outside the set under the test: w38-x-unassigned-055, whose hostile content falls on Muslim members of Congress and which mentions El-Sayed in one clause, and w38-sunday-007, dated September 20, which starts from El-Sayed’s positions on ICE, DHS and the police and assigns a violent motive to Muslim people as a class. On the Kraft billionaires-loyalty card two posts are the handoff, one from Cenk Uygur, a carded carrier on this story, and one from Anya Parampil, recorded as w38-x-late-walk-022 and 020 in the Jewish-collective-control family. In each the hostile content falls on Kraft and on the class of wealthy pro-Israel figures the card’s carriers describe. A Sneako post and a /pol/ thread of September 20, posted after Ed Sheeran’s stage apology in Philadelphia, mention neither Kraft nor any donor and place their hostile content on Jewish people as a whole, so they stay on the harm side and outside the set. On the Kirk death-cult card, Jack Posobiec’s post that ‘These people are shooting at us’ and ‘love the violence’ cleared as w38-x-unassigned-107 and stays on the harm side. It carries the card’s claim in collective form, a violent left, and it names neither Charlie Kirk nor the statue and the vigil that trigger the card, and ‘these people’ names nobody, so under the ruling of the night of September 20 it is no handoff and the card reads on its markers. Across the two cards the handoff references cover seven distinct items, and no item is shared between cards. The routing rule is one rule: an item that clears the harm bar is scored and reported only on the harm side; it sets the ignition status of a narrative only under the two handoff tests above, recorded on that card as its handoff reference; and it never counts here as a below-bar example or as evidence for the charge priority score. The card keeps the framing contest and the below-bar remainder, and the harm instrument keeps the harm.
Six narratives sit at near ignition, each on three markers and none on a cleared item. The Woke Right cabal frame returned to it from warming on fresh instances of a trait recode, guilt spread to anyone who stays friends with Carlson, and coinages in daily use by two carriers. A Laura Loomer post of September 20 that cleared the harm bar mentions the faction in one clause, while its subject and the harm finding concern Muslim people; it was weighed for a handoff and left uncredited, and the card stays at near ignition. The China-psyop backlash frame returned to it from warming on a generic ‘they’ working for hostile interests, standing trait labels on the camp, and Levin’s extension of the charge to Republicans and conservatives whose only link is holding the position. The British two-tier frame reached it for the first time on a collectivized ‘they’, guilt spread to the media and politicians as a class, and a mirror lock in which each side cites the other’s worst voices as its proof. That card now records a Labour MP’s condemnation of the Holborn violence, which Rupert Lowe reposted on September 20 with the words ‘Credit where credit is due’; one backbench statement does not settle how the party leaderships responded, so the card’s check on the parties’ response stays mixed. The Kirk death-cult frame holds at near ignition for a fourth week on a generic camp pronoun, one assassin’s act hardened into the camp’s standing intent, and guilt for two unidentified vandals spread to a Senate nominee through her criticism of Kirk alone. Two further cards read near ignition under the arena test after reading ignited a week earlier, and in both the difference follows from the test and from no change in the narrative. The Clancy sympathy-as-endorsement frame was ignited in each of its three earlier weeks. The 4chan post that set its status this week, w38-4chan-fringe-06, assigns the killing of children to women as a sex; it mentions neither the case nor anyone in it, and its tie to the case is its date. Its hostile content falls on women as a sex, where the card’s target is the camp of the defendant’s sympathizers, so it stays on the harm side and the card reads on three markers: a generic ‘leftists’, a standing trait claim about ‘toxic feminists’, and guilt spread to a governor through an unrelated law. The Sánchez frame was ignited a week earlier. Three cleared posts mention Sánchez, w38-x-right-lanes-047, 055 and 150, and in each the content that cleared the bar, a vermin simile, a forecast of crime and the word ‘beasts’, falls on the people who crossed into Ceuta, with the prime minister blamed in a separate sentence or clause that stays below that bar. None is a handoff, and the card reads on a trait recode, guilt spread to the Red Cross, and eliminationist drift, a marker that now holds three RadioGenoa posts dated September 18, 19 and 20, the third reading ‘Capture him and let him rot in prison for crimes on Spanish and European people’. Eleven narratives sit at warming and two at cold, the Massie frame on the FDA and the Allison inquiry frame, which moved from warming to cold with no marker observed in its carrier set. One moved from near ignition to warming under the decay rule with its earlier marker instances retained: Piker’s frame, where no item cleared the harm bar in the review window and last week’s trait register did not recur. Several markers were weighed, left uncredited and recorded on their cards: the president’s ‘Treasonists, Traitors, and Leakers, BEWARE!’ against eliminationist drift, because it states no end state; Massie’s demand to prosecute one former official, because it extends to no camp; Rupert Lowe’s and Steve Laws’s call to proscribe Stand Up To Racism, a group outside that card’s framed camp; and two insults weighed against the dehumanizing-metaphor marker, Loomer’s mental-illness label on people who chant for Palestine and Piker’s ‘cattle’ for defenders of the bomb sale, which sit outside the disease, vermin and machine framings the marker keys to. Each marker instance is logged on the card that carries it with its own source and date, and clicking a marker opens that card.
Routing also ran in the other direction. Candidates that cleared the harm gates were routed to The Weather Report, which publishes its own counts for this window, and none of them is carded here. Several adjacencies were screened there and sit on no card. Posts by Laura Loomer and one by Mark Levin dated inside the window sit in the confirmed harm corpus, and none is cited on the Woke Right card. The reading of the Swedish election result as a left that imported its voters sits almost entirely in that corpus, in a script on the votes of citizens of immigrant background, and is carded nowhere here; its below-bar remainder is one commentator, under the clustering bar. Rupert Lowe’s post on the home secretary cleared the harm bar as w38-x-right-lanes-113 on its migrant-directed clause and is excluded from the two-tier card. The Piddington frame’s two largest posts, from Katie Hopkins and Jayda Fransen, cleared the harm side on their migrant-directed language and are cited nowhere on that card. Posobiec’s one-word dehumanizing label over a headline about Governor Healey cleared as w38-x-unassigned-077 and is credited to no marker on the Clancy card. Loomer’s post on Florida Senate nominee Angie Nixon cleared as w38-x-unassigned-096; it concerns a different candidate and race, so it is outside the El-Sayed handoff set, and it is uncited on the Kirk card. Greenwald’s post on Miriam Adelson was screened on the harm side and rejected there as w38-x-late-walk-024, and appears on the serve-Israel card as an edge item only. When lane screening ran, twenty-eight of the seventy posts fed to the education-and-culture lane sat in the confirmed harm corpus and were excluded from every carrier set before coding.
Near ignition and ignited are review statuses. They escalate our checking cadence and record where an item crossed to the other instrument, and they carry no verdict about anyone. They do not make anyone on these cards an extremist, and we ask readers to leave that use aside as well.
Seventeen of the 21 narratives drew a response we could find: thirteen record a mirror that answered, and four record a locked pair. Four ran unanswered in our collection: the Aslan casting frame, where the pushback we found was mockery in entertainment columns; the Piddington frame, where the Liberal Democrats and the villagers argued the site and framed nobody’s motives; the Massie frame on the FDA, where no collected post defends Marks or answers the prosecution call; and the Allison inquiry frame, where every in-window post on the hearings came from the two outlets that carry it. Answered and locked are distinct statuses, and the three counts together account for all 21 cards. A lock means each side runs on the other’s statements as its primary evidence. The unanswered count needs a caution this week that it does not usually need: our X collection never includes replies, the week was rebuilt from account timelines, and Bluesky and news for September 16 to 18 are thin, so an answer made in a reply thread, by an untracked account, or on one of those days would not have reached us. Read carefully, the answered count is a finding about this instrument as much as about the discourse: what our gate declines to code is the form the response took, most often merits argument, documentation or reporting with no counter-imputation in it. The Sánchez frame is again the cleanest case, answered by a government arguing substance and electoral ground, and the Clancy frame the next, answered through six newsrooms carrying the holdout’s statement as news. The recess frame moved the other way: the accused camp now answers with a motive claim of its own, Speaker Johnson’s ‘a publicity stunt’ and Levin’s charge that Massie is doing the Democrats’ work, each aimed at one named member and recorded on the card as the answer.
A polarization monitor that codes one side more than the other needs to say so in the same place it reports its findings. The camp label is derived from each card's carrier set. The label is coarse. The carrier-set paragraph on each card remains the authoritative account of who carried a narrative.
The distribution by origin side, which our methodology requires in every report, is set out here. Each card is classed from its origin camp as right-origin, left-origin, or cross-party or other, and the table above lists every card with the basis for its classing, so a reader can class any card differently and recount.
| Origin side | Cards | Mixed evidence | Contradicted by our checks | Core unsubstantiated by our checks | Low confidence | Ignited | Near ignition |
|---|---|---|---|---|---|---|---|
| Right-origin | 13 | 10 | 2 | 1 | 7 | 1 | 6 |
| Left-origin | 3 | 3 | 0 | 0 | 1 | 0 | 0 |
| Cross-party or other | 5 | 5 | 0 | 0 | 1 | 1 | 0 |
| Total | 21 | 18 | 2 | 1 | 9 | 2 | 6 |
Thirteen of the 21 cards are right-origin, three are left-origin, and five are cross-party or other. The right-origin cards are the Aslan casting frame, the Woke Right cabal frame, the El-Sayed association frame, the Clancy frame, the Kirk death-cult frame, the British two-tier frame, the Piddington frame, the Sánchez frame, the open-borders frame, the Allison inquiry frame, the Massie frame on the FDA, the China-psyop backlash frame and the doomer-op frame. The left-origin cards are Piker’s media-operation frame, the Trump dividend frame and the AI industry-money frame. The cross-party or other cards are the reckless-war frame, the serve-Israel frame and the Kraft frame, each carried by left-populist voices together with figures from the right; the recess frame, carried by the bloc around the Epstein petition; and the Free Palestine hate-mob frame, whose pro-Israel advocacy carriers form a camp defined by the Israel question. Every verdict other than mixed runs in one direction this week, and all three fall on right-origin cards: the Sánchez frame and the China-psyop backlash frame read contradicted by our checks, and the Aslan casting frame reads core unsubstantiated by our checks. All eight left-origin and cross-party cards read mixed, as do ten of the thirteen right-origin cards. Seven of the nine low-confidence verdicts sit on right-origin cards. One of the two ignited cards is right-origin, the other is the cross-party Kraft frame, and all six cards at near ignition are right-origin. Two caveats govern how far that split can be read. Our monitored roster is weighted toward right-of-center accounts on X, and an engagement floor sits under every lane, so the card set holds more right-origin narratives before any check is run, and a side with more cards has more chances at every verdict and every status. The split is measured and disclosed; it is never enforced, and no verdict was set to move it. We do not present it as a measurement of where polarizing framing happens. The Michigan Senate fight, the Kraft venue story and the president’s AI posts each produced carded narratives on both sides, and where the opposing frame existed and we could not card it, the card says so.
8 narratives we carded last week are not carded this week. We do not read that as decay. A narrative can be absent because it lost its carriers, because it did not clear the bar this week, or because this week's collection did not surface it. These records cannot separate those, so we report the absence and stop there.
| Narrative | Continues | Tier |
|---|---|---|
| The Woke Right cabal frame, fourth week: Carlson's bomb-transfer report and Kasparian show read as 'a coordinated Islamic influence operation' and 'two hands of one Woke beast' | The Woke Right cabal frame hardens into 'an Islamic movement' as Carlson defends El-Sayed on 9/11 | W3 to W2 |
| The vice president tells House Republicans to run against El-Sayed everywhere, and the NRSC, AIPAC, the RJC, and Levin carry Piker's al-Qaeda line as proof of what El-Sayed and the Democrats funding him are | The Rogers campaign and its super PAC carry the Piker association into the 9/11 anniversary as proof that El-Sayed gives comfort to the hatred behind the attacks | no change |
| Piker, Uygur, and Blumenthal read the Axios al-Qaeda headline and the disavowal questions as a coordinated media and party-establishment operation against the left | Piker and allied left carriers recast the Rogers anniversary attack and the Mamdani ceremony fight as proof the right does not care about 9/11 and uses it only to spread hatred | no change |
| The reckless-war frame at official tier: Warren alleges a cover-up of the number of troops killed, Sanders attacks the president's '$4.44 a gallon' remark, Jeffries calls it a 'reckless war of choice' that failed, Khanna says Johnson canceled votes to avoid one on stopping the war, and Massie files impeachment articles against Hegseth | Warren, Jeffries, Murphy, and Khanna carry the reckless-war frame as crude tops $100, the Saudi bypass shuts, and Trump promises an end right after the midterms | no change |
| The serve-Israel frame on the $2.8 billion bomb sale: Massie says Adelson's money 'worked' and 'we're fighting the war in Iran for them,' Uygur says 'We serve Israel over our own citizens,' Piker calls it an 'israel first admin,' and Greenwald calls Jeffries 'AIPAC's most loyal soldier' | Greenwald, Piker, and Blumenthal read the state-level threats against the UK and Jeffries' Moskowitz pick as US politicians serving Israel over their own constituents | no change |
| The holdout's statement and the threats against him recast as a female and left-coded camp that sides with a child-killer and wants the dissenting juror dead | The juror interviews recast as a confession that a female and left-coded camp rigged the trial, then as a leftist smear of a Black holdout | no change |
| The vandalized Kirk statue and a protested vigil recast as proof that a violent left wanted Kirk dead and is choosing who to murder next | The anniversary memes recast as proof that the left celebrates Kirk's murder and is a death cult that wants conservatives dead | no change |
| The $5,000 dividend's second week: the bribe reading survives among left-coded commentators as Democratic officials go quiet and Senate Republicans shrug | The conditional $5,000 dividend recast as proof that a losing Trump is bribing voters he cannot win and lying about the checks | W2 to W3 |
| Lowe and Robinson, second window: the Portsmouth mask charge and the Holborn attack recast as an establishment that criminalises patriots, labels them nazis and looks away when they are hurt | Lowe, Robinson and Anderson: the mask ban and the policing of Dover and Portsmouth recast as a British establishment protecting migrants, the left and the burqa while it targets patriots | no change |
| RadioGenoa and Robinson, third window: Sánchez as a criminal to be arrested and jailed for flooding Europe, in the week the European Parliament blamed his regularisation policy as a pull factor | RadioGenoa, second window: Sánchez and the socialists as criminals flooding Europe, widened to the Red Cross and the NGOs in Ceuta as enemies, against the declassified record | no change |
| The Senate Republican Conference captions a Schmitt clip 'Socialist Democrats would make America a borderless welfare state,' and Cotton writes 'Democrats want to defund ICE': the immigration predicate back in force in a week when a House Democrat posted 'Abolish ICE,' seventh consecutive window | SenateGOP/SenJohnBarrasso: 'Democrat socialists' and a Canada preview around the Dallas convention, with the ICE predicate reduced to one post, sixth consecutive window | no change |
| Press For Truth and Rebel News carry the Allison Inquiry into a second week as the hearings the media ignored and the testimony the CBC does not want Canadians to see | Independent-media carriers recast the Allison Inquiry's empty benches and thin coverage as proof that Ottawa and the legacy press do not care about vaccine-injured Canadians | no change |
| The president calls the data-center backlash a SICK conspiracy that only China is happy about and warns 'Treasonists, Traitors, and Leakers,' while Levin extends the charge to conservatives | The data-center backlash as a Chinese and far-left manufacture, restated by the Treasury secretary as Chinese agitprop, by the president as a China PR thing, and syndicated to forty local stations | W2 to W1 |
| The president calls AI fear a hoax and his science adviser calls the safety push Trust & Safety 2.0, as the doomer-op frame becomes the administration's line | Jacob Coxon's resignation read as a funded doomer op run through mega donors and the corporate media | W2 to W1 |
| Van Hollen calls the administration's refusal to act on AI 'part of an implicit deal,' Sanders faults AI CEOs' super PAC spending, Warren calls the labs' antitrust ask 'transparently self-serving,' and Khanna says 'the tech lords came after me': the industry-money frame after the president's 'hoax' posts | The Coxon warning read as proof the AI industry's oligarchs are buying Congress to escape any check | no change |
A logic verdict rarely moves between weeks. The Senate Republican Conference captions a Schmitt clip 'Socialist Democrats would make America a borderless welfare state,' and Cotton writes 'Democrats want to defund ICE': the immigration predicate back in force in a week when a House Democrat posted 'Abolish ICE,' seventh consecutive window moved from contradicted by checks to mixed evidence.
Each line is one narrative compared with itself a week earlier, joined on a key assigned when the narrative was first carded rather than on any similarity between the two weeks' text. The two instruments are reported separately and never combined into a single direction, because they measure different things on different scales.
One line in the ledger above records a verdict that differs from a week earlier: the open-borders narrative, from contradicted to mixed. The difference follows from the verdict rules adopted this week, dated the evening of September 19 and the evening of September 20, and it reflects no new evidence for or against the frame. The Clancy narrative reads mixed, as it did a week earlier, so the ledger carries no line for it. One of the eight narratives listed as not carded has a known cause: the demonization-defense card was folded into the El-Sayed association card under a ruling dated the evening of September 20, and its material is carried there.
Every item first runs through our harm methodology's gates. Items that clear them are reported in The Weather Report, under the routing rule stated in the ignition section above. Items that do not clear them run through a two-prong polarization gate: the commentary's primary object must be the motives or character of the camp being framed, which may be the carrier's own, with the event's merits set aside, and at least one of eight named polarizing operations must be present. Heat alone never codes, and documented, sourced motive claims never code. This instrument states whether an item cleared our harm bar and how a camp-versus-camp narrative scored. It makes no legal finding in any jurisdiction, and the countries on this board do not share one speech standard.
Each narrative's base score is the five-factor geometric mean, CPS_base = (R·I·P·E·L)^(1/5): Reach, Intensity, Penetration, Elite adoption, and Lock, each anchored 1 to 5, combined by geometric mean so no single factor can dominate. The published charge priority score is that base times a trajectory multiplier when one is applied, capped at 5: CPS = min(5, CPS_base × τ), where τ is 1.3, 1.0 or 0.6. No multiplier is applied this week, so every published score is its base score. The score is plotted on a 0 to 5 axis. Watch tiers: W1 at 4.0 and above, W2 at 3.0, W3 at 2.0, W4 below. The ignition floor acts on the tier only: a card at near ignition or ignited is listed no lower than W2, and its computed score is shown beside it unchanged.
Scores are published to one decimal. Each of the five factors is a whole-number judgment on a 1 to 5 anchor, and the geometric mean of five such judgments carries no second decimal a reader can act on, so a second decimal stated a precision this instrument does not hold. Rounding puts some cards close to a tier cut, and the page says so where it happens: a score within 0.15 of a tier edge is marked at the boundary, and the mark names the single factor that would carry the card across the edge, the value that factor would have to take, and the score the card would then read. Six cards carry a boundary mark this week. A boundary mark records how thin a tier placement is and is no forecast that the factor will move.
Coordination is a separate verdict on every card, and every card in this week’s set reads no coordination identified. That verdict says what our record holds: no linkage between carriers beyond shared timing, no bridge account, no phrasing copied from one account to another, no named hub and no funder we could identify. It replaces organic parallel, which said more than we had shown. Independent origin is a positive finding and needs positive evidence for it, so the phrase is now reserved for cards that carry such evidence, and an absence in our record stops reading as a finding about the carriers. Convergence on one story within hours is the ordinary behavior of a commentary field watching the same trigger, and this instrument records it as nothing more.
The ignition instrument runs beside the score. It never changes the five factors or the computed score. It can change where a card is listed: a card at near ignition or ignited is listed no lower than W2, shown with its computed score beside it, and this week the Kirk death-cult card is listed at W2 over a computed 2.9. Seven markers, each keyed to a construct our harm methodology already defines, produce a status: cold, warming, near ignition, or ignited. Ignited means an item actually cleared the harm bar and is reported there.
The logic test states each frame's claim as a prediction-generating hypothesis and checks the predictions against what is observable, with sources. The predictions are our own operationalizations of the frame, and reasonable readers may formulate different checks; where a frame supports both a narrow and a strong reading, we test both and say which one a check touches. A card’s verdict aggregates its predictions under a published rule, tightened on the evening of September 19, on September 20, on the evening of September 20 and on the night of September 20, and every prediction is shown with its own result in the drill. The verdict scale carries five values: supported by our checks, mixed evidence, contradicted by our checks, core unsubstantiated by our checks, and not testable from available evidence. A card reads supported when the imputed motive survives the checks we ran. It reads mixed when at least one check on its narrow reading came back true or mixed on independent evidence. Three things add no such support: a narrow reading we did not test or could not resolve; material produced by the frame’s own carrier, unless its underlying evidence was separately verified; and political adoption of the claim, such as a resolution, a vote or an endorsement, which shows a claim entering the mainstream and measures nothing about whether it is true. Where no narrow reading holds, the next two verdicts turn on the claim’s core proposition, the load-bearing thing the frame asserts, which on these cards is usually a motive or a standing trait imputed to a camp. A card reads contradicted by our checks when a probative check reached that core proposition and failed against what is observable. A card reads core unsubstantiated by our checks when no probative check reaches that core proposition in either direction, so our record neither carries the claim nor tells against it, and a reader learns what we looked for and did not find. The core-unsubstantiated verdict opened on the night of September 20. Before it existed, a card whose core proposition our checks never reached read contradicted, which stated more against a frame than our record carries. The record stores the verdict as unsubstantiated and every page shows it to readers as core unsubstantiated by our checks. A check marked context only records a fact that cannot tell the claim’s readings apart; it carries no weight in the verdict, and the drill shows it as ‘context only, carries no weight’. A check whose observation would hold in ordinary politics whatever the truth of the claim, such as a party funding its own nominee or a party’s senators objecting to a proposal on fiscal grounds, is marked the same way. Where no probative check bears on a wide reading, the card says that reading is unsubstantiated on what we found, and it counts neither for the claim nor against it. A check on a narrow reading is judged against that reading as the card states it: where it is true or mixed on independent evidence and could have come out the other way, it is probative and the card reads mixed, even where it lends nothing to the wide reading, and the same fact carries the same weight on both cards of a pair. A mixed verdict on a compound claim substantiates none of its clauses on its own. Verdicts report how each frame's predictions fared against what is observable, as we operationalized them; they are never a ruling on the underlying policy question, and a camp's broader concern can remain entirely legitimate while its narrative's checks come back contradicted.
Every quoted item passed a two-stage verification: a first check against its source, then an independent adversarial re-check of every quote, link and date. Items that did not meet that standard were excluded before scoring. The screening ledger for this week, as lane screening ran on September 19: 11,015 posts from X and Bluesky and 318 news items sat inside the window, with 4chan left out of lane screening because an anonymous board has no named carrier to offer, and the trajectory denominators in the queue section add its 820 opening posts for 11,835; 7,111 posts cleared the engagement floor, a deterministic narrowing pass fed six research lanes 70 posts each and the public-health lane the 53 that matched it, 623 candidates were carried into the seven lanes with their news rows, and the 21 narratives that stand after one fold cite 88 distinct carrier items between them. The narrowing pass ranks the floor-clearing posts by lexicon-and-beat match and then by engagement, capping each lane feed at roughly seventy posts. A post matching more than one lane goes to the lane it matches best. Lane screening ran on September 19, after the backfill, over the record as it then stood. Collection continued through September 20 and added 1,693 posts and 131 news items dated inside the window, which brings the in-window news items to 449, 318 of them collected before September 20; they were read against the cards and for any new narrative, several cards carry dated additions from that reading, and no new card cleared the bar. The record therefore runs from September 14 to about 19:00 UTC on September 20, and September 20 is a partial day. Candidates drop out for several reasons, and the polarization gate is only one of them: no codable text, no readable source, an object the identification bar puts out of reach, or a primary object that is the event’s merits, with no camp’s motives in view. Every item was also checked against the harm methodology’s gates before it could be carded here; candidates that cleared them were routed to that instrument and carded nowhere here, and the ignition section above records where that happened this week. Clustering requires a shared canonical claim about the same trigger event on the same fault line, and a card requires either two independent carriers or a single carrier at official or major-outlet tier. Research lanes are collection beats; fault lines are the terrain the narratives land on. We ran the same queries for both camps in every lane, and we searched for a mirror before any card recorded one as unanswered. Research and drafting run through Claude under this methodology; every published classification passes the documented verification chain, and Jerm holds final editorial responsibility.
The roster is the frame we track. What we collected inside it is smaller, and what returned a post is smaller again. Every number here is counted by script from the roster file, the handle lists and the collection database for the window of 2026-09-14 to 2026-09-20.
| Source | Handles on the roster | Handles collected | Handles that returned a post | Posts collected |
|---|---|---|---|---|
| X | 46 | 137 | 137 | 6,972 |
| Bluesky | 4 | 729 | 591 | 5,270 |
| 4chan boards | not applicable | 2 | 2 | 1,286 |
| News outlets | not applicable | 596 | 322 | 449 |
The roster holds 1,034 accounts across 23 categories, and 136 of them sit in more than one category, so the category column below counts more accounts than the roster holds. Handles collected counts the handles that stored a post during the window; a handle the walk read that returned nothing leaves no row, so a walk that reached an account and found silence looks the same in this table as one that never ran.
| Roster category | Accounts |
|---|---|
| Anti-Establishment, Socialist and Anti-War Media | 74 |
| Great Replacement and White Nationalist Accounts | 73 |
| Far-Right, Nationalist and Culture War Figures | 65 |
| Pro-Israel and Zionist Accounts | 65 |
| Alternative Media, Sovereign and Conspiracy Theorists | 63 |
| Far-Left, Anti-Imperialist and Communist Commentators | 62 |
| Pro-Immigration, Sanctuary and Migrant Rights Accounts | 58 |
| Anti-Semitic and Syncretic Conspiracy Accounts | 56 |
| Black Radical, Pan-Africanist and Marxist-Leninist Factions | 56 |
| Islamophobic, Anti-Islam and Counter-Jihad Accounts | 56 |
| Anti-Establishment Tech, Finance and Geopolitical Dissidents | 55 |
| Anti-Illegal Immigration and Border Nationalist Accounts | 55 |
| Anti-China and Dissident Accounts | 50 |
| Pro-China and State-Affiliated Media Accounts | 49 |
| Extreme Christian Nationalist and Dominionist Accounts | 47 |
| Anti-Vaccine and Medical-Freedom Extremist Accounts | 46 |
| Anarchist, Eco-Radical and Left-Libertarian Organizers | 45 |
| Radical Islamist and Salafi-Jihadist Ideologue Accounts | 44 |
| Pro-Feminist and Intersectionality Accounts | 43 |
| The Manosphere and Red Pill Ecosystem | 41 |
| Anti-Feminist, Traditionalist and Counter-Feminist Accounts | 39 |
| Anti-Christian, Militant Atheist and Secularist Accounts | 38 |
| Anti-Technology and Anti-AI Extremist Accounts | 26 |
The X walk below covers the 46 of 152 X handles we can match to a roster category by name. The other 106 are handles our records hold without a category, and they are collected and read the same way. 14 categories returned an X post this week.
| Roster category | X handles on the roster | Collected | Returned a post | Posts |
|---|---|---|---|---|
| Islamophobic, Anti-Islam and Counter-Jihad Accounts | 8 | 7 | 7 | 761 |
| Anti-Illegal Immigration and Border Nationalist Accounts | 12 | 12 | 12 | 727 |
| Far-Right, Nationalist and Culture War Figures | 12 | 10 | 10 | 419 |
| Great Replacement and White Nationalist Accounts | 7 | 6 | 6 | 343 |
| Alternative Media, Sovereign and Conspiracy Theorists | 2 | 1 | 1 | 195 |
| Anti-Establishment, Socialist and Anti-War Media | 5 | 5 | 5 | 108 |
| Far-Left, Anti-Imperialist and Communist Commentators | 5 | 4 | 4 | 82 |
| Pro-Israel and Zionist Accounts | 3 | 3 | 3 | 65 |
| Extreme Christian Nationalist and Dominionist Accounts | 2 | 2 | 2 | 57 |
| Pro-China and State-Affiliated Media Accounts | 4 | 4 | 4 | 49 |
| The Manosphere and Red Pill Ecosystem | 1 | 1 | 1 | 31 |
| Anti-Establishment Tech, Finance and Geopolitical Dissidents | 2 | 2 | 2 | 26 |
| Anti-Semitic and Syncretic Conspiracy Accounts | 2 | 1 | 1 | 8 |
| Pro-Immigration, Sanctuary and Migrant Rights Accounts | 5 | 1 | 1 | 2 |
| Anarchist, Eco-Radical and Left-Libertarian Organizers | 1 | 0 | 0 | 0 |
This week’s collection has a hole in it, and every finding above should be read through it. Our collectors read nothing between 06:33 UTC on September 15, the last successful read before the outage, and 22:55 UTC on September 19, the first after it. X was backfilled after that hour by reading the timelines of all 152 tracked accounts; 137 of them have posts inside the window. A timeline read stops at 200 posts, so an account that reached the cap has a permanent hole at the start of its week: JackPosobiec’s collected week begins on September 16, Rebel News’s on the evening of September 15 and RadioGenoa’s partway through September 14, and all three are carriers on this queue. Replies and reposts are never collected, in this or any week, so an answer made in a reply thread is invisible to us. Bluesky was backfilled at 100 posts per account. 4chan holds /pol/ threads for September 14, 15, 19 and 20 and none for September 16, 17 or 18: nothing created on /pol/ between about 06:29 UTC on September 15 and 09:11 UTC on September 19, about four days, is in the record. The 45 threads dated September 16 to 18, which are 7, 4 and 34 by day, are all from /g/, and the /pol/ threads that cycled off the board during the outage cannot be recovered. News for September 16 to 18 is a partial backfill, in places titles only, and the trigger scan compensated by web search. Collection ran live again from 22:55 UTC on September 19, and the record closes at about 19:00 UTC on September 20, so that day is partial, and reaction to the German state results posted after that hour is outside it. The consequences run through the page. Every reach figure is a floor. A low count for a camp’s carriage, an unanswered mirror, a single-carrier reading and a lineage recorded as not recurring may each partly reflect the gap. The Clancy lineage’s lead carrier has six posts in the collection this window and none on the case, and the right-populist accounts that Levin answers on the data-center card have no in-window posts in it, so the dissent inside the right on that card is documented from news reports. Outside the gap, our footprint covers the open web and X; surfaces we cannot systematically see, including several large closed platforms, are underrepresented, and a narrative living mainly on those surfaces would read here as smaller than it is. Where a first-party post was not retrievable, quotes are cited to the outlet that carried them.
Penetration is scored from carrier-set breadth as a stated judgment; no within-camp sample was drawn, so every penetration score is an unvalidated qualitative estimate, and with no defined denominator behind it the true spread within a camp could sit above it or below it. Reach is scored on qualitative anchors, and platform-reported view counts, where shown, are floors over time and do not count unique readers. This week they are floors twice over, because the collection behind them has a gap from 06:33 UTC on September 15 to 22:55 UTC on September 19.
The trajectory multiplier moved no score this week. Ten readings were measured and none was applied. On three cards both windows clear the five-post floor and the card is past its first appearance, and under a ruling dated September 19 those readings are published beside the score as unapplied diagnostics. They rest on two-window pairs of 13 and 35 posts, 18 and 6, and 46 and 8, and the two windows were collected in different ways: the prior one mostly live, night by night, with 843 of its 6,238 X posts arriving through the same backfill, and this one by timeline reads after 22:55 UTC on September 19 that exclude replies and reposts, stop at 200 posts per account, and sat beside a Bluesky backfill of 100 posts per account and three days of /pol/ when the readings were measured on September 19. The 1,693 posts collected on September 20 are outside both the counts and the denominators. A share of one corpus set against a share of the other measures the collection as well as the discourse, and we cannot say in what proportion. On seven cards the reading is held under the standing rule, on a first appearance or on a window under the floor. On the other eleven the multiplier is a default with no readable base. One effect carries across weeks: the Woke Right card’s published score a week earlier included an applied 0.6, so most of the difference between 2.1 and 3.7 is that multiplier, and the Trump dividend and Allison inquiry cards sit at W3 on their factors where an applied reading would have put them at W4. The five-post floor is a chosen threshold below which a count is treated as noise, a standing rule under review that has not been validated, recorded as tunable with a date; readings this early say more about small windows than about acceleration.
The published product carries a selection bias of its own. Not-testable verdicts are expected to be common, W4 narratives are logged and go unwritten, and false positives are the first-class error we optimize against, so these reports over-represent charged narratives that happened to generate falsifiable predictions, which is no random subsample of charged narratives. That effect lives in the publication layer, apart from any collection or platform effect, and we name it here so readers can weigh it.
Escalation markers this week are recorded as marker instances, each with its own source URL and date on the card that carries it. The status a set of markers produces is still a judgment under the marker rules, and clicking a marker opens the card so the judgment can be read against its evidence.
The charge priority score is an operational priority score for our own review queue. It does not measure polarization. Its reach, penetration, and elite-adoption factors move together in practice (a narrative carried by elites tends to score up on all three), so nearby scores should be read as a tier. Scores are displayed to one decimal, because five whole-number judgments combined by geometric mean support no finer reading than that; a gap of a tenth or two is inside the noise of those five factors, and cards inside a tier are listed by score for reference, with no ranking drawn from a gap that small. Rounding can place a card either side of a tier cut, so a score within 0.15 of a tier edge carries a boundary mark naming the one factor that would move it, and six cards carry that mark this week. The factor anchors, the marker list and the tier cutoffs in this instrument are conventions we are still calibrating, and every one of them is recorded as tunable with a date attached, so a later change to any of them stays visible against this week.
This research is unpaid volunteer work. If you want it to continue, the fundraiser is the way to help.
Support the fundraiserBlind verdict agreement: 25 of 28 as of 2026-07-24