Crowd Counting Consortium

register entry crowd-counting-consortium-aio-20261031 OBS-1d7507d26f51 · file as first run sha256 8dc7b14b4e6c31ca…

• T1-1 definition: 'a joint research project that tracks and collects public data on political protests, marches, rallies, strikes, and acts of civil disobedience across the United States.' (two markers)
• T1-2 leadership & origin: 'Co-founded in 2017 by Erica Chenoweth (Harvard University) and Jeremy Pressman (University of Connecticut) following the Women's March on Washington.' (two markers)
• T1-3 institutional home: 'Operated in partnership with the Nonviolent Action Lab at the Harvard Kennedy School’s Ash Center for Democratic Governance and Innovation.' (one marker)
• T1-4 scope: 'Systematically logs public, nonviolent (and broader political) crowd events across all U.S. states and territories with no minimum size requirement. This includes historic movements such as the 2017 Women's March, 2018 March for Our Lives, 2020 Black Lives Matter protests, campus encampments, and subsequent political demonstrations.' (three markers)
• T1-5 scale: 'The project has tracked hundreds of thousands of individual demonstrations and rallies across the U.S. since 2017.' (one marker)
• T1-6 monthly data: 'Published regularly on Harvard Dataverse.' (one marker)
• T1-7 weekly updates: 'A compiled and augmented version of the dataset is updated weekly on Wednesdays via the Nonviolent Action Lab GitHub Repository.' (one marker)
• T1-8 interactive tools: 'You can explore data visuals and project updates directly through the Counting Crowds Website.' (one marker)

• B1 Harvard University — Crowd Counting Consortium – Ash Center card; B1-refetch.html sha256 2837906be77ea4ef, HTTP 200 at 2026-10-31T22:58:49Z; B1.html the census copy crowd-counting-consortium--home (sha256 bd7b38580b7baa3e), text identical
• B2 Counting Crowds — About - Counting Crowds card; B2.html sha256 be076d67b1488910, HTTP 200 at 2026-10-31T22:58:49Z
• B3 boB Rudis — Crowd Counting Consortium 2025 U.S. Protest Event Data card; B3.html sha256 c8f7088d06b5055f, HTTP 200 at 2026-10-31T22:58:50Z
• B4 Counting Crowds — crowdcounters – Counting Crowds card; B4.html sha256 d9a2fe6edc796efd, HTTP 200 at 2026-10-31T22:58:51Z
• B5 Policy Commons — Crowd Counting Consortium - Policy Commons unreachable: https://policycommons.net/orgs/crowd-counting-consortium/ HTTP 403 (bot check); URL undetermined; card title and snippet only
• B6 X·crowdcounting — Crowd Counting (@crowdcounting) / X card; B6.html sha256 5e56ffb881defc7a, HTTP 200 at 2026-10-31T22:58:52Z
• B7 Wikipedia — Crowd Counting Consortium (no article) no article: raw wikitext HTTP 404, rendered page HTTP 404; full-text search HTTP 200 (five articles mention the name); Crowd counting raw HTTP 200

E_A, the archive's entity: the same project, in the maryleelabor census record, the census's anchor and the capture record: a public interest and scholarly project publishing an event dataset, in the census's area of labor action and tagged 'NOT a labor dataset', whose unit is a political protest event reported in the United States with crowd numbers; whose record keeps political crowds (occurrence, themes and scale, as reported or submitted with a verifiable source), held by a university project funded by foundations and university centres; whose losing side is not recorded, the positions argued inside a movement left unanswered; read by the census as a holder independent of the movements and governments it counts, which keeps the crowd and leaves the movement's internal argument unrecorded; under the question #1725 draws from the survival of a losing position through organ, holder and route, with #1725's limit that no cross-organization dataset of left splits was found; and whose composition at its own name is faithful on founders, home and data, puts 'nonviolent' first and leaves out the verification and privacy rules (Capture Registry, 2026-10-31) ML-01 ML-04 ML-06 ML-08 ML-09 ML-10 ML-17 ML-18 ML-23 ML-25 ML-28 O1725-01 O1725-02 CAP-1-02 CAP-1-03 CAP-1-04; defining papers #1725

E_F, the field's entity (type C — a conventional reading holds it (public entity: a university project publishing a dataset of US protest events)): the joint project of Harvard Kennedy School and the University of Connecticut, part of the Nonviolent Action Lab at the Ash Center, that emerged from Jeremy Pressman and Erica Chenoweth's effort to estimate the participants in the Women's March of 21 January 2017 and collects publicly available data on political protest events reported in the United States (rallies, marches, demonstrations, riots and acts of civil disobedience; crowds at meetings, teach-ins, workshops, panels, fundraisers and town halls excluded), posting only records confirmed through fact-checking and sharing no identifying information not already public, with aggregate crowd numbers for each event published monthly on the Harvard Dataverse in three phases since 2017 (and, on its blog, in monthly Google Sheets and a weekly compilation on GitHub); co-directed by Chenoweth and Pressman; 200,000 protests and rallies tallied by August 2024; its data used to show the size and the nonviolence of named mobilizations; described by a third party's page, whose text carries a model's interface string, as a dataset of nonviolent public events; with no Wikipedia article

ρ = extends: The archive reads the Crowd Counting Consortium as its own page presents it and adds, on the field's own object: the census's question of whose record it keeps, who holds it and whether the losing side is kept; the answer that it keeps the crowd and records nothing of the positions, splits or dissent inside the movements it counts, and so answers neither labor action as such nor which tendency organized an event; the census's reading of it as a holder independent of the movements and governments it counts; #1725's conditions for a losing position's survival and its stated limit that no dataset of left splits was found; and the composition layer's narrowing of its scope and omission of its rules at its name. ML-25 ML-10 ML-17 ML-18 ML-23 O1725-02 O1725-01 CAP-1-03 CAP-1-04. Guard: ρ = extends claims an addition and no more (§1.2a). It does not make the field's sources wrong at their dates. Where an archive claim stands against a field claim (CAP-1-03 and CAP-1-05 against F34, the third party's 'nonviolent'; ML-13, no total stated, beside F44, the X account's 200,000), both are composed under P_coh within the one ontology. #1725's claims are claims on the records of the left and its movements, the class into which the census enters the consortium, never on the entity by the composer's judgment.

The case: equal terms, ontology included (v0.9). The archive reads the field's entity as the entity. The maryleelabor census describes the Crowd Counting Consortium from its own page, each field with its basis ('operator'), and answers for it the census's question of whose record it keeps, who holds it and whether it keeps the losing side in a form that could be taken up later; its one reading of its own, ml_reading, takes the dataset as a holder independent of the movements and governments it counts, which keeps the crowd and leaves the movement's internal argument unrecorded. The census's anchor, #1725, supplies the question's terms (organ, holder, route; the losing position) and states that no cross-organization dataset of left splits was found; it does not name the entity, and the census enters the entity under its question. The capture record checks the composition against the consortium's page. No other deposit names the entity or says anything of protest event data or of nonviolent action as a class (D 0, R 0; three class matches read and rejected; the hop from #1725 rejected). No defining paper or record contests the field's ontology of the entity or supplies one the field lacks: the archive adds the census's question and its answer on the field's own object (a dataset of political crowds that keeps no record of the positions argued inside the movements it counts) and the composition layer's account at the name. ρ = extends, the equal-terms case (§0.9, §1.2a).

#ML census.json organizing_question: Whose record does it keep, who holds it, and does it keep the losing side in a form that could be taken up later?

#ML census.json rows[slug == 'crowd-counting-consortium'].record['losing_side']: not recorded. The data are used to study protest, counter-protest and state response; the saved page shows nothing of the positions, splits or dissent inside the movements counted

#ML census.json rows[slug == 'crowd-counting-consortium'].ml_reading: none stated. The archive reads the dataset as a holder independent of the movements and governments it counts, whose record has been used against prevailing narratives (that the 2020 uprising was violent; that there was 'no resistance' in 2025). It keeps the crowd and leaves the internal argument of the movement unrecorded.

#1725 Conclusion, ¶1 (line 369): The decisive variable in the record is what happens to the losing position after the decision. The amount of conflict matters far less. Synthesis followed where the defeated critique kept three things: an organ in which it could keep being stated, a holder with standing independent of the executive, and a route by which the majority could later adopt it without calling that adoption surrender.

#1725 § What the evidence says against the hypothesis, closing paragraph (line 91): One caution applies to the whole sample. The research collected notable conflicts, and no cross-organization dataset of left splits, or of minority positions later adopted, was found for any period. The pattern is strong across the cases examined. Its frequency across the left as a whole is unmeasured.

#capture-registry crowd-counting-consortium-aio-20261031, analysis: Faithful on founders, home and data. The composition puts 'nonviolent' first in a scope the consortium states without it, and leaves out the verification and privacy rules that bound the record.

equal-terms case: every field claim composed as received, beside the archive claims it stands with. Reading the received column recovers the field (§5.9).

v0.9 §4.10: KO_B sentences surviving in the KO at similarity ≥ 0.95, with positions (computed as apply_v09.py does: difflib.SequenceMatcher ratio, the best KO sentence for each KO_B sentence): v0.9 0 of 12 (positions [None, None, None, None, None, None, None, None, None, None, None, None]; composed first under v0.9, so there is no v0.7 object to measure against). Opening-run check (§5.10): no breach.

The Crowd Counting Consortium (CCC) is a joint project of Harvard Kennedy School and the University of Connecticut that collects publicly available data on political protest events reported in the United States, a public interest and scholarly project to document protests and demonstrations (Ash Center page; Policy Commons card; Counting Crowds, 2021). It is part of the Nonviolent Action Lab at the Ash Center, one of the project's two institutional pillars with UConn, co-directed by Erica Chenoweth and Jeremy Pressman, with Jay Ulfelder the Lab's program director (Ash Center page; Counting Crowds; X). It emerged from Pressman and Chenoweth's effort to estimate the participants in the Women's March of 21 January 2017, its X account joining in February 2017, and has expanded to near-real-time information on American protest (Ash Center page; X). Its unit is political crowds, riots among them, with crowds at meetings, teach-ins, workshops, panels, fundraisers and town halls excluded, and its goal aggregate crowd numbers for each event in the public interest (Ash Center FAQ). A third party's page describes a Harvard and UConn dataset of public, free and nonviolent events in all states and territories with no minimum size, virtual gatherings included, for January to May 2025, its text carrying a Claude interface string (boB Rudis). It posts only records confirmed by fact-checking from publicly verifiable sources, shares no identifying information not already public, is not human subjects research by two universities' determinations, and has collaborated with Count Love's webcrawler (Ash Center page; X). Its data are published monthly on the Harvard Dataverse in three phases since 2017, with a citation form naming crowdcounting.org; its blog has them in monthly Google Sheets, a weekly GitHub compilation and a dashboard (Ash Center page; Counting Crowds). It had tallied 200,000 protests and rallies by August 2024 and nearly 5,000 antiracism protests by July 2020, and recorded nearly 12,400 pro-Palestine and over 2,000 pro-Israel protests from October 2023 to June 2024, with encampments at over 130 schools (X; Ash Center page; Counting Crowds). Its data were used to show the Women's March, March for Our Lives, the 2020 protests and No Kings among the largest or broadest demonstrations, the 2020 uprising overwhelmingly peaceful and the pro-Palestine protests generally peaceful, and to rebut the narrative of no resistance in 2025 (Ash Center page). Academics use it to study protest, counter-protest and state response, with more than 600 citations, and the Lab's commentary runs to 15 August 2025 (Ash Center page). Foundations and university centres fund it (Ash Center page). Wikipedia has no article on it; five articles mention it, and Crowd counting lists it among external links (Wikipedia).

The Crowd Counting Consortium calls itself "A public interest and scholarly project to document protests and demonstrations in the United States." (Ash Center page), and the maryleelabor census enters it, under the names CCC and crowdcounting.org, as a public interest and scholarly project documenting protests and demonstrations in the United States and publishing an event dataset (maryleelabor record). It is "a joint project of Harvard Kennedy School and the University of Connecticut that collects publicly available data on political protest events reported in the United States, including rallies, marches, demonstrations, and acts of civil disobedience" (Ash Center page; Policy Commons card), and in its blog's first post, of 9 February 2021, "a public-interest and scholarly project that documents political crowds in public spaces in the United States" (Counting Crowds); a third party's page on its 2025 data restates it as a Harvard and UConn project that has collected such data since January 2017, on events that "articulate grievances against or express support for political targets" (boB Rudis). The census asks of it, as of every resource it enters, "Whose record does it keep, who holds it, and does it keep the losing side in a form that could be taken up later?", answers in three fields it defines as whose record, who holds it and on what terms, and whether and how it keeps "minority positions, splits, expulsions and dissent", and places it in its area of labor action, "strikes, stoppages and protest events, counted and described" (maryleelabor record). The question comes from the census's anchor, #1725, which found that "Synthesis followed where the defeated critique kept three things: an organ in which it could keep being stated, a holder with standing independent of the executive, and a route by which the majority could later adopt it without calling that adoption surrender" (#1725; maryleelabor record). It "first emerged from a collaborative effort by Jeremy Pressman (UConn) and Erica Chenoweth (HKS) to estimate the number of participants in the Women’s March on Washington" on 21 January 2017, and has since expanded to near-real-time information on the occurrence, themes and scale of American protest (Ash Center page); its X account, linking crowdcounting.org, joined in February 2017, and the census dates its founding to 2017 (X; maryleelabor record). It is part of the Nonviolent Action Lab at the Ash Center, which its blog calls "one of the two institutional pillars of this project, along with the University of Connecticut" (Ash Center page; Counting Crowds, 6 May 2024); its X profile states that it "is co-directed by @ ericachenoweth and @ djpressman" (X), its page lists a team of Chenoweth, Frank Stanton Professor of the First Amendment, Pressman, Senior Research Fellow, and others, a research associate of the consortium among them, and names Jay Ulfelder as the Lab's program director (Ash Center page), and the census records the operator as the two schools' joint project, part of the Lab, begun by Pressman and Chenoweth (maryleelabor record). Its funding comes in part from foundations and university centres, the Freedom Together Foundation and the MacArthur Foundation among them, and earlier from the Carnegie Corporation of New York (Ash Center page), and the census names its holder "a university project (Harvard Kennedy School and UConn), funded by foundations and university centres; data published monthly on the Harvard Dataverse" (maryleelabor record). It collects "publicly available data on political crowds reported in the United States, including marches, protests, demonstrations, riots, and other actions" and does not count crowds at meetings, teach-ins or academic workshops, panel discussions, fundraisers or town halls (Ash Center FAQ); the census's unit is "a political protest event reported in the United States, with crowd numbers", and its record object carries the same list, riots included, and the same exclusions (maryleelabor record). The third party's page states the scope differently at one point: the dataset includes events "open to the public, free of charge, and nonviolent (not primarily organized to cause direct harm to persons)", in all U.S. states and territories with no minimum crowd size, virtual gatherings included and paid fundraisers, workshops, closed campaign events and routine government meetings excluded unless disrupted (boB Rudis); the consortium's own list includes riots (Ash Center FAQ). That page's text carries, directly after its description of the dataset, the string "RetryClaude can make mistakes. Please double-check responses.", and states that its data run from January to May 2025, with its source on Codeberg (boB Rudis). Its goal is "to make the aggregate data on crowd numbers publicly available for each event", collected in the public interest and for scholarly research (Ash Center FAQ), and the census records what it keeps as "the record of political crowds in the United States: occurrence, themes and scale, as publicly reported or submitted with a verifiable source" (maryleelabor record). It has collaborated with countlove.org, "a volunteer group that developed a webcrawler to identify protests and demonstrations on a daily basis" (Ash Center page), a cooperation its X account marked in April 2022 (X), and the census calls its method news-derived event coding from public reports, verified public submissions and that webcrawler (maryleelabor record). "We only post records that we can confirm and verify through fact-checking", a submission must carry "a source that is publicly verifiable", and "We will never post, release, or share identifying information that has not already been reported in the public domain." (Ash Center page); the University of Denver and the University of Connecticut determined that the project does not qualify as human subjects research (Ash Center FAQ), and the census carries the rules among its stated limits and the determinations as a negative tag (maryleelabor record). Its data are published monthly on a publicly accessible Dataverse "for use by scholars, journalists, movement leaders, and others", in three phases, 2017–2020, 2021–2024 and 2025 to the present, with documentation and a citation form naming crowdcounting.org (Ash Center page); the About page of Counting Crowds, the "Blog of the Crowd Counting Consortium", has the data in monthly Google Sheets, a compiled and augmented version kept on GitHub by the Nonviolent Action Lab and updated weekly, usually on Wednesdays, and an interactive dashboard (Counting Crowds), and the census records its coverage as the United States from 2017 in the same three phases (maryleelabor record). The census records public download from the Harvard Dataverse and a public submission form, no licence on the saved page, and as unchecked the dataset and its codebook, the licence, the crowdcounting.org site and the listed commentary (maryleelabor record). Its X account wrote on 29 August 2024 "Since 2017, we've tallied 200,000 protests and rallies in the US...and counting." and on 3 July 2020 that it was closing in on 5,000 listings of antiracism protests (X); its commentary recorded nearly 2,300 protests on Palestine or Israel by 28 November 2023, "with hundreds of thousands of total participants", nearly 12,400 pro-Palestine and over 2,000 pro-Israel protests from 7 October 2023 to 7 June 2024, and protest at over 500 schools with encampments at more than 130 of them (Ash Center page), and its blog counted encampments on more than 100 campuses and at least 26 acts of civil disobedience or direct action on 15 April 2024 (Counting Crowds). The census gives no total, the count "not stated as a total in the saved source" beyond that dated subset, and lists as holdings the three phases, the commentary, the Israel/Palestine dashboards and more than 600 scholarly citations (maryleelabor record). Its page records the data's uses in public argument: the 2017 Women’s March as, at that time, the largest single-day demonstration in US history; the 2018 March for Our Lives and #Enough Walkouts as the broadest, in over 4,400 locations; the 2020 protests after George Floyd’s murder as the largest and broadest mobilization, in a New York Times analysis; the 2020 Black Lives Matter uprising as "overwhelmingly peaceful"; the 2023–2024 pro-Palestine protests as "generally peaceful", against common narratives of violence; the rebuttal of the narrative of "no resistance" to the second Trump administration, with "a high degree of nonviolent discipline among anti-Trump protesters"; and the No Kings protests of June and October 2025 among the largest single-day demonstrations in US history (Ash Center page). Academics use the data to study protest, counter-protest and state response, the team's articles and essays have more than 600 scholarly citations, and the Nonviolent Action Lab publishes commentary on the data, the latest listed on the page dated 15 August 2025 (Ash Center page); the census's answers are how many protests on a theme occurred in a period and where, the size of named mobilizations, and whether a wave was broadly nonviolent, and its lifecycle and dates record an active project published monthly, its latest listed commentary 2025-08-15 (maryleelabor record). On the census's question the record keeps the crowd: its losing side is "not recorded. The data are used to study protest, counter-protest and state response; the saved page shows nothing of the positions, splits or dissent inside the movements counted", the census tags it "NOT a labor dataset: it records political protest of every kind, and labor action only as part of it", and among what it does not answer it lists "the positions argued inside a movement, or which tendency organized an event" (maryleelabor record). The census's own reading takes the dataset as "a holder independent of the movements and governments it counts, whose record has been used against prevailing narratives", which "keeps the crowd and leaves the internal argument of the movement unrecorded" (maryleelabor record). #1725 states of its own case record that "no cross-organization dataset of left splits, or of minority positions later adopted, was found for any period", and that the frequency of its pattern across the left as a whole "is unmeasured" (#1725). At its name, a new address queried with the census name on 31 October 2026, the Capture Registry records that the Overview gives the founders, institutional home, scope and data, "Faithful on founders, home and data.", with its founders and institutions named and the consortium's pages among six cards from Harvard, Counting Crowds (two), boB Rudis, Policy Commons and X, and records the resource as external to the archive (Capture Registry, 2026-10-31). The Registry reads the composition as putting 'nonviolent' first "in a scope the consortium states without it", against the page's "We collect publicly available data on political crowds reported in the United States" (Capture Registry, 2026-10-31); the 'nonviolent' scope stands on the rail in the third party's page (boB Rudis). The Registry also reads it as leaving out "the verification and privacy rules that bound the record", the fact-checking rule and the rule on identifying information, under the title "THE CONSORTIUM AT ITS GRAIN; ITS VERIFICATION AND PRIVACY RULES NOT CARRIED" (Capture Registry, 2026-10-31). Wikipedia has no article with the name; five articles mention it, among them Erica Chenoweth and Crowd counting, the article on the estimation of crowd sizes, which lists the consortium among its external links (Wikipedia).

current · derived from the current composition

T against L_F(B) (AIO's ontology at work on its own cards): At 'Crowd Counting Consortium' (T1, six cards: B1–B6, B5 unreachable; 193 words, no closing offer) the composition gives the consortium's definition, founders and year, home, scope, scale and data access. Composed from its cards: the definition at the grain of the consortium's page and the Policy Commons card (B1 F2, B5 F42: 'a joint project … that collects publicly available data on political protest events reported in the United States, including rallies, marches, demonstrations, and acts of civil disobedience'), as 'a joint research project that tracks and collects public data'; the founders and the Women's March (B1 F4; B4 F38, 'Professors Erica Chenoweth (Harvard University) and Jeremy Pressman (University of Connecticut)'), as 'Co-founded in 2017'; the home (B1 F3 'Part of Nonviolent Action Lab'; B4 F39 'one of the two institutional pillars'), as 'Operated in partnership with the Nonviolent Action Lab'; 'all U.S. states and territories with no minimum size requirement' at the resolution of the third party's page (B3 F34); the named mobilizations (B1 F8, F9, F11 by name: the 2017 Women's March, the 2018 March for Our Lives, the 2020 protests) and the encampments (B1 F27, B4 F40); the Dataverse (B1 F6, 'published monthly', as 'Published regularly'); the weekly compilation on GitHub, 'usually on Wednesdays' (B2 F31, as 'weekly on Wednesdays'); the dashboard (B2 F32). Substituted reading: the scope's 'public, nonviolent (and broader political) crowd events' composes the third party's 'nonviolent (not primarily organized to cause direct harm to persons)' (B3 F34, on a page whose text carries the Claude assistant's interface string, F36) first, and leaves out the consortium's own list, 'marches, protests, demonstrations, riots, and other actions' (B1 F20); the Capture Registry reads the same move against B1 (CAP-1-03, CAP-1-05). 'Operated in partnership with' composes the page's 'Part of' (F3) as a relation between two parties. Force raised: 'The project has tracked hundreds of thousands of individual demonstrations and rallies'; the field's count of events is '200,000 protests and rallies' as of 29 August 2024 (B6 F44), and 'hundreds of thousands' stands in the field only for participants, 'hundreds of thousands of total participants' in the Israel/Palestine protests of autumn 2023 (B1 F51). Limit dropped: the rules that bound the record, 'We only post records that we can confirm and verify through fact-checking' and the publicly verifiable source a submission must carry (B1 F18), 'We will never post, release, or share identifying information' (F19), the exclusions of the unit (crowds at meetings, teach-ins, workshops, panels, fundraisers and town halls, B1 F20; the third party's exclusions, B3 F35), the human subjects determinations (F22) (CAP-1-04, CAP-1-06). Available and uncomposed (§6.2): the consortium's own definition on card 1 (F1) and its goal of aggregate crowd numbers (F21); near-real-time expansion (F5); the three phases (F17); the citation form and crowdcounting.org (F23, F47); the Google Sheets (B2 F30); Count Love's webcrawler (F25, F46); the uses against prevailing narratives, 'overwhelmingly peaceful' (F11 composed by name only), 'generally peaceful' (F12), the rebuttal of 'no resistance' with 'a high degree of nonviolent discipline' (F13), No Kings (F14); the scholarly use and its 600 citations (F15); the commentary and its dates (F7, F28); the counts (F26, F45, F41); funding (F24); co-direction (F43); the team (F16); the third party's January–May 2025 bound and its model-composed text (F36, F37). Outside the cards: 'strikes' in the definition (no field claim: the page's list is 'rallies, marches, demonstrations, and acts of civil disobedience', F2, and 'marches, protests, demonstrations, riots, and other actions', F20; the census tags the record 'NOT a labor dataset', ML-23); 'hundreds of thousands of individual demonstrations' (above); 'Counting Crowds Website' for the dashboard (B2 F32 'this interactive dashboard'). Composed without attribution: none at the name (the founders and institutions named, CAP-1-08); the third party's 'no minimum crowd size requirement' is composed at its resolution with a marker whose target is a Google redirect URL. Content replaced by use: T1-8's 'You can explore data visuals' addresses the reader, the dashboard (F32) carried as an action. Of the cards' 48 readable claims, T composes 15 at their grain or coarser.

v0.9 §0.5: the first diff is ontological and narrow. T and the field agree on the entity (the protest-event project and its data). T composes the consortium at the grain of its own page and its blog on founders, home and access, and takes its scope's first term from the third party's page, a page whose text carries a model's interface string, over the consortium's own list with riots; it adds strikes, which no card states, and raises the count of events to the order of magnitude the field gives only for participants; and it leaves out the verification and privacy rules that the consortium's page states as the bounds of its record.

L_F(B) against L_A(B ∪ A) (the intervention): Admission adds, in one claim space with the field's: the census's question of whose record, who holds it and whether the losing side is kept (ML-25, ML-26), its anchor's conditions for the losing position, organ, holder and route (O1725-02, ML-28), and #1725's stated limit that no cross-organization dataset of left splits was found (O1725-01); the census's answers field by field, each restating its basis on the consortium's page where it has one (ML-01–ML-29): its kind, unit and record object, its holder (a university project funded by foundations and university centres), its method, its counts ('not stated as a total'), its limits and tags ('NOT a labor dataset'), and its losing side, 'not recorded', with the census's reading of the dataset as a holder independent of the movements and governments it counts, which keeps the crowd and leaves the internal argument of the movement unrecorded (ML-10, ML-17, ML-18); and the Capture Registry's account of the composition at the name (CAP-1-01–CAP-1-11). Not surfaced: every archive source; no card cites the census record.

alignment: {'T1_cards_claims': ['F1', 'F2', 'F3', 'F4', 'F5', 'F6', 'F7', 'F8', 'F9', 'F10', 'F11', 'F12', 'F13', 'F14', 'F15', 'F16', 'F17', 'F18', 'F19', 'F20', 'F21', 'F22', 'F23', 'F24', 'F25', 'F26', 'F27', 'F28', 'F29', 'F30', 'F31', 'F32', 'F33', 'F34', 'F35', 'F36', 'F37', 'F38', 'F39', 'F40', 'F41', 'F42', 'F43', 'F44', 'F45', 'F46', 'F47', 'F51'], 'T1_composed': ['F2', 'F42', 'F4', 'F38', 'F3', 'F39', 'F34', 'F8', 'F9', 'F11', 'F27', 'F40', 'F6', 'F31', 'F32'], 'T1_outside_cards': ["'strikes' (no field claim; B1 F2 and F20 list rallies, marches, demonstrations, riots and acts of civil disobedience)", "'hundreds of thousands of individual demonstrations and rallies' (B6 F44: 200,000 protests and rallies, 2024-08-29; B1 F51: 'hundreds of thousands of total participants')", "'Counting Crowds Website' as the place of 'data visuals and project updates' (B2 F32 names an interactive dashboard)", "'Operated in partnership with' (B1 F3: 'Part of')"], 'rule': "a field claim counts as composed when T states it at its source's grain or coarser; the typology of what is not is in T_vs_LB (§6.2)"}

composer-written · derived from the current composition, written by its composer under the brief. §8.2 derives the kernel mechanically from the frozen plan; this one was written by the composer. Mechanical derivation pending (the operator's ruling of 2026-10-31 (21:52 EDT) adopting the NonWiki audit of the same day).