Consolidation run — 2026-07-27

Fourth consolidation run (OPS-015; workflow v0d.24/25). Run branch: claude/consolidate-2026-07-27, off master (5ff1778). Worksheets: consolidation-pregate-2026-07-27 · claim-retirement-queue-2026-07-27.

Run scope (fixed at C0)

Register at C0: 303 entries — 23 supported, 143 live, 130 candidate, 2 contested, 5 retired. normalize_claims_table.py --check: in sync at C0. Seat config per the environment register (2026-07-26): all subagent seats Opus 5 — adjudicators (claim-promotion-reviewer, claim-refuter) at high effort, citation-verifier at medium, Fable main-thread only.

  • C1 (subtraction): nothing to apply. Retirement queue at the continuity window (--staleness-days 60): 109 rows, all keep/staleness, 0 concrete-defect rows (downgrade/retire/merge) — fourth consecutive clean queue at register 303. keep/review rows route into batch selection.
  • C2 (candidate triage): 30 oldest candidates (batch cursor = Updated:, per the run-1 rule). Following the run-1/2/3 discharge pattern: the 11 multi-source candidates get the two-agent gate (3-test claim-promotion-reviewer + rival-reading claim-refuter, peer-isolated); the 19 single-source candidates discharge mechanically as KEEP-with-condition — promotion is impossible under the single-source rule and the concrete-defect scan returned 0 rows.
  • C3 (supported jury): first 5 of 85 eligible, oldest-first — ambivalence-as-unconscious-replacement-predates-wahl-1951 · broad-psychoanalysis-as-named-camp-1949-sorbonne · spontaneous-structuration-as-institution-precursor · bukharin-as-mp-exemplar-phenomenological-political-ethics · culture-as-mediating-milieu-webbed-causality. Full jury per claim: citation-verifier on every Evidence anchor, independent 5-test review, unanimity default-deny refuter panel (evidence-chain / rival-reading / scope). Cap ≤5 promotions; no same-run substitution for blocks. (Run 2's and run 3's juried claims are correctly excluded by the pregate's 14-day settle window.)
  • C4: skipped (C2+C3 consume the run's edit budget, run-2/3 precedent). Speculate-survivor backlog recount stays an ops-ledger job (NEXT.md OPS-015); the ~82-row harvest reservoir stays queued.
  • Parked ledger: read at C0 — empty; nothing to unpark.
  • Carry-forwards assessed this run (queued from runs 2–3 in NEXT.md OPS-015): the Malraux Voices pagination re-derivation pass (mechanical locator legs only) and the pbp-as-pivot revision-set sizing — outcome recorded below.
  • Adjudication workflow: wf_0c146f6b-f92 (consolidate-c2-c3-adjudication, 47 agents: 11×2 C2 + 5×5 C3), peer-isolated, launched before any register write.

Run mechanics

First fully clean adjudication run across runs 2–4: wf_0c146f6b-f92 completed 47/47 seats, 0 dropped, 0 errored, in one launch (17.8 min, ~4.15M subagent tokens); verdicts taken from the workflow's returned result; no salvage, no rerun, no schema false starts. gen_adjudications.py ran on the clean-run provenance path (PR #114 parameterization) and its REFUTED⇒objection / STANDS⇒tried payload validation passed 26/26.

Adjudication outcomes (applied same-day)

Jury: 47/47, decisive, 0 promotions. Verdicts: consolidation-2026-07-27-verdicts.json; per-claim apply-doc: consolidation-2026-07-27-adjudications.md. All 26 refuters returned REFUTED (11 C2 rival-reading + 15 C3 across three lenses) — the fourth consecutive all-REFUTED panel (100/100 across runs 1–4). C2 reviewers: 4 PROMOTE (all refuter-blocked), 3 HOLD, 4 REQUEST-REVISION. C3: no claim had a clean verifier sheet (1/2/0, 1/0/2, 1/2/0, 2/1/0, 2/2/0 P/U/F) and no 5-test said PROMOTE (2 REQUEST-REVISION, 3 HOLD) — every C3 claim was blocked on at least two independent legs before the refuter panel voted. The PR #114 demotion discriminator fired once, on ambivalence-as-unconscious-replacement-predates-wahl-1951 (the 5-test sheet's explicit contested recommendation).

Absence-claim re-verification (main-thread, before apply). 225 absence assertions collected across the 47 seats; tiered re-verification (178 scripted needle re-greps — 75 absences hold mechanically, 103 hits filtered to 64 and eyeballed — plus targeted raw reads on every objection-load-bearing anchor). Catches, both directions: (a) the ambivalence verifier's claim that the E&C II raw is absent from raw/ is a false absence — raw/datalab-output-Etre et Chair II.md is present (the seat's own method note recorded its Glob tool failing); its UNCERTAIN stands on the unattested page-span only. (b) The broad-psychoanalysis verifier's "Mexico-III co-naming in MP's own hand at Inédits II raw 4177" is an over-read — the psychanalyse des choses attributions live in the editorial apparatus (notes 1711–1712: Sartre's formula; Ponge, not Baudelaire); the Counterpressure addition was applied in corrected form. (c) Conflicting seat-level folio readings for the CPP volume were adjudicated by the volume's own index: the raw {N} markers are PDF pages at +16 from printed folios (Gestaltists 150; ambivalence-as-substitute 73, 124; contrasted-with-ambiguity 242, 286) — the 5-test's +16 diagnosis was right and both refuters' mechanisms were wrong in opposite directions.

Status changes

One status change (every other entry stays at its pre-run status, with objections and repairs applied). Standing veto covers every write below.

Slug Change Rationale Veto
claims#ambivalence-as-unconscious-replacement-predates-wahl-1951 live → contested (confidence medium → low) The priority thesis ("first lexical use / earliest named technical usage") is refuted from inside the ingested corpus, main-thread re-verified in raw: SB 1942 Ch III raw 1686 carries the substitution doctrine verbatim in the Politzer-anchored Freud critique seven years earlier; Inédits II raw 6577 carries the Politzer/second-récit/« – l'ambivalence – » complex (1947–49) while the Inédits II extraction note (lines 184/484) quotes the neighboring phrases and silently drops the lexeme — the 2026-05-05 live promotion's negative check ran against that elision (Rule-18 extraction-adequacy finding); and CPP itself teaches the Klein ambivalence/ambiguïté distinction pre-Wahl (raw 3397 ≈ printed 242, raw 3891 ≈ 286, indexed), undercutting the two-sense periodization. 5-test reviewer explicitly recommended contested/low; refuters 3× REFUTED. Corrective restatement (the survivable continuity thesis) filed to the Harvest reservoir; Supersedes: is the route back. veto: say "veto ambivalence-as-unconscious-replacement-predates-wahl-1951" — restores live + Confidence: medium (pre-run state)
  • C2 gated (11): all stay candidate — every rival-reading refuter REFUTED under default-deny (all four reviewer-PROMOTEs refuter-blocked). Each objection appended verbatim to Counterpressure; prose Status History line with failed legs + an explicit would-promote condition; reviewer-named mechanical repairs applied in the same edit; Updated: 2026-07-27.
  • C2 single-source (19): mechanical KEEP-with-condition discharge — promotion impossible under the single-source rule; each Status History now names its specific second-source promotion path (incl. the two SPENT-re-test entries, whose condition is a new source entering the corpus, not another search); Updated: bumped.
  • C3 blocked, stays live (4): broad-psychoanalysis-as-named-camp…, spontaneous-structuration…, bukharin-as-mp-exemplar…, culture-as-mediating-milieu… — verifier UNCERTAIN/FAIL legs + 5-test non-PROMOTE + 3× REFUTED each; 12 objections appended; verifier-named repairs applied; confidence unchanged (bukharin and culture keep medium per the reviewers' explicit recommendations).
  • Parked ledger: nothing parked this run; ledger remains empty.

Rule 16 findings (prominent — citation-fabrication-pattern class)

  1. broad-psychoanalysis-as-named-camp… — the "Freud-Lacan" false conjunct (the run's decisive Rule-16 finding). The 1954-55 terminus "invocation of Freud-Lacan as institution-resources" is raw-falsified: zero Lacan occurrences in the I&P raw, source page, and extraction note (main-thread re-verified) against 207 Freud hits. Struck from the entry (Title, Claim, Evidence, Payoff) and from the propagated surfaces (institution ×2; the CPP source page ×3); the CPP extraction note's line-224 copy is outside write-scope, routed to the next audit. Bullet 2 replaced with the anchored I&P Freud-side bullet; bullet 3 restated as the anchored Inédits negative result.
  2. ambivalence… cluster — the extraction-note lexeme elision (see Status changes), a paraphrase-in-quotes in the cardinal bullet (restored to raw-verbatim), an unattested "p. 61-85" span (narrowed to the attested p. 64), and a circular concept-page Evidence bullet (deleted).
  3. ion-enthousiasmos… — untraceable quotation in the Claim. The Phaedrus 245a "sane… is not complete" wording exists in no ingested text; replaced with the verified Cooper wording (raw 14478) in the entry and on poetic-inspiration.
  4. bukharin… — a quote-marked hedge attested nowhere in the HT raw (de-quoted as maintainer paraphrase); the AdV-self-revision sentence annotated unanchored-as-stated (AD has zero trial-defense Bukharin engagement); the Sartre Anti-Semite and Jew payoff parallel tagged untraceable (extraction-note self-recycle).
  5. sophist… — Evidence attributed the GW 21 locus to the wrong Hegel volume (1816 Begriff → 1832 Sein; Sources synced); a Counterpressure bullet refuted by the raw was rewritten (GW 21 raw 922 generalizes the ends-in-nothing charge — which cuts for the claim); the Hegel-not-innocent-of-the-Sophist apparatus finding added (GW 11 raw 7133 → Werke Bd 14 on Sophistes 259c–e).
  6. khora… — Evidence bullet 2 cited motifs.md (not a permitted anchor type) and bundled a khôra-empty Kaushik attribution; re-anchored to the Carbone extraction note with its counter-valence recorded (Poimandres closer than Timaeus; "not 'third'"); the Derrida counterpressure noted unanchored in-corpus.
  7. ungrund… — both Evidence bullets were bare topic pointers (distinctAnchoredSources = 0); anchored (Schelling raw 561 disapplication
    • note 72; Hegel GW 21 raw 1801) with the refuter's valence caution.
  8. parole-masque… — the L2472 bracketed substitution inverted Goldstein's two terms; corrected against the PbP raw.
  9. Smaller repairs: nihilism p.68/p.69 locator split + circular sub-anchor re-pointed + "/§4" leg dropped + cap-sentence retired; culture 4084→4086 off-by-two; spontaneous-structuration p. 167→150 (the CPP +16 finding) + two to-do bullets anchored + Title date 1949-50→1949-52; callicles/statesman/self-is-soul locator additions and the unanchored "built to answer" rider softened; the dead claim-link [[claims#nihilism-as-inverse-of-recognition-institution]] corrected in the sibling entry.

Malraux Voices locator re-derivation (run-2 parked carry-forward, executed)

The queued pass ran as planned: per-locator re-derivation against the Princeton/Bollingen 1978 raw's own apparatus (TOC part-starts, the Synopsis's per-topic printed pages, the index), never blanket offsets. The wiki's cites were three mixed conventions (block-as-page, block+1, and an alien pagination for Part III/IV). ~45 locators corrected across andre-malraux, metamorphosis-art, two-historicities, silence, art-as-revolt-against-fate, aftermath-of-the-absolute, museum-without-walls, coherent-deformation, the source page, and two claim entries — headline conversions: metamorphosis-law 72→68, dialogue 73→69, severance 69→65, schema 374→339, sieve/style 376→348, equivalences 392→355, style-freedom 411→359, gramophone/destiny 662→630, revolt-against-fate 671→639, crowning-victories 672→640, trembling-hand 673→642, Hercules/museum 666→635, Cassis 293→280, influences/chrysalis 247→171, lonely-smile 240→171, lines-of-silence 218→155. Includes the supported claim claims#malraux-schema-precedes-mp-coherent-deformation-philological-refinement (mechanical locator repair only; absence core untouched — veto: say "veto malraux-schema locators" to restore the former numbers). The terminology row's "voix du silence" attestation list was withdrawn (the conjoined phrase occurs only in the raw's front/back matter — in the French raw's front matter too); one silence.md quote flagged unlocatable as worded. Signs-side syncs: the run-2 p. 66→60 correction completed at its two missed sites; the "transforming efforts into 'works'" quotation inflection + p. 64 locator fixed (Signs raw 904). The pbp-as-pivot revision set (interpretive) stays queued.

Caps consumed / backlog remaining

C1 0/15 (0 concrete-defect rows, fourth consecutive clean queue) · C2 30/30 triaged (11 gated + 19 single-source) · C3 5/5 juried, 0/5 promotions (cap unconsumed; 1 reviewer-recommended demotion live→contested) · C4 skipped. Register after run: 303 entries — 23 supported, 142 live, 130 candidate, 3 contested, 5 retired. Remaining backlog: 10 candidates untriaged by C2 batches (130 − 30×4), 80 supported-eligible live claims (85 − 5), ~54 speculate survivors (recount still queued to ops-ledger), the ~82-row harvest reservoir + 1 new reservoir item (the ambivalence corrective restatement). New audit-routed items from this run: the corpus-wide CPP marker-vs-folio (+16) locator sweep; the extraction-note copies of the Freud-Lacan conjunct, the 4084 off-by-two, and the Inédits II lexeme elision (all outside consolidation write-scope). The register is not yet dry: next trigger is another /consolidate run (run 5).

Files touched

wiki/claims.md (37 entries: 11 gated + 19 single-source + 5 C3 + the 2 Malraux locator entries + the sibling dead-link fix; 26 verbatim refuter objections into Counterpressure; ~30 reviewer/verifier-named citation repairs; 1 restatus) · label/citation sync on 15 pages (the 9 Malraux-sweep pages + poetic-inspiration + institution + the CPP source page + ambiguity-vs-ambivalence + broad-vs-narrow-psychoanalysis + spontaneous-structuration) · derived surfaces (claims Summary table; index Claims section + counts + page tables) · baselines (claim-status fixture 143 live/2 contested → 142/3; overview stats; graph SHA; the Paper-A dashboard_contested_claim_slugs pin + test_paper_a_a2.py, hand-patched — patcher-list carry-forward still queued) · run artifacts (verdicts JSON, adjudications MD, this digest, pregate + queue worksheets) · wiki/log.md · NEXT.md (OPS-015 progress). Test suite: 590 passed / 10 skipped; lint baseline OK at 906 pages / 303 entries.

Errata (post-run, PR #115 review)

Codex's PR #115 review returned four P2 findings, all incomplete propagation; all four are accepted and repaired in a follow-up commit on the same branch. Standing veto applies to each — say "veto PR-115 errata N".

  1. Locator sweep left half-converted pages (errata 1–2). The Malraux pass above re-derived correctly but propagated partially, so single passages carried two folios. Completed to every site: equivalences 392→355 (5× on coherent-deformation); gramophone/inner-voice 662→630; Hercules 666→635; crowning-victories/torch 672→640; closing image 673→642; revolt-thesis 671→639; dialogue 73→69; chrysalis 247→171; Cassis 293→280; lonely-smile 240→171; lines-of-silence 218→155; schema 374→339; sieve/style 376→348; style-freedom 411→359; §VII range 661–673→630–642. Sites: art-as-revolt-against-fate, andre-malraux, metamorphosis-art, silence, aftermath-of-the-absolute, coherent-deformation, the source page, and claims#art-as-absolute-replaces-religion-malraux-thesis.

  2. One in-run conversion corrected. The beast-within / "rediscover Man" quotation was written to p. 640 in the run; it is printed 642. §VII's marker-to-folio offset is a uniform −31, pinned four ways against the volume's Synopsis ("The inner voice, 630" / "Man as revealed in the museum, 635" / "Art, a revolt against man's fate, 639" / "The noblest legacy…, 640"), and the quotation sits under raw marker {673}. The Synopsis's own "'To rediscover Man…,' 641" line indexes where the topic opens (marker {672}), not where the sentence lands — a per-locator distinction that a blanket read of the Synopsis would have gotten wrong.

  3. Un-re-derived residue made explicit rather than guessed. pp. 14, 16, 53, 70, 71, 130, 272, 386 and the annexation pair 117, 396 remain in the raw-{N} convention inside broader "etc." attestation lists. A dated locator-convention note now names them on the source page and routes them to the next audit Citations phase with the CPP +16 sweep. No blanket offset: the drift is non-uniform through the volume. deformation-coherente-mp-coinage's 2026-04-29 Status History entry keeps its superseded numbers as a dated record, with a 2026-07-27 erratum line appended beneath it.

  4. Cite-back sync completed on both restatused claims (errata 3–4). ambiguity-vs-ambivalence still labelled the demoted claim (live) at two sites and kept the priority framing; both now expose contested/low + the pre-Wahl counterevidence, with the surviving weaker point (a distinct Politzer-derived sense, not a first one) stated in place of the refuted superlative. The raw-falsified "Freud-Lacan as institution-resources" conjunct was struck at its two surviving live sites — the broad-vs-narrow-psychoanalysis summary paragraph and the entry's own Counterpressure bullet 3. Refuter-objection bullets keep the phrase because they quote it in order to refute it; the extraction-note copies stay routed to audit.

Process note for run 5. Both defect classes are propagation, not judgement: a verified conversion or restatus that lands on the register but not on every dependent surface. The run's own cite-back step greps by claim slug, which catches [[claims#slug]] references but not locator strings or prose paraphrases of a struck conjunct. Worth adding a post-apply grep on every changed locator number and on the distinctive lexeme of any struck phrase.

Consolidation — 2026-07-27, run 5 (second run of the date)

Same-day deviation (user-authorized). Run 4 (PR #115, merged 26b7a02) consumed the 2026-07-27 date key earlier this day. The maintainer authorized a second same-day run ("Run 5 now, same-day") under the skill's same-day mechanics: this section APPENDS to the run-4 digest (run 4's content above is untouched); the pregate and retirement-queue worksheets were regenerated in place (--force; run-4 copies preserved in git history at 26b7a02); the run branch is claude/consolidate-2026-07-27-run5; run-5 verdict artifacts carry a -run5 filename suffix so run 4's are not clobbered; every run-5 Status History tail carries a run 5 marker plus the run's workflow id (wf_a0b34b6a-bc5). The C4 batch was likewise user-selected ("Speculate-signoff batch"). The run crossed midnight into 2026-07-28 mid-execution and keeps its start-date key throughout, per the run-3 precedent. Veto protocol unchanged: say "veto <slug>" — reverts to the pre-run state.

Run scope (fixed at C0)

  • C1: retirement queue → 0 concrete-defect rows (fifth consecutive clean queue). The register's least-recently-touched active entry sat at 89d, one day under the 90d staleness default — the keep / review heuristic starts firing naturally from the next run; no --staleness-days override (parity with runs 1–4). (Corrected 2026-07-28 — see the Erratum at the end of this digest: the parity clause was mistaken — run 4's worksheet had used a 60d override — and the 90d-default regeneration emptied it; regenerated at 60d, 103 keep/review rows, still 0 concrete-defect rows.)
  • C2: the 10 never-triaged candidates (the pregate batch rows at 29–30d) — 1 gated (multi-source) through the two-seat gate; 9 single-source mechanical KEEPs. This exhausts the never-triaged candidate backlog.
  • C3: the 5 oldest supported-eligible live claims (all Updated: 2026-05-05): h-and-t-political-articulation-of-mps-open-ontology, no-causality-between-psych-and-soc-as-stiftung-precondition, institution-as-middle-term-1953-55, institution-as-paradigm-shift-from-stiftung-to-instituting-subject, mp-1955-three-coordinated-mutations. Cap 5/5 consumed.
  • C4: speculate-signoff batch 3 (2026-07-27-batch3, first consolidation-C4-embedded batch): 6 survivors from the 21 remaining v1.2-pilot pool (motif ×3, subtext ×1, aporia ×1, genealogy ×1; the two Knight-2024-dependent survivors deliberately skipped). No convergent finding; juror-C not spawned.
  • Seats: C2/C3 via workflow wf_a0b34b6a-bc5 — 27 peer-isolated Opus 5 seats (adjudicators high, verifier medium, per .claude/environment-register.md), 27/27 clean, second consecutive fully clean adjudication workflow; C4 via the speculate-promote-jury team (lead = session top-tier, jurors + interp Opus 5), 6 agents, phases 0–7, two debate rounds.

C1 — Subtraction

No rows to apply. 0/15 cap consumed.

C2 — Candidate triage (10/10 discharged; 0 promotions)

Single-source mechanical KEEPs (9) — per the run-2/run-4 precedent, each with its entry's own recorded promotion condition restated in Status History and Updated: bumped; none consumed agent seats: algorithm-as-special-case- of-language-1950, painting-language-asymmetry-designed-to-fail, objective-drawing-as-one-expressive-operation-among-others, i-speak-as-methodical-subjectivist-foundation, universality-of-feeling-grounds-first-speech-pre-ontologically, descartes-as-institution-reflexively-applies-expression-thesis-to-philosophy (the six PdM/PW-cluster entries — conditions are the named missing second sources: 1953–54 courses, "Eye and Mind", PbP 1960, PhP/VI mechanism passages), gm-genealogy-is-herkunft-not-purpose (Husserl-side Stiftung anchor via weave/motifs-delta), gm-ascetic-priest-fuses-ressentiment-and- bad-conscience (second source on fuses-vs-sequences, or the What-the-Concept-Does route), bge-philosopher-as-legislator-closes-will-to- truth-to-power (second source on §211's hinge function).

Gated (1): gm-bad-conscience-is-internalized-will-to-power (2 sources, same author — GM 1887 + BGE 1886) — blocked, stays candidate: reviewer REQUEST-REVISION (test 2 CONCERNS — a propagated off-by-one locator: the pregnancy sentence at raw 2876 is GM II.19, not II.18, main-thread re-verified by block-counting and the Chouraqui GM-II-19 corroboration; test 3's strongest objection, the two-souls rival, was unwritten), refuter REFUTED (R1 synchronic-vs-diachronic: BGE §19's plural soul is constitutive of all willing and cannot receive GM II.16's internalization as its genealogy; R2 same-force-different-material deflation of the "third register"). Applied: II.18→II.19 per-locator corrections in the entry, propagated to bad-conscience (blockquote, Key Points, Open Questions, Sources) and — caught by the new post-apply grep, beyond the reviewer's named list — the GM source page (2 grouped-attribution sites split per-locator). The extraction-note copies (l. 76 + the 2b header l. 22) are outside consolidation write-scope — routed to the next audit Citations phase. Would promote: Counterpressure answering R1/R2 + the Chouraqui Ch. 2 third-source anchor (loc 740, 984–1110) verified for the will-to-power-turned-inward reading (the reviewer's unchecked lead).

C3 — Supported jury (5 juried, 0 promoted, 0 demoted)

All five blocked; every entry stays live at medium, all repairs applied in-run, all 15 refuter objections + reviewer would-promote paths recorded on the entries. Per claim (verifier P/U/F; 5-test; outcome):

  1. h-and-t-political-articulation-of-mps-open-ontology — 3/2/0; HOLD; 3x REFUTED. Both UNCERTAINs were the cross-source bullets: the PhP pointer named a 'horizon' apparatus absent from the cited note (0 hits — Rule-18 extraction-adequacy gap, not fabrication) and the V&I bullet had no locator. The 5-test's homonymy objection (three "opennesses" sharing one adjective; the entry's own Counterpressure concedes the parity thesis its Claim denies) is now recorded. Applied: both bullets re-anchored (PhP raw p. 96 l. 1864 + the note's theme-weight motif; the May-1960 "Metaphysics—Infinity / World—Offenheit" note, V&I p. 251, raw 2566–2575), the "unfinished" half struck as a manuscript-state false friend (main-thread NUL-tolerant re-verification: 1 occurrence in the primary raw, editorial — python is the tool of record for that NUL-containing file; BSD grep under-reports it), the garbled clause restated. Would promote: the two-register restatement (interpretive, queued); an AD 1955 bullet discharging the 2026-05-05 precondition; the homonymy objection answered.
  2. no-causality-between-psych-and-soc-as-stiftung-precondition — 4/1/0; HOLD (confidence stays medium per the reviewer — defect gate-shaped, not confidence-shaped); 3x REFUTED (SB-1942 priority rival; constitution-not-causation rival; reduction-vs-causation scope equivocation). Applied: the Claim's Comte/methodological-individualism attribution replaced with the Freudians MP names (raw 164/3249; Comte = 2 occurrences in the raw, "methodological individualism" = 0 — main-thread re-verified); bullet 1 repointed (the former anchors named the decentration item and a nonexistent Pass 2b entry); bullet 3 softened to MP's own raw- 4646 hedge; bullet 4 anchored to the I&P hinge + endogenous-causality passages with the not-same-predicate caution; bullet 5 reclassified as a coordination note; raw 4646 added to Counterpressure as author-attested. Would promote: the genealogical rivals engaged; the Crisis/Heidegger prerequisite discharged; the wording narrowed or defended; the webbed-causality twin adjudicated.
  3. institution-as-middle-term-1953-55 — 6/0/1 FAIL — the run's Rule-16 finding (see below); 5-test HOLD; 3x REFUTED (Pagan's "as is well known"/"it could be argued" hedges; the coherent-deformation rival mediator; the Weber/Lukács first-relatum gap). Applied: the Halák bullet rewritten as two correctly-attributed legs with the Endstiftung-divergence caution and the unanchored "wild Being / flesh of life" bridge dropped; the Claim's reverse-order splice corrected (I&P 59 impersonal + I&P 62); missing locators added (raw 1606/1812); bullets 6–7 upgraded to direct primary anchors; the "novel" mischaracterization corrected; wild-being dropped from Wiki homes (no cite-back, no anchoring bullet). Would promote: non-M-C corroboration (Larison 2023) or the softened "a load-bearing middle term" restatement; the mutual-warrant loop with the PPH claim adjudicated.
  4. institution-as-paradigm-shift-from-stiftung-to-instituting-subject — 4/1/0 (the UNCERTAIN a composite paraphrase in quotation marks — the extraction note's own Pass-2a sentence presented as M-C's); 5-test HOLD; 3x REFUTED (León's "more than a development" is additive, not exclusive; the Brunschvicg/Sartre-target rival; the from-term equivocation — every anchor excludes constitution, not Stiftung). Applied: bullet 3 de-quoted with M-C's real raw-1600/1606 sentences; Lefort attribution; I&P-8 quote tail restored; three stale Payoff labels live→supported (Summary-table-verified); propagation on [[constituting-vs-instituting- subject]] (composite de-quoted; §-labels; Caraus Ch 2→Ch 12), institution (near-antagonism re-attributed from Stiftung to constitution per I&P 8's own text), interdependence-claim (Halák→Popa). Would promote: the from-term restatement (queued); a Husserl-continuity Counterpressure bullet; non-M-C corroboration + the synthesis-reading reconciliation.
  5. mp-1955-three-coordinated-mutations — 2/2/0; 5-test REQUEST-REVISION (the reviewer found the unrecorded I&P raw-2332 primary anchor); 3x REFUTED (SA's own word is "coïncident"; Alquié Fontaine is 1947, not "1948–49"; Counterpressure bullet 2's "not in raw/" was false — the Passivity course is in raw/ and attests mutation (i) directly). Applied: all four reviewer items — bullet 4 anchored to raw 2332/2282/2292 with the fn-27 two-series folio caution (118, not the extraction note's drifted 218); datings corrected in Claim + bullet 3 with home-bullet sync on institution and indirect-ontology; the false Counterpressure statement rewritten to the true limit (the archive-bound residue is the coordination thesis + the 1958–61 litany); leg (iii) restated per SA raw 635 — plus the graded-hedges fix and the unanchored Lefort-as-1957-dater annotation. Would promote: the "coordinated" modality operationalized or softened; non-SA corroboration; the "dominates" quantifier anchored or dropped.

Refuter streak: 16/16 REFUTED this run (runs 1–5 cumulative: 116/116 under default-deny unanimity). The pattern across five runs is stable: reviewers rate the cores sound at live; supported bids fail on cross-source anchor quality, self-set undischarged preconditions, and Counterpressure that concedes rather than defeats.

C4 — Speculate-signoff batch 3 (2026-07-27-batch3)

First consolidation-C4-embedded batch; first zero-promotion batch (bottom-tier selection from a twice-harvested pool — both deliberately placed calibration tests caught real defects). Verdicts: 6/6 unanimous — 3 HOLD-AS-SPECULATIVE (erwachen, realisation, passence) + 3 ROUTE-TO-METHODOLOGICAL (schritt-zurueck → MQ-5, two-registers-collapse → MQ-6, hyppolite → MQ-7); 0 promotions, 0 escalations (the parked ledger stays empty), 0 lead overrides. Four converged at round 1, two at round 2; 14/24 seats moved, every move attributed to a named finding; the round cap was never reached. Phase 6 (main thread): six SPECULATIONS.md jury-verdict annotations written; no claims.md or concept-page writes — the batch's one near-promotion (schritt-zurueck ⟷ anti-survol, 3-1 promote at round 1) died on E's find that SA's own body text glosses MP's relation to the Identität und Differenz movement as "parfaitement inverse" (the cousin-pair's alignment conjunct counter-attested) plus the doubled instrument objection (the has cross-tradition cousin relation is weave-reserved, and MP annotated the parallel himself at OntoCart 167–168) — it routes as MQ-5's open-direction philological check with a false-friend caution. MQ-8 records the batch-wide anchor-stripping signal (every survivor anchored to sources, not passages — the MQ-3 family's third firing; codification bar met). Process findings recorded in the cohort fingerprint for future batches: the speculate baseline's exclusion of all 72 candidates blinded its restatement filter to this batch's three sharpest adjacencies (R); two merge-heuristic blind spots with worked examples (re-axing one source; disjoint home-sets over a shared anchor text); G's proposed Test-1 alignment-statement pre-check for cousin-pair survivors; S's alignment-not-availability probe refinement + the clean-probe-is-not- promotion-support rider; interp's proposed fourth map axis (repair-variants). Open liability, self-flagged by juror E: the PdM 142 integration/accumulation check underlying the passence axis finding was never run against raw — carried in the batch record for whichever batch revisits the retyped repair. Execution notes: the jury made no git operations (this C5 commit carries the batch dir); team-agent completion notifications routed to the session main thread rather than the lead throughout — the main thread relayed each phase transition (a harness-routing quirk worth knowing for future C4 batches, not a protocol defect; peer isolation was never breached by the relays).

Status changes (veto surface)

Slug Change Rationale Veto
(none — 0 promotions, 0 demotions, 0 retirements across C1–C4)

15 register entries were touched with repairs/KEEPs/objections only (no status moved); 6 SPECULATIONS.md annotations and 4 methodological-queue entries (MQ-5..MQ-8) were written by C4. Standing veto applies to every one of these writes: say "veto <slug>" (register entry — reverts the repair/KEEP line), "veto batch3-<survivor-slug>" (removes the annotation), or "veto MQ-N" (strikes the queue entry); each is a single revert on this run's one commit.

Rule-16 finding (prominent, per the escalation triage)

The institution-as-middle-term-1953-55 Halák bullet carried a misattributed quotation: quotation marks around "between vital, collective, and transindividual domains," attributed to Halák Ch 5 §2.1, when the phrase is Mendoza-Canales's (Ch 4, raw 1610, reading "traverses… of experience"). Adjudicated repair-class, not halt-class: the volume is real and ingested, the phrase occurs in it, the provenance chain is identifiable (the extraction note's own paraphrase re-attributed at register-writing time), and run 4 handled the same class (the Kee pseudo-quotations) as repairs. The paradigm-shift entry's composite-in-quotes (the extraction note's Pass-2a paraphrase presented as M-C's sentence, propagated to one concept page) is the same class, same treatment. Both repaired in-run with dated markers; both named here for the standing veto. The halt-class pattern (invented citation to an unread work) did not occur.

Caps consumed / backlog remaining

C1 0/15 · C2 10 discharged (1 gated blocked + 9 single-source KEEPs; the 20 re-enumerable run-1 KEEPs untouched) · C3 5/5 juried, 0/5 promotions · C4 one batch, 6 survivors (cap ceiling). Register after run: 303 entries — 23 supported, 142 live, 130 candidate, 3 contested, 5 retired (no status moved this run). Remaining backlog: 0 never-triaged candidates (the C2 discharge is complete — every candidate now carries a KEEP-with-condition or a block-with-objection), 75 supported-eligible live claims (80 − 5 juried; the five juried re-enter the pre-gate after the 14d settle window, repaired), ~15 remaining v1.2-pilot survivors (21 − 6; recount still queued to ops-ledger), the ~82-row harvest reservoir. New audit-routed items from this run: the GM extraction-note II.18→II.19 copies (l. 76 + l. 22); the CPP/PhP/ V&I extraction-note gaps the juries named (the 'horizon' vocabulary absence; the I&P 218 folio drift); the h-and-t out-of-scope locator drift (motifs-delta-2026-05-05 + the v1.2-pilot motif draft). The register is not dry on the C3 side: next trigger is another /consolidate run (run 6).

Files touched

wiki/claims.md (15 entries: 1 C2-gated + 9 single-source KEEPs + 5 C3; 16 verbatim refuter objections into Counterpressure; ~30 verifier/reviewer-named repairs; 0 status changes) · label/citation propagation on 8 pages (bad-conscience, the GM source page, [[constituting-vs-instituting- subject]], institution, interdependence-claim, indirect-ontology, plus the institution/indirect-ontology mp-1955 home-bullet syncs) · C4 batch dir wiki/.speculative/promotion/runs/2026-07-27-batch3/ (MANIFEST, baseline, interpretation map, 24 initial + 24 final verdict files, 6 debate transcripts, ADJUDICATIONS.md, cohort-fingerprint.md) · promotion-ledger batch-3 triage entry · methodological-queue MQ-5..MQ-8 · 6 SPECULATIONS.md annotations · run artifacts (run5-verdicts JSON with clean-run provenance, run5-adjudications MD, regenerated pregate + retirement-queue worksheets) · derived surfaces (claims Summary table; index Claims section + page-table cells) · baselines (home_audit_count 145→146; atlas contract event counts + legacySemanticHash; two graph SHAs; test_paper_a_a3.py audit-count pins) · wiki/log.md · NEXT.md (OPS-015). Test suite: 590 passed / 10 skipped. Digest generator note: gen_adjudications.py derives its provenance header from the verdicts JSON's provenance field — regenerating before that field is set produces the salvage-narrative header (hit and corrected this run); a verdicts-filename parameterization for suffixed runs is queued scaffolding.

Erratum (2026-07-28) — C0 retirement-queue regeneration used the unscaled default

Caught by the Codex review of PR #116 (P2). The C0 same-day regeneration of claim-retirement-queue-2026-07-27.md ran with the 90d staleness default, against the scaled-window mandate (CLAUDE.md §Subtraction Infrastructure young-register note; consolidate SKILL C0 step 4) and against the artifact's own printed young-register warning. The scope bullet's "parity with runs 1–4" clause was mistaken: run 4's worksheet had used a 60d override (109 keep/review rows). With the least-recently-touched active entry at 89d, the 90d default produced a 0-row artifact, overwriting run 4's worksheet in place and leaving C1 and this digest to report an empty queue as if it were information.

Repair (2026-07-28, on the run branch, Codex-review fixup): regenerated from the C0-time register state (master 26b7a02, detached worktree) with --date 2026-07-27 --staleness-days 60 --force → 103 rows, all keep/review, 0 concrete-defect rows. The C1 "nothing to apply" conclusion and every run-5 outcome stand unchanged. The 6 rows present in run 4's 109-row worksheet but absent here are entries whose last-touch was legitimately reset by run 4's own C1–C3 edits and the PR #115 fixup. Standing reminder the queue file itself prints: a 0-row queue on a young register is a threshold artifact, not a subtraction-health certificate.