docs(taxonomy): #498 AIF chantier — coverage-status synthesis for jsboige ratification - #707
Merged
Merged
Conversation
…oige ratification Synthesis note (read-only, 0 prod write) consolidating the #498 chantier after 4 PR, to support jsboige ratification of (a) the méthode and (b) the I/RA/CA CSV schema decision. Content: - §1 Coverage code=truth on master 70bd160: 70/1408 (5.0%) per family. Flags a data-quality gap (48 rows with empty Family — out of chantier scope, separate hygiene item). - §2 Chantier progress table: 4 PR, 14 leaves, 70->81 fully-modeled (+11 DoD). Fallacious comparison complete (3/3 sub-subs), Inexact definition opened (1/3). - §2b FAIL-LOUD accounting: 4 leaves across 2 layers (CA-missing PR-1 840, RA-missing PR-2 834/835/837), all honest gaps documented in AIF_skosOther, 0 fabrication. - §3 Méthode recap (cluster selection, 2 patterns exception/direct-conflict, native-vocab discipline #677, fail-loud rules). - §4 DoD enrichi jsboige (I/RA/CA): recorded in each PR §7 but NOT YET SERIALIZED. Decision needed: (a) ratify 2 new CSV columns AIF_attackType + AIF_attackedNode (reco — serialize early avoids late back-fill), (b) defer, (c) alternative schema. - §5 Trajectory: ~87 sub-subs, recommend focus on Misleading language (12.6%) + Faulty logics (10.8%) = best anchor/leaves ratio. Next worker-able targets listed. Derived code=truth from CSV scan + chantier PRs (#699/#701 merged, #703/#705 OPEN). No token fabricated in the synthesis. Co-Authored-By: Claude-Code <noreply@anthropic.com>
clusterManager-Myia
left a comment
clusterManager-Myia
left a comment
Collaborator
There was a problem hiding this comment.
[NanoClaw] — LGTM (COMMENT only, décision merge + §4 schema = Emerjesse/jsboige)
Docs synthèse coverage-status byte-vérifié firsthand au head 783e6b9 (+176/-0, 1 NEW docs/taxonomy/498-coverage-status.md, 0 prod write confirmé par file-list = add-only docs).
§1 Coverage code=truth PROUVÉ : les 2 anchors chiffrés les plus concrets vérifiés par parse quoted-field-aware (le CSV a 102 cols + cellules HTML multi-lignes, le naive split(\n) over-compte 1544/299 — il FAUT un parser quoted-field) :
- 1408 data rows = EXACT firsthand (Fallacies Taxonomy CSV 4 070 984 bytes, parser quoted-field).
- 48 rows empty-Family = EXACT firsthand (même parser,
Familycol idx 20). - 70/1408 = 4,97% ≈ 5,0% ✓, arithmetic sound.
§2 Chantier table byte-accurate : PR states tous confirmés firsthand via pulls/N — #699 merged=true ✓, #701 merged=true ✓, #703 merged=false OPEN ✓, #705 merged=false OPEN ✓. Cumulative 70→81 (+11) : leaves 4+2+2+3=11 ✓. FAIL-LOUD accounting honnête (CA-missing 840 PR-1, RA-missing 834/835/837 PR-2, 2 layers distinctes, DirectRef pattern + AIF_skosOther doc, 0 token fabricated).
§4 Decision framing sound (je n'opine pas sur le choix) : les 3 options (a 2 new columns AIF_attackType+AIF_attackedNode / b defer / c single JSON column) sont clairement énoncées avec reco (a) + raison (serialize early avoids back-fill, structured data for EPITA harness). Décision = jsboige. La reco (a) est défendable mais c'est un appel produit, pas un check review.
§5 Trajectory : ~87 sub-subs / ~30-40 PR = estimate planning sound (basé sur leaves/PR empirique du chantier 4-PR). Reco focus Misleading language (12,6%) + Faulty logics (10,8%) = best anchor/leaves ratio — cohérent avec §1 coverage data.
Convention respectée : read-only synthesis (0 Cards/ touch, 0 CSV modifié, 0 OWL regen), dispatch mkzt2v idle-de-secours, framing « 0 fabrication » tenable (tous chiffres dérivés CSV scan master 70bd1605 + chantier PRs). △ Caveat : les per-PR leaves/DoD counts (4/2/2/3, fully-modeled +4/+2/+2/+3) sont empruntés au body sans re-vérifier chaque PR-§7, mais la row-denominator + empty-Family + PR-states = 4 anchors indépendants exacts = méthodologie code=truth validée. Sound, honnête, utile pour la ratification jsboige.
jsboige
added a commit
that referenced
this pull request
Jul 5, 2026
…t, 7 leaves) (#708) Sixth cluster of the #498 AIF chantier. Models the Inconsistent definition sub-sub (pk 826-832) of the Inexact definition subfamily, borrowing the d2 root pk 799 (exception pattern: VerbalClassification_Inference defeated by BiasedClassification_Conflict + ArbitraryVerbalClassification_Inference). Largest and most varied cluster of the chantier so far — semantically heterogeneous (contradiction / stolen-concept / circularity / lexical rigidity / recursive vagueness / vague meta-fallacy), so the leaves split across three patterns rather than forcing one: - Exception pattern (826 VerbalClassification + Logical_Conflict, 830 VerbalClassification + ArbitraryVerbalClassification) - Direct-conflict pattern (827 Logical_Conflict, 828 InconsistentCommitment, 831 VagueVerbalClassification) - FAIL-LOUD (829 circularity CA-missing like PR-1 840, 832 vague meta-fallacy RA-missing like PR-2 834/835/837) 7 leaves proposed: 5 fully-modeled + 2 FAIL-LOUD (documented in AIF_skosOther, 0 fabrication). Cross-cluster scheme reuse: 831 reuses PR-4's VagueVerbalClassification_Inference. §7 I/RA/CA decomposition recorded but not serialized (pending jsboige schema decision #707 §4). Borrow-root shape = independent of that decision. PROPOSITION ONLY — GATED, 0 prod CSV write, no DB write, no OWL regen. Native vocabulary validated programmatically (all backticked scheme/conflict tokens native; Circularity_Conflict appears only in the "do not fabricate" prohibition, never as a proposal). Relates: dispatch mkzt2v (tertiary), #498, PR-1 #699, PR-2 #701, PR-3 #703, PR-4 #705, coverage-status #707, #133/#130 (OWL), #499, #677 (0 fabrication), #192, #458. Co-authored-by: Claude-Code <noreply@anthropic.com>
This was referenced Jul 5, 2026
docs(taxonomy): data-quality gap inventory — 48 rows empty Family (i18n propagation, gated fix)
#712
Merged
jsboige
added a commit
that referenced
this pull request
Jul 5, 2026
…k of 5 clusters (ratification-ready) (#709) Independent read-only audit of po-2023's 5 #498 cluster propositions (PR-1 #699 False analogy, PR-2 #701 Faulty comparison, PR-3 #703 Association fallacy, PR-4 #705 Vague definition, PR-6 #708 Inconsistent definition), triggered by ai-01 dispatch `awhj8g` (PRIMARY) to de-risk jsboige's ratification before any prod CSV write. Method: a deterministic stdlib script (498_audit_extract.py, read-only, scratchpad) extracts the CSV columns for every cited pk (25 pks: 799, 800-803, 826-832, 833-845) and builds the native AIF vocabulary inventory (every *_Inference/*_Conflict token in any AIF column). Verdicts computed by crossing the docs' claims against this extraction — no LLM judgment on the data, only on the semantic legitimacy of FAIL-LOUD cases. Findings — the 5 propositions are code=truth-clean: - 25/25 cited pks present; cluster boundaries match Soussousfamille/Subsubfamily_en for all 5 clusters. - 4/4 mapped anchors (799, 800, 833, 839) match byte-exact on text_fr + DirectRef + ExceptionRef + MappingType. - 10/10 proposed tokens native (inventory of 54 tokens) → 0 fabrication (#677 discipline upheld). - 21/21 cited desc_fr match byte-close (1 cosmetic apostrophe diff on 841). - 6/6 FAIL-LOUD cases legitimate (Comparison_Inference + circularity node genuinely absent from inventory). Two FAIL-LOUD layers verified: CA-missing (840, 829 circularity) + RA-missing (832, 834/835/837). 1 FLAG raised (not a rubber-stamp) — coverage-accounting consistency: - PR-1 counts 840 (CA-missing: RA exists, CA absent) in its "+4 mapped", but PR-6 excludes 829 (same CA-missing shape) from its "+5 fully-modeled". The chantier headline "86 fully-modeled" depends on which PR's convention applies; uniform strict criterion → 85. Recommend jsboige ratify a single criterion. Does not block ratification of the propositions themselves. Secondary (ratification-ready note) integrated as §7 of the audit: verified native-token inventory + the two FAIL-LOUD layers, for jsboige's schema decision (I/RA/CA serialization, per #707 §4). Read-only: 0 write under Cards/, audit doc only. Adversarial (1 FLAG, not rubber-stamp). Dispatch awhj8g (PRIMARY + SECONDARY integrated). Base 34c7702. Co-authored-by: Your <your.email@example.com> Co-authored-by: Claude-Code <noreply@anthropic.com>
jsboige
added a commit
that referenced
this pull request
Jul 5, 2026
#711) Seventh cluster of the #498 AIF chantier. Opens the Ambiguïté subfamily (Abus de language / Misleading language family). Models the Amphibologie sub-sub (pk 847 anchor + 848-854), borrowing the d2 subfamily root pk 846 (direct-conflict pattern: ArbitraryVerbalClassification_Inference + OppositeConsequences_Conflict + SignFromOtherEvents_Conflict). Structurally coherent cluster — all leaves are syntactic/structural/prosodic ambiguity mechanisms (sentence-level), distinct from the Equivoque sub-sub (term-level). This coherence drives an honest scheme-divergence from the borrow-root: amphiboly is syntactic, the borrow-root's classification framing is term-level, so leaves split across native schemes that honestly fit their mechanism rather than forcing the borrow-root's scheme onto syntactic defects: - Exception pattern (849 modal scope, 850 quantifier shift): Deductive_Inference + Logical_Conflict (genuine logical-scope defects, first chantier use of Deductive_Inference for a fallacy) - Direct-conflict pattern (851 accent, 852 illicit contrast): ArbitraryVerbalClassification_Inference (borrowed from 846 — prosody is term-level, classification framing fits) - Direct-conflict + cross-cluster reuse (854 barbarism): VagueVerbalClassification_Inference (PR-4 scheme — barbarism is vagueness not ambiguity; the desc word 'incertain' signals it) - FAIL-LOUD (847 amphiboly anchor, 848 punctuation, 853 solecism): structural/syntactic ambiguity — THIRD recurring AIF ontology gap after circularity (PR-1/PR-6); no native syntactic-ambiguity CQ, documented in AIF_skosOther, 0 fabrication 8 rows proposed: 5 fully-modeled + 3 FAIL-LOUD (most FAIL-LOUD in a single cluster). §7 I/RA/CA decomposition: all typed leaves undercuts (consistent jsboige). Decomposition recorded but not serialized (pending jsboige schema decision #707 §4). Borrow-root shape = independent of that decision. PROPOSITION ONLY — GATED, 0 prod CSV write, no DB write, no OWL regen. Native vocabulary validated programmatically (all backticked scheme/conflict tokens native; Ambiguity_Conflict and Amphiboly_Conflict appear only in the 'do not fabricate' prohibition L105, never as proposals). Relates: dispatch ih617l (primary), #498, PR-1 #699, PR-2 #701, PR-3 #703, PR-4 #705, PR-6 #708, coverage-status #707, verification-audit #709, #133/#130 (OWL), #499, #677 (0 fabrication), #192, #458. Co-authored-by: Claude-Code <noreply@anthropic.com>
jsboige
added a commit
that referenced
this pull request
Jul 5, 2026
…8n propagation, gated fix) (#712) Inventory + gated fix proposition for the data-quality gap flagged by #707 §1: 48 rows in Cards/Fallacies/Argumentum Fallacies - Taxonomy.csv with empty Family (EN) column. Triggered by ai-01 dispatch awhj8g idle-de-secours ("Pioche une autre vérif code=truth freeze-safe (ex : le data-quality gap '48 rows Family vide' — inventorie-les, propose un fix gated)"). Finding (code=truth on 34c7702, deterministic read-only extractor): - The 48 rows are NOT orphans, NOT headers, NOT ambiguous. They are a pure i18n-propagation gap: each has Soussousfamille (FR) filled but Family (EN) + Subfamily (EN) empty. - 100% mechanical fix: each of the 48 rows' Soussousfamille_fr maps to exactly one canonical (Family_en, Subfamily_en) pair (100% consensus across filled rows, zero conflict). - 6 distinct FR subfamilies in the gap: Jeu de pouvoir (33 rows -> Influence/Psychological manipulation), Conditionnement (6 -> Influence/Psychological manipulation), Mensonge (4 -> Cheating/Spin doctoring), Langage persuasif (2 -> Influence/Rhetorical device), Ambiguïté narrative (2 -> Misleading language/Ambiguity), Repoussoir (1 -> Influence/Appeal to emotion). - Also clarifies the 56 Subfamily-empty count = 48 gap + 8 family-header rows (pk 0 + 7 family roots, where Subfamily_en is legitimately empty) — no additional gap beyond the 48. Gated fix proposition (NOT executed, release freeze on Cards/): one-liner lookup-table fill of Family_en + Subfamily_en for the 48 rows, same byte-targeted apply discipline as #654. Validation gates: re-derive mapping on current master (abort if conflict), row count unchanged, BOM+CRLF byte-preserved, no Soussousfamille_fr drift. Post-tag CSV-hygiene window. Why not during freeze: Cards/ frozen pre-tag; gap is pre-existing (3.4% of rows), does not block generation (pipeline uses FR Soussousfamille for hierarchy). Low value during freeze. Read-only: 0 write Cards/, 0 build, 0 test. Reproducible via family_gap_extract.py (scratchpad). Dispatch awhj8g (idle-de-secours). Base 34c7702. Co-authored-by: Your <your.email@example.com> Co-authored-by: Claude-Code <noreply@anthropic.com>
jsboige
added a commit
that referenced
this pull request
Jul 5, 2026
…in-sub-sub anchor, 4 leaves) (#713) Eighth cluster of the #498 AIF chantier. Second cluster of the Ambiguïté subfamily (opened by PR-7 Amphibologie). Models the insinuation mechanism of the Ambiguïté narrative sub-sub: in-sub-sub anchor pk 876 (ConflictingGoals_ Conflict, direct-conflict) + 4 leaves (pk 877-880). In-sub-sub anchor shape (cleanest, PR-1/PR-4-like). The sub-sub has 10 leaves spanning 3 mechanisms (insinuation / propaganda / interrogation), so PR-8 scopes to the insinuation mechanism (4 leaves); propaganda + interrogation deferred to PR-9 (different AIF schemes). All 4 leaves are say-without-saying variants — the narrative serves conflicting communicative goals (benign surface + deniable insinuated payload), the anchor 876 mechanism. Two patterns: - Direct-conflict reusing anchor token (877 false implication, 878 innuendo): ConflictingGoals_Conflict (borrowed from 876), narrow/closeMatch - Direct-conflict with semantic-closer native conflict (879 poisoned compliment): OpposedCommitment_Conflict (praise/discredit opposition beats conflicting-goals) - Genuine exception pattern with distinct scheme (880 anonymous sourcing): PositionToKnow_Inference + LackOfCompleteKnowledge_Conflict (epistemic defect, anonymity hides whether source is in position to know) 4 leaves, 4 fully-modeled, 0 FAIL-LOUD (cleanest since PR-3). First chantier use of OpposedCommitment_Conflict and LackOfCompleteKnowledge_Conflict (both native, 2x/1x in existing usage). Palette now 17 distinct native tokens. §6 accounting transparency (re po-2024 audit #709): this PR's +4 is criterion-independent (all leaves have both scheme AND conflict). Cumulative reported under both strict (94) and inclusive (95) criteria — the 1-row discrepancy is the PR-1 baseline (pk 840, RA-exists/CA-missing), pending jsboige criterion decision. §7 I/RA/CA: all 4 undercuts (insinuation defeats the audience's inference, not the surface statement). RA-nodes untyped for 877/878/879 (Gricean implicature — no native Implicature_Inference scheme, 4th ontology gap shape, documented not fabricated). Decomposition recorded, not serialized (in-sub-sub anchor = independent of I/RA/CA schema decision #707 §4). PROPOSITION ONLY — GATED, 0 prod CSV write, no DB write, no OWL regen. Native vocabulary validated programmatically (Implicature_Inference appears only as named gap L281/L283/L303, never as a proposal). Relates: dispatch ih617l (tertiary), #498, PR-1 #699, PR-2 #701, PR-3 #703, PR-4 #705, PR-6 #708, PR-7 #711, coverage-status #707, verification-audit #709, family-gap #712, #133/#130 (OWL), #499, #677 (0 fabrication), #192, #458. Co-authored-by: Claude-Code <noreply@anthropic.com>
jsboige
added a commit
that referenced
this pull request
Jul 5, 2026
…leaves, completes subfamily) (#714) Ninth cluster of the #498 AIF chantier. Completes the Ambiguïté subfamily (3/3 sub-subs). Second half of the Ambiguïté narrative sub-sub: the deferred mechanisms from PR-8 — source-deception (881-883), interrogation (884-885), and polytélie (886). In-sub-sub anchor pk 876 (ConflictingGoals_Conflict). Source-deception and interrogation leaves honestly diverge from the anchor's ConflictingGoals_Conflict to epistemic/action schemes that fit their semantics: - Source-deception (881 grey, 883 black): ExpertOpinion_Inference + ExpertiseInconsistency_Conflict. Omission (concealment, narrow) vs commission (falsification, close) distinguished by MappingType — graded source-deception. - Whisper campaign (882): PositionToKnow_Inference + LackOfCompleteKnowledge_ Conflict (cross-PR-8 reuse of 880 epistemic pair, organized-campaign flavor). - Clandestine interrogation (884): PracticalReasoning_Inference + ConflictingGoals_Conflict (anchor reuse — friendly surface + extraction goal). - Preach the false (885): Bias_Inference direct-conflict (scheme as defect — evidence biased by false-premise collection, degenerate undercut like PR-4 800). - Polytélie (886): ConflictingGoals_Conflict direct reuse (literal multiple goals). 6 leaves, 6 fully-modeled, 0 FAIL-LOUD (2nd consecutive clean cluster). 4 new chantier native tokens: ExpertOpinion_Inference, ExpertiseInconsistency_Conflict, PracticalReasoning_Inference, Bias_Inference (all native 1x/1x/2x/2x). Palette now 21 distinct native tokens across 9 PR. §6 completes Ambiguïté subfamily 3/3 (18/18 leaves). Second fully-complete subfamily after Fallacious comparison. Cumulative 70→100 strict / 101 inclusive (+30). §6 dual criterion re po-2024 audit #709 (this PR's +6 criterion-independent). §7 I/RA/CA: all 6 undercuts (deception defeats the inference audience/target is manipulated to draw). Decomposition recorded, not serialized (in-sub-sub anchor = independent of I/RA/CA schema decision #707 §4). PROPOSITION ONLY — GATED, 0 prod CSV write, no DB write, no OWL regen. Native vocabulary validated programmatically — PASS, 0 fabrication. Relates: dispatch ih617l (tertiary), #498, PR-1 #699, PR-2 #701, PR-3 #703, PR-4 #705, PR-6 #708, PR-7 #711, PR-8 #713, coverage-status #707, verification-audit #709, family-gap #712, #133/#130 (OWL), #499, #677 (0 fabrication), #192, #458. Co-authored-by: Claude-Code <noreply@anthropic.com>
This was referenced Jul 6, 2026
Merged
jsboige
added a commit
that referenced
this pull request
Jul 7, 2026
…e / Causalités cluster (proposition, gated) (#741) Pilot for serializing the ratified #707 §4(a) schema (AIF_attackType + AIF_attackedNode) on the Virtues side, mirroring the po-2023 Fallacies chantier format (#498 PR-1 #699). Dispatch ai-01 ompwhx [primaire] #499 Virtues AIF. One cluster modeled: Raisonnement valide (family 4) / Causalités bien identifiées (subfamily 4.1) — anchor pk 80 + leaves 81/82/83. Chosen as the cleanest inverse-paradigm mirror: the Fallacies subfamily Causalité douteuse sits at path 4.1.x (698/707/719), the Virtues subfamily sits at path 4.1 — same path, opposite tenor (structural corroboration of jsboige's inverse paradigm, independent of the crossLink_Opposes mapping). The Virtues 12-col relational layer is already generated + validated CLEAN (222/222 nodes, 9 PRs merged June 2026, validator #518). This PR does NOT redo it; it extends it with the I/RA/CA decomposition the ratified schema asks for, reusing the existing scheme/CQ/opposed-PK grounding (no re-derivation). Inverse-paradigm adaptation FLAGGED for jsboige validation (not decided alone): the ratified columns are attack-shaped, but a Virtue is the good holding of a scheme, not an attack. Three options documented (A attack-resisted [reco], B structurally-empty, C AIF-native support). Pilot written under Option A; only the last two table columns change if jsboige picks B or C. 0 fabrication (undercut + RA-node are native AIF). Native vocab only (discipline #677). 0 prod CSV write, 0 DB, 0 OWL regen, 0 Cards/. Proposition doc only. Co-authored-by: Your <your.email@example.com> Co-authored-by: Claude-Code <noreply@anthropic.com>
jsboige
added a commit
that referenced
this pull request
Jul 7, 2026
…Fidélité aux faits cluster (proposition, gated) (#742) Second cluster of the Virtues I/RA/CA chantier, advancing the ratified #707 §4(a) schema (AIF_attackType + AIF_attackedNode) on the Virtues side. Dispatch ai-01 ompwhx idle-de-secours mandate "Continue #499 cluster par cluster". One cluster modeled: Honnêteté intellectuelle (family 6) / Fidélité aux faits (subfamily 6.1) — anchor pk 153 + leaves 154/155/156. Path-matched mirror: Fallacies Arranger les faits (Tricherie) at path 6.1.x (889/942/953), Virtues Fidélité aux faits at path 6.1. Second independent path-prefix corroboration of the inverse paradigm (PR-1 was 4.1 Causalités). Point of this PR = generalization evidence: PR-1 was 4/4 undercut/RA-node. This cluster shows a MIXED distribution (1 undermine/I-node + 2 undercut/RA-node), derived per-case from the fallacy mechanism (falsehood→undermine the premise; source-distortion/bias→undercut the inference). Proves the columns are derived, not templated. 1 honest nuance (942 multi-variant) classified by primary mechanism, edge flagged in AIF_skosOther, 0 fabrication. Written under Option A (same FLAG as PR-1, same one-line adjustability). I/RA/CA decomposition is Option-independent. Reuses the validated 12-col layer (scheme/CQ/opposed-PK from 499-scaleup-phase2-honnetete annotations). Native AIF vocab only (undermine/undercut/I-node/RA-node/Witness Testimony/Bias). 0 prod CSV write, 0 DB, 0 OWL regen, 0 Cards/. Proposition doc only. Co-authored-by: Your <your.email@example.com> Co-authored-by: Claude-Code <noreply@anthropic.com>
jsboige
added a commit
that referenced
this pull request
Jul 7, 2026
… rebut-rarity finding (#743) Extends Virtues AIF I/RA/CA decomposition (schema #707§4(a)) with 3rd cluster, covering family 7 Échange enrichissant / subfamily 7.3 Respect de la personne (3 leaves pk 208/212/216, scheme: Argument from Commitment). Mirror Fallacies: Obstruction 7.x primary (1398 ad hominem, 1371 genetic, 1352 poisoning-the-well) + cross-family (322 repoussoir/Influence, 942 fausse attribution/Tricherie). Path-prefix 7.3↔7.3 PARTIAL (4/6 Obstruction) — scheme- divergence documented honestly (#677), not forced into 1:1. PK 942 cross-reference PR-2: resisted by BOTH 155 (Fidélité 6.1) and 212 (Échange 7.3) — 2 distinct CQs (epistemic credibility vs dialogic loyalty), modeled as 2 arcs not deduplicated. Distribution: 6/6 undercut/RA-node (interpersonal fallacies short-circuit the commitment→evaluation/dialogue inference, don't attack facts or conclusions). Uniform like PR-1. KEY CUMULATIVE FINDING (load-bearing for Option decision): across PR-1/PR-2/PR-3 = 3 families, 3 subfamilies, 9 leaves, ~14 fallacy-instances — rebut/CA-node NEVER cleanly instantiated (0 occurrences). Structural explanation: fallacies are inference-errors (undercut) or false-premises (undermine), rarely independent counter-arguments. PK 1398 (ad hominem) is defined in the taxonomy itself as "au lieu de réfuter" — the explicit negation of rebuttal. Implication: Option A AIF_attackType column will be undercut+undermine-dominated with rebut as a sporadic tail case. Recommend not over-investing rebut coverage in validator tooling. Reinforces Option A: "good holding of a scheme" = resistance to inference-breakdown (undercut, dominant) + premise-falsity (undermine). Limit: 3 clusters don't cover all 8 families; rebut may surface in unsampled families (1.x Argument pertinent, 2.x Présentation). Finding is directional. Recommend 1-2 more clusters on varied-scheme families before locking Option. Option A/B/C unchanged (doc written under A; only 2 final columns are Option- sensitive). 0 fabrication (#677), 0 write CSV/DB/OWL/Cards, docs-only. Co-authored-by: Your <your.email@example.com> Co-authored-by: Claude-Code <noreply@anthropic.com>
jsboige
added a commit
that referenced
this pull request
Jul 10, 2026
…ks (59%) + AIF attack wiring (#141/#498) (#763) The transverse cross-link layer was stuck at 1.5% (21 nodes) and the AIF attack-typing was inert because the Fallacies OWL generator only emitted the AIF_skos* Walton mapping — the crossLink_* and AIF_attack* CSV columns were never read (unmapped in the Fallacy entity) nor emitted. Populating the CSV alone left them dead. "Complete integration" = code path + curated data + regen. Code (makes the columns emittable in OWL): - Fallacy.cs: map the 8 crossLink_* columns + AIF_attackType/AIF_attackedNode (10 props + 10 .Optional() ClassMap lines, mirroring the AIFSkos* block). - OwlGeneratorConfig.cs: 2nd emission pass (concepts dict fully populated) — declares 8 crossLink ObjectProperties, resolves ';'-split dotted targets via fallaciesByPath, emits AnnotateConceptWithResource (symmetric verbs opposes/inverts/mirrors/isRelatedTo emitted bidirectionally), and for each AIF-typed leaf emits aifAttackType (literal) + aifAttackedNode (RA/I/CA-node, per ASPIC+ #707§4: undercut→RA, undermine→I, rebut→CA). - Program.cs: add --generate-owl entry point (single-stage OWL regen, no harvest /PDF), mirroring --pdf-cmyk — the reproducible way to refresh the committed OWL. - OwlE2EGenerationValidationTests: regression guard asserting crossLink + aifAttack assertions survive the round-trip (#133 known-fail left untouched). Data (additive, byte-exact — tools/141-crosslink-apply.py, gated): - Layer A crosslinks: from the 0-fabrication scan candidates, transverse subset (target neither ancestor nor descendant of source) at confidence >= 0.6, no warns → 1210 relations; 1 dropped (source path absent from taxonomy), 1 seed collision skipped → 1059 empty cells filled (1208 relations, max 3/cell). Coverage 1.5% → 59.9% (844/1408 nodes). All 22 manual seeds preserved untouched; independent byte-check confirms only crossLink columns changed, 0 non-crossLink cells across all 104 columns, CRLF+BOM preserved. - AIF attack: the 93 typed leaves already present (merged #760) become emittable via the wiring above — no new CSV write. Distribution 61 undercut / 29 undermine / 3 rebut. Regen: docs/ontology/argumentum.owl regenerated via --generate-owl — a pure- addition diff (+10851/-0), faithful superset (prefLabel 2816, hierarchy, skos matches identical). Now carries isRelatedTo 643, mirrors 721, leverages 403, …, aifAttackType 93, aifAttackedNode 93, +3 AIF node classes. Ontology metadata only — not rendered on cards, orthogonal to the v0.9.0 visual GO. Tests: 595 pass / 1 fail (pre-existing #133 OWLSharp round-trip known-fail, proven identical on old OWL) / 5 skip. Zero-warning solution build. Deferred (gated follow-ups, not part of this chantier's two ratified decisions): - Layer B — AIF_skos bulk Walton expansion (default mappingType is a semantic call not yet ratified; existing 42/50 refs still emit). - Layer C — generative attackType fill for the ~866 untyped leaves. - crossLink emission on the Virtue side (already loaded, deferred there). Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
jsboige
added a commit
that referenced
this pull request
Jul 10, 2026
…t rebut cluster, gated proposal) (#766) First dedicated cluster doc for a rebut family (Obstruction / Relativisme abusif). Opens 4 unmapped leaves (1283-1286) under anchor 1282 (serialized rebut/CA-node by #760). Proposition only — GATED, 0 prod CSV write. Deliverables (docs/taxonomy/): - 498-aif-relativisme-abusif-cluster.md (prose §1-7 + gate boundaries) - 498-relativisme-abusif-annotations.csv (12-col machine-readable, 5 rows) Model (first rebut cluster = attack-columns-only pattern): - All 4 leaves: AIF_attackType=rebut + AIF_attackedNode=CA-node (deterministic map #707§4 option a: rebut -> CA-node). - AIF_skosExceptionRef=Commitment_Inference (native) for 1283 only (self-exemption honestly rebuts the arguer's own commitment); the other 3 have no clean native rebutted scheme -> ExceptionRef empty, fail-loud in AIF_skosOther. - AIF_skosDirectRef empty for all 4 (bare relativist rebut = structural conflict, no native Walton CQ-conflict token). #677 discipline (0 fabrication, code=truth verified): - CounterConclusionConflict_Conflict / IssueAvoidance_Conflict / TuQuoque_Conflict verified NOT native (absent from all filled skos cols) -> correctly excluded (as #760 did for the 3 merged rebuts). - Commitment_Inference verified native. - Annotation CSV uses honest structural descriptor in AIF_CA_node (no fabricated token) — improvement over phase-3 placeholder practice. Honest per-leaf boundary flags: 1284 undermine-flavor, 1285 undercut-flavor, 1286 false-equivalence — documented in prose + why_not_others, not silently forced. Coverage: 93/1408 mapped (code=truth rescan) -> 97 projected once applied; rebut population 3 -> 7 (Relativisme abusif becomes the rebut nucleus, 5 of 7). Gated review ai-01; jsboige ratifies -> gated apply. Relates #498 #760 #757 #753 #755 #763 #707 #677. Co-authored-by: Your <your.email@example.com> Co-authored-by: Claude-Code <noreply@anthropic.com>
jsboige
added a commit
that referenced
this pull request
Jul 10, 2026
…ows typed (gated proposal + dry-run apply) (#769) Réconciliation P1 (skos-only -> attack columns), GO ai-01 `msg-20260710T180845-5i1v03`, couvert par le pilote GO #498. Proposition gated : docs + apply-script dry-run, 0 write prod CSV dans ce diff. Correction load-bearing du cadrage #768 : il n'existe PAS de sous-ensemble "inherit mécanique" parmi les 52 skos-only. La classif initiale "19 inherit du sous-sous anchor" est un artefact — 0/19 ont une signature skos alignée sur leur anchor (les anchors du même sous-sous sont skos-VIDES : couche attack-only #753/#760, arguments différents). Hériter leur attackType serait une fabrication. L'attackType est un jugement neuf : 0-risque token (les tokens skos existent), pas 0-risque modélisation. Raffine le "0 fabrication risk" de #768. Méthode : attackType dérivé de la propre signature skos de chaque ligne, ancré sur les 18 lignes fully-modeled (ground truth token->attackType). attackedNode déterministe #707§4 (a) : undercut->RA-node, undermine->I-node, rebut->CA-node. Tiering des 52 : - 14 PRECEDENT (token exact d'un précédent fully-modeled + contrôle sémantique) -> tranche 1, distribution 11 undermine / 2 undercut / 1 rebut. - 2 PREC-TIE (777 Inconsistance, 633 Relation infondée — votes divergents). - 36 SUFFIX-ONLY (aucun précédent ; le prior de suffixe vers undermine est non fiable — contre-ex. pente glissante 705/677 penchent undercut) -> modélisation Walton au cas par cas, sous-lots aux prochains ticks. Livrables : - docs/taxonomy/498-reconciliation-p1.md — proposition (correction, méthode, tranche 1 avec justif par ligne, reste scopé, bornes du gate). - docs/taxonomy/498-reconciliation-p1-annotations.csv — 12 cols, 14 lignes, BOM+CRLF (factuel lu du CSV master, raisonnement authored). - tools/498-p1-apply.py — apply gated (--write dry-run par défaut), mirroir de 498-phase13-apply.py (#757). Re-vérifie 14/14 vs annotation CSV, splitters byte-exact, preuve byte-preservation. Dry-run : 0 mismatch, 93 -> 107 attack-typed, delta 207 bytes, CRLF+BOM préservés, 0 write prod. Sérialisation prod = étape suivante gated (relais ai-01), pas dans ce PR. Co-authored-by: Your <your.email@example.com> Co-authored-by: Claude-Code <noreply@anthropic.com>
This was referenced Jul 10, 2026
jsboige
added a commit
that referenced
this pull request
Jul 11, 2026
…(foundation of the derivations) (#770) Adds §2bis to the P1 reconciliation proposition (#769): a read-only audit of the 18 fully-modeled anchors on which every P1 attackType derivation rests. If an anchor is mis-typed, the error propagates into every derived tranche and, eventually, prod — so the anchor set is validated before more rows are derived. Audit (hand-Walton "what does the CQ defeat" + code=truth cross-check): - layer census re-confirmed vs #768: fully-modeled 18 / attack-only 75 / skos-only 52. - node-map #707§4a (undercut->RA / undermine->I / rebut->CA): 18/18 clean, 0 violation. - skos<->attackType coherence: 16 CLEAN, 2 SOFT (defendable), 0 outright error. Foundation VALIDATED — the 14 tranche-1 derivations all rest on CLEAN anchors; the only SOFT dependency is pk1281<-pk1313 (same family, same rebut nucleus). Two SOFT anchors flagged (choices, not errors): pk1313 Evasion (rebut as global dialogue-level conflict) and pk1361 Proces en incoherence (rebut vs undercut/ undermine judgment call — it is the anchor 777->rebut leans on, so 777 must be tranched by desc_fr + co-token, not mechanically). Three findings that HARDEN the method (beyond validation): 1. suffix _Conflict/_Inference does NOT determine attackType — in-set proof: pk804 PropertyNotExistant_Conflict -> undercut (vs _Conflict->undermine elsewhere). Empirically grounds the "suffix-prior unreliable" claim (§2/§5), previously argued only from 677/705. 2. slippery-slope leans undercut — in-set anchor pk858 VerbalSlipperySlope_ Inference -> undercut grounds the 677/705 counter-example (previously CQ- semantic only). 3. CircumstantialAdHominem_Inference is non-decisive (in both 1361 rebut and 1371 undermine) -> 777/633 tie-break must read co-token + desc_fr. Deliverables (read-only vs prod taxonomy — 0 CSV write in the diff): - docs/taxonomy/498-reconciliation-p1.md — §2bis audit + TL;DR point 5. - docs/taxonomy/498-audit-18-anchors.csv — 10 cols, 18 rows, BOM+CRLF (factual cols read from master e748735; verdict+note authored). - tools/498-audit-anchors.py — reproducible read-only audit (re-derives the 18 from prod, checks node-map, cross-verifies the audit CSV). RESULT: OK. Co-authored-by: Your <your.email@example.com> Co-authored-by: Claude-Code <noreply@anthropic.com>
jsboige
added a commit
that referenced
this pull request
Jul 11, 2026
…k-typed (GO ai-01, 93->107) (#771) Sérialisation prod de la tranche-1 P1 (14 lignes PRECEDENT), après merge de la proposition #769 (master daa7176) et GO write d'ai-01 (msg-20260710T230135-kpea48, couvert par le pilote GO #498 jsboige 2026-06-17). Flow #760 MODE 2. Back-fill des 2 colonnes attack (AIF_attackType / AIF_attackedNode) pour 14 lignes skos-only dont l'attackType est dérivé de leur PROPRE signature skos, ancré sur les 18 fully-modeled (ground truth), node déterministe #707§4(a). Aucun token AIF créé (#677) : on TYPE des lignes déjà skos-vettées. attackType = jugement neuf (0-risque token, pas 0-risque modélisation, cf. correction #769). Distribution : 11 undermine / 2 undercut / 1 rebut. undercut (RA-node) : 4 Appel à l'ignorance, 621 Transfert illicite rebut (CA-node) : 1281 Refus du débat undermine (I-node) : 1198 1083 1090 1092 1104 1 3 70 133 799 846 2 lignes MED flaggées (traçables, non cachées) : pk3 (Argument vide) et pk1090 (Biais émotionnels) — vote issu d'un token de scheme secondaire, undermine cohérent sémantiquement mais à re-confirmer. Les 12 autres = HIGH. Write byte-exact, vérifié indépendamment (backup tmp/ vs prod) : - 14 lignes changées == les 14 PK de l'annotation ; 0 changement hors les 2 colonnes cibles ; pré-état 14 cellules vides (fill, pas overwrite). - 1409 lignes x 104 colonnes inchangées ; BOM + CRLF + 144 LF intra-cellule préservés ; delta +207 bytes (exact). - couche attack : 93 -> 107 ; fully-modeled (skos & attack) : 18 -> 32. Co-authored-by: Your <your.email@example.com> Co-authored-by: Claude-Code <noreply@anthropic.com>
jsboige
added a commit
that referenced
this pull request
Jul 11, 2026
…PREC-TIE + 5 SUFFIX-ONLY, gated dry-run) (#773) Back-fill proposition for 7 of the 52 skos-only rows in the two-layer AIF reconciliation (#498 P1), continuing tranche-1 (#769/#771, 14 PRECEDENT rows now in prod). Docs + gated dry-run apply script only — 0 prod CSV write. Tranche-1b = the rows with NO exact-token precedent among the 18 fully-modeled anchors (audited CLEAN #770): - 2 PREC-TIE (777 Inconsistance, 633 Relation infondée): tokens vote for divergent attackTypes; arbitrated by co-token + desc_fr + family (tie-break rule hardened by the #770 audit — never a shared/divergent token alone). - 5 SUFFIX-ONLY (337 Appel à la terreur, 432 Engagement, 356 Manipulation mentale, 1280 Obstruction, 1360 Ad hominem): modeled per-row from each fallacy's OWN Walton scheme + desc_fr. The suffix prior (_Conflict/_Inference→undermine) is PROSCRIBED (in-set disproof pk804 _Conflict→undercut, #770). Distribution: 4 undermine / 2 undercut / 1 rebut. NODE is the deterministic ASPIC+ map (#707§4 Option a): undercut→RA-node, undermine→I-node, rebut→CA-node. No AIF token fabricated (#677) — we TYPE rows already carrying a vetted skos. Files: - docs/taxonomy/498-reconciliation-p1b.md — proposition (arbitrage + per-row) - docs/taxonomy/498-reconciliation-p1b-annotations.csv — machine-readable source - tools/498-p1b-apply.py — byte-exact cell-fill, GATED (--write held for ai-01) Dry-run (master f34ac77, post-#771): 107 → 114 attack-typed, 0 byte-preservation mismatch, 1409×104, CRLF+BOM preserved, +102 B. Serialization gated on ai-01. Co-authored-by: Your <your.email@example.com> Co-authored-by: Claude-Code <noreply@anthropic.com>
jsboige
added a commit
that referenced
this pull request
Jul 11, 2026
… 107→114, GO ai-01) (#776) Prod write of the tranche-1b attack-layer back-fill (proposition merged #773), under ai-01's write relay (msg-20260711T070142, pilote GO #498). Fills the 2 AIF attack columns (AIF_attackType / AIF_attackedNode) on 7 skos-only rows via `tools/498-p1b-apply.py --write`. Rows (2 PREC-TIE arbitrated + 5 SUFFIX-ONLY modeled), distribution 4 undermine / 2 undercut / 1 rebut: - 777 Inconsistance -> undermine / I-node - 633 Relation infondee -> undercut / RA-node - 337 Appel a la terreur -> undermine / I-node - 432 Engagement -> undercut / RA-node - 356 Manipulation mentale-> undermine / I-node - 1280 Obstruction -> rebut / CA-node - 1360 Ad hominem -> undermine / I-node Byte-exact verified 2x (script self-proof + independent backup-vs-prod): exactly 14 cells changed (7 rows x 2 cols), 0 outside AIF attack cols, 0 outside target PKs, all empty->value fills (no overwrite), node/type 7/7 consistent, 1409x104, BOM+CRLF preserved, +102 B. Attack-typed total 107 -> 114 (fully-modeled 32 -> 39). Backup tmp/Fallacies-backup-pre-p1b.csv (gitignored, not committed). Mirror of #771 (tranche-1 write). NODE deterministic (#707§4a). 0 token fabricated (#677) — types rows already carrying a vetted skos. Co-authored-by: Your <your.email@example.com> Co-authored-by: Claude-Code <noreply@anthropic.com>
This was referenced Jul 11, 2026
feat(taxonomy): #498 Fallacies AIF — P1 tranche-1e (5 SUFFIX-ONLY, famille Insuffisance, gated)
#780
Merged
jsboige
added a commit
that referenced
this pull request
Jul 11, 2026
…FIX-ONLY, famille Insuffisance, gated) (#780) P1 reconciliation back-fill, tranche-1e = 5 SUFFIX-ONLY rows (2/33/34/43/71), the whole "Insuffisance" family (Argument bâclé + Préjugé). All 5 tokens are novel (no in-set typed precedent). Modeled per-row Walton from own scheme + desc_fr ("what does the CQ defeat"): all epistemic-insufficiency -> the cited evidence is too weak to support the conclusion -> undercut/RA-node (5/5). attackType = NEW judgment; NODE deterministic (#707§4a); 0 token fabricated. 2 MED flags for ai-01 review: - pk33 (Justification triviale): skos token AlternativeMeans_Conflict is a loose map vs desc_fr (induction-from-example); both readings converge on undercut but the disconnect is documented. - pk43 (Pratique courante): token PopularPractice_Inference; tension with PopularOpinion_Inference typed undermine (3/3, pks 3/70/177). Practice-vs- opinion / justification-vs-truth distinction argues undercut; ai-01 may prefer aligning on the 'Popular*' -> undermine pattern. 0 prod write (gated dry-run). Baseline dynamic (write-driven, disjoint PKs): 114->119 (master c1ed77d) / 121->126 (if 1c written) / 125->130 (if 1c+1d). Byte-exact: 0 mismatch, 1409x104, CRLF+BOM preserved, +75 B. - docs/taxonomy/498-reconciliation-p1e.md (proposition) - docs/taxonomy/498-reconciliation-p1e-annotations.csv (5 rows, 12 cols) - tools/498-p1e-apply.py (gated apply, dry-run default, mirrors p1c) Co-authored-by: Your <your.email@example.com> Co-authored-by: Claude-Code <noreply@anthropic.com>
jsboige
added a commit
that referenced
this pull request
Jul 11, 2026
…FIX-ONLY, famille Abus de langage, MIXTE gated) (#781) P1 reconciliation back-fill, tranche-1f = 7 SUFFIX-ONLY rows (800/808/814/833/839/856/876), the whole "Abus de langage" family (Definition inexacte + Comparaison fallacieuse + Ambiguite). UNLIKE the inference-uniform families (1c/1d/1e all-undercut), this family is HETEROGENEOUS: definitional/classificatory manipulations attack the PREMISE (term application / exhaustivity / comparison-aptitude contestable) -> undermine/I-node (6 rows); only the analogy scheme attacks the INFERENCE rule -> undercut/RA-node (pk839). Distribution: 6 undermine / 1 undercut. 5/7 tokens novel; 2 with undermine precedent (BiasedClassification_Conflict-> pk799 anchors pk833; ConflictingGoals_Conflict->pk356 anchors pk876). attackType = NEW judgment; NODE deterministic (#707§4a); 0 token fabricated. 4 MED flags (800, 808, 856, 876) — honest cost of a heterogeneous family: - pk800 (Acception vague): vagueness = ill-defined premise -> undermine; alt undercut if read as equivocation-by-shifting. - pk808 (Sophisme des correlatifs): DISCONNECT skos/desc (token AlternativeMeans_Conflict suggests means-end undercutter; desc is definitional). desc-driven -> undermine. SAME TOKEN as tranche-1e pk33 (typed UNDERCUT, induction desc). Opposite verdicts justified by desc divergence (definitional vs inferential), NOT by token — flagged as the most delicate point for ai-01 (tests the desc-driven principle). - pk856 (Expression vague): same token as pk800 -> undermine; alt equivocation. - pk876 (Ambiguite narrative): ConflictingGoals precedent -> undermine; alt undercut if ambiguity read as equivocal inference. 3 HIGH (814 false dilemma, 833 fallacious comparison precedent, 839 false analogy scheme undercutter). Clean premise-vs-inference split documented (833 classification-biased vs 839 analogy-transfer). 0 prod write (gated dry-run). Baseline dynamic (write-driven, disjoint PKs): 114->121 (master c1ed77d) / 121->128 (if 1c written) / 125->132 (if 1c+1d) / 130->137 (if 1c+1d+1e). Byte-exact: 0 mismatch, 1409x104, CRLF+BOM, +105 B. - docs/taxonomy/498-reconciliation-p1f.md (proposition, MIXTE) - docs/taxonomy/498-reconciliation-p1f-annotations.csv (7 rows, 12 cols) - tools/498-p1f-apply.py (gated apply, dry-run default, mirrors p1e) Co-authored-by: Your <your.email@example.com> Co-authored-by: Claude-Code <noreply@anthropic.com>
jsboige
added a commit
that referenced
this pull request
Jul 11, 2026
…FIX-ONLY, famille Tricherie, MIXTE gated, LAST FAMILY) (#784) P1 reconciliation back-fill, tranche-1g = 8 SUFFIX-ONLY rows (888/973/1020/1023/1066/1087/1148/1175), the whole "Tricherie" family (Arranger les faits + Changement de cap + Pensee biaisee). LAST SUFFIX-ONLY family -> closes the 36-row back-fill (5 1b + 7 1c + 4 1d + 5 1e + 7 1f + 8 1g = 36). HETEROGENEOUS (MIXTE): 3 undermine / 5 undercut. Three defect natures: - relational manipulation (deceptive arrangement, goalpost-moving, biased reasoning) -> premise-level (undermine/I): 888, 973, 1023 (precedents PositionToKnow pk70, ConflictingGoals pk356, Bias pk70). - sunk-cost / effort-justification -> inference rule undercut via Waste scheme (undercut/RA): 1020, 1148 (precedent Waste_Inference pk432). - cognitive induction biases (mere exposure, anecdotal, attribution) -> weak induction (undercut/RA): 1066, 1087, 1175 (1066/1175 Example_Inference novel; 1087 desc-twin of tranche-1e pk34). 6/8 tokens have a typed precedent; 2/8 novel. attackType = NEW judgment; NODE deterministic (#707§4a); 0 token fabricated. 3 MED flags (888, 1066, 1175): - pk888 (Presentation trompeuse): SAME-TOKEN DIVERGENCE with tranche-1e pk2 (LackOfPTKReliability_Scheme). pk2->undercut (induction); pk888->undermine (deceptive arrangement = biased premise, reinforced by PositionToKnow pk70). 2nd such divergence (after 1f pk808/pk33) — reinforces desc-driven principle. - pk1066 (Effet de simple exposition): Example_Inference novel; undercut (familiarity->acceptable defective rule); alt undermine. - pk1175 (Biais d'attribution): Example_Inference novel; undercut (attribution over-systematic); alt undermine. 5 HIGH: 973 (ConflictingGoals precedent), 1020 (Waste precedent sunk cost), 1023 (Bias+Presumptive precedent), 1087 (desc twin pk34), 1148 (Waste precedent). 0 prod write (gated dry-run). Baseline dynamic (write-driven, disjoint PKs): 121->129 (master cbf9b7c) / 137->145 (if 1d+1e+1f #783 merged). Byte-exact: 0 mismatch, 1409x104, CRLF+BOM, +120 B. - docs/taxonomy/498-reconciliation-p1g.md (proposition, MIXTE, last family) - docs/taxonomy/498-reconciliation-p1g-annotations.csv (8 rows, 12 cols) - tools/498-p1g-apply.py (gated apply, dry-run default, mirrors p1f) Co-authored-by: Your <your.email@example.com> Co-authored-by: Claude-Code <noreply@anthropic.com>
jsboige
added a commit
that referenced
this pull request
Jul 11, 2026
… 137→145, GO ai-01 gc40e5) (#785) P1 reconciliation tranche-1g WRITE = 8 SUFFIX-ONLY rows (888/973/1020/1023/1066/1087/1148/1175), the whole "Tricherie" family (Arranger les faits + Changement de cap + Pensee biaisee). LAST SUFFIX-ONLY family -> closes the 36-row back-fill (5 1b + 7 1c + 4 1d + 5 1e + 7 1f + 8 1g = 36). attack-typed 137 -> 145. MIXTE distribution: 3 undermine / 5 undercut (no rebut). Three defect natures: - relational manipulation (deceptive arrangement, goalpost-moving, biased reasoning) -> premise-level (undermine/I): 888, 973, 1023. - sunk-cost / effort-justification -> inference rule undercut via Waste scheme (undercut/RA): 1020, 1148. - cognitive induction biases (mere exposure, anecdotal, attribution) -> weak induction (undercut/RA): 1066, 1087, 1175. GO from ai-01 (msg gc40e5): concur on all 3 MED verdicts: - pk888 (Presentation trompeuse) -> undermine/I. SAME-TOKEN DIVERGENCE with tranche-1e pk2 (LackOfPTKReliability_Scheme): pk2->undercut (induction), pk888->undermine (deceptive arrangement = biased premise, reinforced by PositionToKnow pk70). desc-driven, token non-decisif (aif-no-inherit). - pk1066 (mere exposure, Example_Inference novel) -> undercut/RA. - pk1175 (attribution bias, Example_Inference novel) -> undercut/RA. 5 HIGH: 973 (ConflictingGoals pk356), 1020 (Waste pk432 sunk cost), 1023 (Bias+Presumptive pk70 x2), 1087 (desc twin pk34), 1148 (Waste pk432 // 1020). Byte-exact cell fill (cols AIF_attackType/AIF_attackedNode only): - self-proof: 0 byte-preservation mismatch, 1409x104, CRLF+BOM, +120 B. - INDEPENDENT verify (backup-vs-prod, master 7c2c963 state 137): exactly 8 PKs changed, 16 cells (8 PKs x 2 cols), 0 stray, all empty->value, ATI==P1G_MAP, ANI==NODE map. - pre-state: all 8 target PKs empty + carry skos (skos-only back-fill). - 0 token fabricated (#677); NODE deterministic (#707 section 4 option a). Base master 7c2c963. Verdict QA = ai-01 (no self-merge). After this write the 36 SUFFIX-ONLY are closed -> the attack layer covers all 52 skos-only modulo the 2 PREC-TIE (777/633), next gated step. Co-authored-by: Your <your.email@example.com> Co-authored-by: Claude-Code <noreply@anthropic.com>
jsboige
added a commit
that referenced
this pull request
Jul 12, 2026
…45) + P1 closure doc (#787) Regenerate docs/ontology/argumentum.owl from master 474607c via the --generate-owl entry-point (single-stage, 0 harvest/PDF/mindmap). The committed OWL was stale (pre-P1): its AIF attack layer counted 93 annotation assertions while prod CSV had advanced to 145 via the #498 P1 reconciliation. 3-layer validation (argumentum-owl-validation-2026-07-12.md): - Layer 1 skos Walton: 70 mappings (stable, +0) - Layer 2 crossLink #763: 1985 relational verbs (stable, +0) - Layer 3 AIF attack #498: 93 -> 145 (+52) == CSV prod attack-typed total PASS Distribution reflected: 87 undercut/RA-node, 53 undermine/I-node, 5 rebut/CA-node. 0 node/type inconsistency in prod (deterministic ASPIC+ Option a, #707 section 4). Diff scope verified AIF-only: 530 insertions / 10 deletions, every added line an <AnnotationAssertion> for aifAttackType/aifAttackedNode. skos + crossLink + concept declarations + multilingual labels byte-stable. argumentum_virtues.owl regenerated byte-identical (not re-committed). Also adds docs/taxonomy/498-reconciliation-p1-closure.md — the P1 closure report: 52 skos-only back-filled (14 PRECEDENT + 2 PREC-TIE + 36 SUFFIX-ONLY), tiering methodology, 7-tranche serialization record (93->145), defect-nature modeling, same-token divergence principle (pk808/pk33, pk888/pk2), anchor audit #770 foundation (16 CLEAN/2 SOFT). Feeds release notes + #133. 0 write prod CSV (--generate-owl reads CSV, emits OWL; Cards/ untouched). Layer C (~750 leaves, generative pass) = jsboige decision, out of scope, not started. Co-authored-by: Your <your.email@example.com> Co-authored-by: Claude-Code <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Read-only synthesis note consolidating the #498 AIF chantier after 4 PR, to support jsboige's ratification of (a) the méthode and (b) the I/RA/CA CSV schema decision (the enriched-DoD serialization). This is the idle-de-secours deliverable named in dispatch
mkzt2v. 0 prod write.§1 — Coverage code=truth (master
70bd1605)70/1408 fallacies mapped (5.0%) across 8 families. Highest coverage: Misleading language 12.6%, Faulty logics 10.8%. Flags a data-quality gap: 48 rows with empty
Family(out of chantier scope, separate hygiene item for jsboige).§2 — Chantier progress (4 PR)
Cumulative: 70→81 fully-modeled (+11, +16%). Fallacious comparison subfamily complete (3/3 sub-subs, 13/13 leaves). Inexact definition opened (1/3).
§2b — FAIL-LOUD accounting (honest gaps, 0 fabrication)
4 leaves, 2 distinct layers:
Comparison_Inferencein the 36-scheme vocabulary.All 4 take a DirectRef (direct-conflict pattern) + document the gap in
AIF_skosOther. No token fabricated.§3 — Méthode recap (validée ai-01 « rigoureuse, 0 fabrication »)
Cluster = 1 sub-sub. 2 cluster shapes (in-sub-sub anchor / borrowed-root anchor), 2 patterns (exception / direct-conflict). Native-vocab discipline (#677): restricted to 26 Conflict nodes + 36 Inference schemes confirmed by existing usage. Validated programmatically per PR.
§4 — ⚠ Decision jsboige requise: I/RA/CA serialization
The enriched DoD (jsboige's last #498 comment — I-node/RA-node/CA-node decomposition, typed undermine/undercut/rebut) is recorded in §7 of each PR but NOT YET SERIALIZED in the CSV. Three options:
AIF_attackType+AIF_attackedNode(reco — serialize early avoids a massive late back-fill, gives the EPITA harness structured data from the start).AIF_decompositionJSON column).§5 — Trajectory & recommandation
~87 sub-subs total; at ~3-4 leaves/PR this is a long chantier (~30-40 PR for anchor-mapped sub-subs). Reco: confirm focus on Misleading language (12.6%) + Faulty logics (10.8%) = best anchor/leaves ratio. Next worker-able targets (post-batch ai-01): Inexact definition/Arbitrary definition (3/19, large — needs splitting), Inconsistent definition (0/7 borrow-root), Ambiguity sub-subs.
Gate (HARD)
70bd1605+ chantier PRs). No token fabricated in the synthesis.Relates
dispatch
mkzt2v(idle-de-secours), #498 (chantier), PR-1 #699, PR-2 #701, PR-3 #703, PR-4 #705, #133/#130 (OWL), #499 (inverse), #677 (0 fabrication), #192, #458.— po-2023