Skip to content

docs(taxonomy): #498 Fallacies AIF — Relativisme abusif cluster (first rebut cluster, gated proposal) #766

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
jsboige merged 1 commit into from
Jul 10, 2026
+291 −0
Merged

docs(taxonomy): #498 Fallacies AIF — Relativisme abusif cluster (first rebut cluster, gated proposal) #766

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
285 changes: 285 additions & 0 deletions docs/taxonomy/498-aif-relativisme-abusif-cluster.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,285 @@
# 2026-07-10 — #498 AIF chantier : cluster « Relativisme abusif » (Obstruction) — **first rebut cluster doc**

**Scope**: first dedicated cluster doc for a **rebut** family in the Fallacies taxonomy. Opens the
**Relativisme abusif** sub-sub (family **Obstruction**) — 4 unmapped leaves (1283-1286) under the
anchor **1282** (already serialized `rebut`/`CA-node` by #760). **Proposition only — GATED, 0 write
to prod CSV.** Continues the cluster-docs authoring backlog (ai-01 dispatches `kxxbqd`/`nnzspt`/`hgicfj`,
primary = ~80 unmapped sub-subs).

**Repo reference**: master `7796c127`. Issue: #498. Predecessors (undercut/undermine clusters):
PR-1 #699 … PR-5 #706 (Arbitrary definition), PR-12 #720 (Equivoque complete). Serialization:
#753 (§7 cluster docs), #755 (Virtues mirror), #757 (phase 1-3 apply/verify), **#760 (phase 1-3
serialize — wrote the anchor 1282 + the 3 first Fallacies rebuts)**. Ontology wiring: #763 (OWL now
emits `AIF_attack*`).

> **⚠ STATUS = PROPOSITION ONLY.** No prod CSV write, no DB write, no OWL regen. jsboige ratifies →
> worker applies the CSV edits in a follow-up PR (gated), same flow as #753/#755/#760. The
> attack-type decomposition is recorded here + in the machine-readable annotation CSV
> `498-relativisme-abusif-annotations.csv`; serialization to the 2 columns (`AIF_attackType`,
> `AIF_attackedNode`) is deferred to the gated apply.

---

## TL;DR

- **Relativisme abusif** (Obstruction, sub-sub, 5 leaves total): anchor **1282** MERGED
(`rebut`/`CA-node`, #760), **4 unmapped leaves** — 1283 Sophisme subjectiviste, 1284 Truthiness,
1285 Droit à mon opinion, 1286 Plus faux que faux.
- **First rebut cluster.** All 4 leaves honestly model the **same structural move as the anchor**:
opposing a **subjective/leveling counter-conclusion** that denies the shared conclusion's scope or
authority → **rebut / CA-node** (deterministic map ratified #707§4 option a: rebut → CA-node).
- **Attack-columns-only pattern (load-bearing #677 finding).** The 3 merged rebut rows (1282/1313/
1361) serialize as **attack columns only** — `AIF_skosDirectRef` **empty** (no native Walton
CQ-conflict token for a bare rebut), `ExceptionRef` = the rebutted *native* scheme *iff one honestly
applies*. The coined `CounterConclusionConflict_Conflict` from the phase-3 annotation notes is **NOT
native** (verified: absent from all filled skos columns) and is **correctly not serialized**. This
cluster follows that reality: **0 fabricated token**.
- **Honest per-leaf serialization**: `AIF_attackType=rebut` + `AIF_attackedNode=CA-node` for all 4;
`AIF_skosExceptionRef=Commitment_Inference` (native) for **1283 only** (the self-exemption honestly
rebuts the arguer's own commitment); the other 3 have **no clean native rebutted scheme** →
ExceptionRef empty, fail-loud documented in `AIF_skosOther`.
- **4 attack-typed rebut leaves proposed, 0 fabrication, 3 documented boundary nuances** (1284
undermine-flavor, 1285 undercut-flavor, 1286 false-equivalence).

---

## 1. Cluster state (code=truth, master `7796c127`)

From `Cards/Fallacies/Argumentum Fallacies - Taxonomy.csv`, family **Obstruction**, sub-sub
**Relativisme abusif** (5 rows):

| pk | text_fr | depth | AIF_attackType | AIF_attackedNode | skos cols | State |
|----|---------|-------|----------------|------------------|-----------|-------|
| **1282** | **Relativisme abusif** (anchor) | d3 | `rebut` | `CA-node` | *(all empty)* | **✅ MERGED #760** |
| 1283 | Sophisme subjectiviste | d4 | — | — | — | unmapped (**this cluster**) |
| 1284 | Truthiness | d5 | — | — | — | unmapped (**this cluster**) |
| 1285 | Droit à mon opinion | d5 | — | — | — | unmapped (**this cluster**) |
| 1286 | Plus faux que faux | d4 | — | — | — | unmapped (**this cluster**) |

**Leaf content (code=truth `desc_fr` / `example_fr`)**:
- **1282 Relativisme abusif** — « Vous affirmez que la vérité est subjective et propre à chaque
individu. » / « C'est peut-être vrai pour vous, mais cela ne l'est pas pour moi. »
- **1283 Sophisme subjectiviste** — « Vous affirmez qu'une idée est vraie pour les autres, mais pas
pour vous, afin d'éviter de vous l'appliquer. » / « Oui, fumer est mauvais pour la santé, mais pas
pour moi. »
- **1284 Truthiness** — « Vous affirmez qu'une chose est vraie parce que vous la ressentez ainsi, sans
preuve concrète. » / « Je n'ai pas besoin de statistiques pour savoir que le changement climatique
est un mythe : je le vois bien. »
- **1285 Droit à mon opinion** — « Vous prétendez que votre avis vaut autant qu'un fait établi, sans
avoir à le justifier. » / « J'ai le droit de croire que ces médicaments sont plus dangereux
qu'utiles. Après tout, chacun a droit à sa propre opinion. »
- **1286 Plus faux que faux** — « Vous assimilez des erreurs de gravité différente en les présentant
comme équivalentes. » / « Il a peut-être grillé le feu rouge à cent à l'heure, mais vous traversiez
en dehors des clous. Vous avez tous les deux tort. »

The sub-sub is **semantically coherent around one mechanism**: abusive relativism *opposes a subjective
or leveling counter-conclusion* to a shared/graded conclusion, refusing its scope or authority. Every
leaf is a specialization of that move (self-exemption, felt truth, epistemic parity, false leveling).

---

## 2. The anchor model (1282, MERGED #760) — attack-columns-only

`1282` was serialized by #760 as `AIF_attackType=rebut`, `AIF_attackedNode=CA-node`, **all skos
relational columns empty**. This is deliberate and #677-honest: a bare relativist rebut is a
**structural counter-conclusion conflict**, not a Walton-CQ conflict. There is **no native
CQ-conflict token** for it — the phase-3 annotation note recorded a *placeholder*
`CounterConclusionConflict_Conflict`, but that token is **not native** and was **correctly dropped**
at serialization.

**Verification (code=truth, `498-native-whitelist` scan of prod CSV)**:

| token | native? (in a filled skos column) |
|-------|-----------------------------------|
| `CounterConclusionConflict_Conflict` | ❌ NOT native (absent) |
| `IssueAvoidance_Conflict` (1313 note) | ❌ NOT native (absent) |
| `TuQuoque_Conflict` (1361 note) | ❌ NOT native (absent) |
| `Commitment_Inference` | ✅ native |

The 3 merged rebut rows confirm the serialized shape:

| pk | AIF_attackType | AIF_attackedNode | AIF_skosDirectRef | AIF_skosExceptionRef |
|----|----------------|------------------|-------------------|----------------------|
| 1282 Relativisme abusif | rebut | CA-node | *(empty)* | *(empty)* |
| 1313 Évasion | rebut | CA-node | *(empty)* | `Dialogue_Scheme` (native) |
| 1361 Procès en incohérence | rebut | CA-node | *(empty)* | `InconsistentCommitment_Inference, CircumstantialAdHominem_Inference` (native) |

**Pattern (rebut cluster)**: `attackType=rebut` + `attackedNode=CA-node` always; `DirectRef` empty
(the fallacy is a bare counter-conclusion conflict, no CQ token); `ExceptionRef` = the **rebutted
native scheme** *iff* one honestly applies, else empty.

---

## 3. Proposed AIF structure for the leaves

For each leaf: `AIF_attackType=rebut`, `AIF_attackedNode=CA-node` (structural — the leaf opposes a
counter-conclusion). `AIF_skosExceptionRef` gets a **native** rebutted scheme only where honest;
otherwise empty with a fail-loud note in `AIF_skosOther`. **0 fabrication (#677).**

### pk 1283 — Sophisme subjectiviste (self-exemption) ✅ rebut
- **Mechanism**: the arguer *accepts* the shared truth (smoking harms) yet opposes a counter-conclusion
that it does not apply *to them* — a self-exemption that blocks the truth's application.
- **Proposal**: `attackType=rebut`, `attackedNode=CA-node`, `ExceptionRef=Commitment_Inference`
(native — the arguer's own commitment to the general truth is the rebutted argument), `DirectRef`
empty (structural conflict, no native CQ token), `MappingType=skos:narrowMatch` (d4 specialization).
- **Why rebut**: not undermine (the admitted premise is not contested — the arguer *grants* smoking
harms), not undercut (no inference rule declared inapplicable); the payload is a counter-conclusion
of self-exemption. The rebutted scheme (`Commitment_Inference`) is native → serializable.

### pk 1284 — Truthiness (felt truth) ⚠ rebut (boundary: undermine)
- **Mechanism**: asserts a counter-conclusion guaranteed by feeling alone (« c'est un mythe, je le
vois »), substituting felt certainty for evidence.
- **Proposal**: `attackType=rebut`, `attackedNode=CA-node`, `ExceptionRef` **empty** (no clean native
rebutted scheme — the felt assertion does not cleanly rebut one named Walton scheme), `DirectRef`
empty, fail-loud note in `AIF_skosOther`, `MappingType=skos:narrowMatch` (d5).
- **Documented boundary**: close to an *undermine* (the ressenti substitutes for the probatory
premise), but the charge is an asserted counter-conclusion, so **rebut**. Recorded honestly for
ai-01/jsboige adjustment.

### pk 1285 — Droit à mon opinion (epistemic parity) ⚠ rebut (boundary: undercut)
- **Mechanism**: claims an unjustified opinion equals an established fact, refusing the burden of
justification.
- **Proposal**: `attackType=rebut`, `attackedNode=CA-node`, `ExceptionRef` **empty** (no clean native
rebutted scheme), `DirectRef` empty, fail-loud note, `MappingType=skos:narrowMatch` (d5).
- **Documented boundary**: an *undercut* flavor (refusing the warrant that claims require
justification), but the payload is a counter-conclusion of epistemic parity (« mon avis vaut un
fait »), so **rebut**.

### pk 1286 — Plus faux que faux (false leveling) ⚠ rebut (boundary: false-equivalence)
- **Mechanism**: equates errors of different gravity as equivalent (« vous avez tous les deux tort »),
relativizing the graded distinction.
- **Proposal**: `attackType=rebut`, `attackedNode=CA-node`, `ExceptionRef` **empty** (no clean native
rebutted scheme), `DirectRef` empty, fail-loud note, `MappingType=skos:closeMatch` (d4 sibling,
distinct leveling mechanism).
- **Documented boundary**: false equivalence is often treated as a *comparison* defect, but here the
charge opposes a leveling counter-conclusion to the graded conclusion, and the leveling relativizes
gravity (the sub-sub theme), so **rebut**.

---

## 4. Cluster summary (proposed)

| pk | text_fr | attackType | attackedNode | ExceptionRef (rebutted scheme) | DirectRef | MappingType | Honest? |
|----|---------|-----------|--------------|-------------------------------|-----------|-------------|---------|
| 1282 | Relativisme abusif | rebut | CA-node | *(empty)* | *(empty)* | — | ✅ MERGED #760 |
| 1283 | Sophisme subjectiviste | rebut | CA-node | `Commitment_Inference` (native) | *(empty)* | narrowMatch | ✅ |
| 1284 | Truthiness | rebut | CA-node | *(empty — fail-loud)* | *(empty)* | narrowMatch | ⚠ boundary undermine |
| 1285 | Droit à mon opinion | rebut | CA-node | *(empty — fail-loud)* | *(empty)* | narrowMatch | ⚠ boundary undercut |
| 1286 | Plus faux que faux | rebut | CA-node | *(empty — fail-loud)* | *(empty)* | closeMatch | ⚠ boundary false-equiv |

**4 attack-typed rebut leaves proposed, 0 fabrication.** 1/4 (1283) carries a native rebutted scheme;
3/4 are attack-columns-only with a documented fail-loud on the rebutted-scheme layer. **4/4 fail-loud
on the CA-conflict token** (no native token for a bare relativist rebut — consistent with the merged
anchor 1282).

---

## 5. Method notes (additions for the chantier)

- **First rebut cluster → attack-columns-only pattern.** Undercut/undermine clusters (PR-1…PR-12)
serialize a scheme (`*_Inference`) + a conflict (`*_Conflict`) in the skos columns. **Rebut clusters
do not**: the load-bearing serialization is the 2 attack columns (`attackType=rebut`,
`attackedNode=CA-node`), the `DirectRef` stays empty (no CQ-conflict token), and `ExceptionRef`
holds the rebutted native scheme only where one honestly applies. This is not a gap — it is the
faithful AIF shape of a rebut (a symmetric conflict between conclusions, structural not CQ-based).
- **Non-native placeholder tokens must NOT be serialized (#677).** The phase-3 annotation notes coined
`CounterConclusionConflict_Conflict` / `IssueAvoidance_Conflict` / `TuQuoque_Conflict` as *decomposition
labels*. Verified against the prod CSV: **none is native**. #760 correctly serialized the 3 rebuts
as attack-columns-only, dropping these. This cluster keeps that discipline — the annotation CSV uses
the honest descriptor `(structural rebut — no native CQ-conflict token, fail-loud #677)` in the
`AIF_CA_node` column rather than propagating a fabricated token. **Improvement over the phase-3
placeholder practice.**
- **In-sub-sub coherence + honest boundary flags.** All 4 leaves inherit the anchor's rebut structure
(the sub-sub is one coherent mechanism). Where a leaf has a genuine secondary reading (1284
undermine, 1285 undercut, 1286 false-equivalence), it is **flagged in prose + `why_not_others`**, not
silently forced. This gives ai-01/jsboige the material to ratify or reclassify per leaf.
- **Rebutted-scheme identifiability varies.** Only 1283 (self-exemption of one's own commitment) cleanly
rebuts a *named* native scheme (`Commitment_Inference`). Truthiness/parity/leveling rebut the
*general* evidenced or graded conclusion without a single named Walton scheme → ExceptionRef empty,
fail-loud. Same honesty as 1313 (`Dialogue_Scheme` identified) vs 1282 (none identified).

---

## 6. DoD status

| DoD item (#498) | Status |
|-----------------|--------|
| Effective cluster-level coverage documented (not "100% leaves") | ✅ §1/§4 (4 leaves attack-typed; sub-sub Relativisme abusif fully opened, 5/5 leaves addressed incl. anchor) |
| Per unmapped leaf: legitimate decomposition (not Latin alone) | ✅ §3 (attack-type + attacked-node + rebutted scheme where honest, 0 fabrication) |
| `AIF_skosMappingType` coherent | ✅ §4 (narrow for d5 specializations, close for d4 leveling sibling) |
| Fail-loud when no native token fits | ✅ 4/4 CA-conflict fail-loud + 3/4 rebutted-scheme fail-loud, documented in `AIF_skosOther` / annotation CSV |
| OWL regen reflects new structures | ⏳ Deferred — jsboige ratifies → CSV edits (gated) → OWL regen (#763 wiring ready) |

### Coverage accounting (code=truth)

- **Master coverage (rescan)**: **93/1408 mapped** (`AIF_attackType` filled, 6.6%).
- **This cluster (proposed)**: **+4 attack-typed rebut leaves** → 97/1408 projected once applied.
- **Rebut population**: currently **3 rebut** in prod (1282/1313/1361, all Obstruction). This cluster
would take it to **7 rebut** (+1283/1284/1285/1286) — the Relativisme abusif sub-sub becomes the
taxonomy's rebut nucleus (5 of the 7).
- **Accounting note**: rebut leaves are counted as **attack-typed** (structural), distinct from the
**strict scheme+conflict** count used for undercut/undermine clusters (this cluster adds 0 to the
strict count — the conflict layer is structural, not a native token — and 4 to the attack-typed
count). This distinction is the §5 method lesson.

**Relativisme abusif sub-sub: fully opened** (anchor 1282 merged + 4 leaves proposed = 5/5 addressed).

---

## 7. Mapping to jsboige's enriched DoD (I-node / RA-node / CA-node decomposition)

### pk 1283 — Sophisme subjectiviste → **rebut** (self-exemption)
- **I-nodes**: premise P (the shared truth, *granted* by the arguer — « fumer nuit ») ; counter-conclusion
C′ « … mais pas pour moi » opposing the conclusion C « donc cela vaut pour vous aussi ».
- **RA-node**: `Commitment_Inference` — the rebutted argument runs from the arguer's own commitment.
- **CA-node**: structural rebut (C′ conflicts with C) — no native CQ-conflict token.
- **Attack type**: **rebut** (the counter-conclusion opposes the conclusion, not a premise or the warrant).

### pk 1284 — Truthiness → **rebut** (felt counter-conclusion; boundary undermine)
- **I-nodes**: counter-conclusion C′ « X est vrai/faux parce que je le ressens » opposing the evidenced C.
- **RA-node**: *(no clean native rebutted scheme — the felt assertion does not rebut one named scheme)*.
- **CA-node**: structural rebut — no native CQ-conflict token.
- **Attack type**: **rebut**, boundary *undermine* documented (feeling substitutes for the probatory premise).

### pk 1285 — Droit à mon opinion → **rebut** (epistemic parity; boundary undercut)
- **I-nodes**: counter-conclusion C′ « mon avis vaut un fait » opposing C « le fait établi prime ».
- **RA-node**: *(no clean native rebutted scheme)*.
- **CA-node**: structural rebut — no native CQ-conflict token.
- **Attack type**: **rebut**, boundary *undercut* documented (refuses the justification-requiring warrant).

### pk 1286 — Plus faux que faux → **rebut** (false leveling; boundary false-equivalence)
- **I-nodes**: counter-conclusion C′ « les deux torts sont équivalents » opposing the graded C « le tort A
est plus grave que B ».
- **RA-node**: *(no clean native rebutted scheme)*.
- **CA-node**: structural rebut — no native CQ-conflict token.
- **Attack type**: **rebut**, boundary *false-equivalence* documented (the leveling relativizes gravity).

### Cluster attack-type distribution
| pk | RA-node (rebutted scheme) | CA-node | Attack type |
|----|---------------------------|---------|-------------|
| 1282 | *(none identified)* | structural rebut | rebut (MERGED #760) |
| 1283 | `Commitment_Inference` | structural rebut | rebut |
| 1284 | *(none)* | structural rebut | rebut (boundary undermine) |
| 1285 | *(none)* | structural rebut | rebut (boundary undercut) |
| 1286 | *(none)* | structural rebut | rebut (boundary false-equiv) |

**All 5 rows are rebuts** — the abusive-relativism mechanism *is* the opposition of a subjective/leveling
counter-conclusion. Serialization = `AIF_attackType=rebut` + `AIF_attackedNode=CA-node` for all;
`AIF_skosExceptionRef=Commitment_Inference` for 1283 only; skos otherwise empty (fail-loud, documented).

---

## Gate boundaries (HARD — proposition only)

- ❌ No prod CSV write, no DB write, no OWL regen, no `aif:` assertion emitted.
- ❌ No #674/#596/#666/#686 merge, no régén launch. No self-merge.
- ✅ Proposition derived code=truth from the taxonomy CSV (master `7796c127`) + AIF vocabulary verified
native (`Commitment_Inference` native; the coined rebut tokens verified **NOT native** and excluded
— 0 fabrication #677).
- ✅ First rebut cluster documented (attack-columns-only pattern); non-native placeholder exclusion
recorded; per-leaf boundary nuances flagged for ai-01/jsboige adjustment; machine-readable annotation
CSV `498-relativisme-abusif-annotations.csv` provided for the gated serialization.

Relates: #498 (chantier), #760 (phase 1-3 serialize — anchor 1282 + first 3 rebuts), #757 (apply/verify),
#753/#755 (serialization precedents), #763 (OWL AIF wiring), #707§4 (attack-type map option a),
#677 (0 fabrication), #499 (inverse: virtues), #133/#130 (OWL). Base master `7796c127`.
Loading
Loading