Skip to content

data(rules): FR clarity pass gpt-5.5 (19/24) - #366

Merged
jsboige merged 1 commit into
masterfrom
data/rules-fr-clarity
May 28, 2026
Merged

data(rules): FR clarity pass gpt-5.5 (19/24)#366
jsboige merged 1 commit into
masterfrom
data/rules-fr-clarity

Conversation

@jsboige

@jsboige jsboige commented May 28, 2026

Copy link
Copy Markdown
Contributor

Summary

  • FR clarity review of 24 Rules records using gpt-5.5 (Category B, SelectEmptyTargets=false)
  • 19/24 Rules improved: grammar fixes, consistent "vous" form, punctuation, conciseness
  • 5 Rules unchanged: Rules_01, Rules_07, Rules_10, Rules_15, Rules_20
  • Cascade drift remap (EN/RU/PT/ES/AR/FA/ZH) ran → 0 cells changed (translations already aligned)
  • Added DatasetUpdater prompt files for Rules FR clarity + cascade tasks

Cascade convergence

Rules cascade drift remap (7 languages × 24 records = 168 cells) produced zero changes — all translations were already consistent with the revised FR text. No cascade PR needed.

Test plan

  • Visual inspection of 19 revised FR rules (grammar, "vous" form, game mechanics preserved)
  • Verify unchanged Rules (01, 07, 10, 15, 20) match master exactly

🤖 Generated with Claude Code

Category B cleanup pass on French game rules text using gpt-5.5.
19 of 24 rules received clarity improvements: grammar fixes,
consistent "vous" form, cleaner punctuation, simplified phrasing.
5 rules unchanged (already clear).

Adds Rules FR clarity task config (disabled) and prompt files.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

@clusterManager-Myia clusterManager-Myia left a comment

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

[NanoClaw] Review: Argumentum#366 — FR clarity pass (19/24 rules)

Scope: 4 files — CSV rules text, DatasetUpdater config, 2 new prompt files.

CSV changes (Rules_02 through Rules_24):

  • Grammar and elision fixes throughout: le embromadorl’embromador, voulusouhaité, scénario accent fix
  • Consistent « vous » form across all rules (was mixed on/impersonal)
  • Trailing whitespace and double spaces cleaned in section headers (### 1. ### 1. )
  • Bullet punctuation normalized (trailing periods)
  • carte d’argument fallacieuxcartes d’arguments fallacieux (plural agreement)
  • Missing periods at end of sentences added

Notable fix: Rules_13 had qu’il ne reste plus qu’une personne est en jeu (extra est) → corrected to qu’il ne reste plus qu’une personne en jeu

New files: Prompt templates for GPT-5.5 FR clarity review task — well-scoped, preserve game mechanics.

Config: New DatasetUpdater entry, Enabled = false, sequential chunks. Clean addition.

Observation: CSV now starts with a BOM (\ufeff). This is common for Excel compatibility but worth noting if any downstream parsers are strict about it.

Security: Clean. No tokens, credentials, or secrets.

Verdict: Solid mechanical clarity pass. Game mechanics preserved, translations untouched as stated.

@clusterManager-Myia clusterManager-Myia left a comment

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

[NanoClaw]

Clean clarity pass on 19/24 FR rules records. The changes are consistent: proper French elision ("le embromador" → "l’embromador"), uniform "vous" imperative, tighter punctuation, trimmed whitespace in headers. No game mechanics altered, no translation columns touched — matches the zero-drift cascade claim.

Minor notes:

  • Rules_13: "carte Scénario" is capitalized mid-sentence, whereas all other rules use lowercase "carte scénario" / "carte de scénario". Probably worth normalizing.
  • The BOM () added to the CSV header is harmless but worth confirming downstream consumers handle it (some parsers choke on it).
  • Prompt files and DatasetUpdater config look fine. The config is Enabled = false, so no risk of accidental execution.

No secrets or sensitive data in the diff.

@clusterManager-Myia clusterManager-Myia left a comment

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

[NanoClaw] — Review: Argumentum#366 — FR clarity pass gpt-5.5

Summary

Clean clarity pass on 19/24 French rules records. Consistent improvements: proper elision (le embromador → l'embromador), uniform imperative vous form (was mixed on/impersonal), trimmed whitespace in section headers, tightened punctuation, and lighter sentence structures. Game mechanics are preserved throughout. No translation columns touched (cascade confirmed zero drift). Two prompt files added for the DatasetUpdater pipeline, plus the matching config entry (disabled by default).

Findings

Consistency issue — grammatical agreement on organisé vs organisées:

  • Rules_02, Rules_11, Rules_16: 1 paquet de cartes d'arguments fallacieux, organisées en 7 classes (agrees with cartes, feminine plural)
  • Rules_21: 1 paquet de cartes d'arguments fallacieux, organisé en 7 classes (agrees with paquet, masculine singular)
    Both agreements are valid in French depending on what you're qualifying, but having both in the same document is inconsistent. Worth picking one convention and applying it uniformly.

Minor — Rules_13 mid-sentence capitalization:
Le piocheur tire une carte Scénario parmi les deux pioches disponibles — Scénario is capitalized mid-sentence here, whereas all other rules use lowercase scénario / carte scénario / carte de scénario.

Minor — Rules_23 section renumbering:
The old duplicate ### 2. (Les enchères d'atout / Les tours de jeu) is now correctly ### 2. → ### 3. This aligns with Rules_24 now being ### 4. Décompte de la manche — internally consistent.

Positive notes:

  • embromador terminology consistently handled with proper French elision everywhere
  • Imperative vous form makes instructional rules much more direct and readable
  • Rules_06 variants section: long run-on sentences split into cleaner bullet points
  • Rules_23 coinche explanation significantly clearer (separated auction mechanics from game phase)

Security

No secrets or credentials found. TOKENS matches are C# config property names (MaxOutputTokens, MaxTokensPerMinute) — false positives.


Reviewed by NanoClaw (ClusterManager) — $(date -u +%Y-%m-%dT%H:%MZ)

@clusterManager-Myia clusterManager-Myia left a comment

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

[NanoClaw] review of #366

Scope: 4 files — CSV rules data (1), DatasetUpdater config (1), 2 new prompt templates.

Count verification: 19/24 rules modified, matching the title. Unchanged: Rules_01 (content unchanged; BOM-only header edit), Rules_07, Rules_10, Rules_15, Rules_20. Confirmed: all 7 translation columns (EN/RU/PT/ES/AR/FA/ZH) untouched — changes land exclusively in the French Text column, consistent with zero-drift cascade claim.

Changes verified as improvements:

  • French elision: le embromadorl'embromador throughout (Rules_04–06, 18–19, 23–24). Correct.
  • Imperative shift: on constitue/choisitconstituez/choisissez (vous form, Rules_03, 12, 17). Consistent with prompt instructions.
  • Punctuation: trailing periods on bullet items, semicolons joining clauses, trimmed whitespace in ### N. headers. Cleaner.
  • Plural agreement: carte d'argument fallacieuxcartes d'arguments fallacieux (Rules_02, 08, 11, 16, 21). Correct plural.
  • Grammar: voulusouhaité, débarasserdébarrasser, courcours (typo fix in Rules_06).
  • Bug fix: Rules_13 — qu'il ne reste plus qu'une personne est en jeu...qu'une personne en jeu (spurious est removed). Good catch.
  • Conciseness: Rules_02, 21 — long run-on sentences broken into shorter, clearer ones. Rules_23 — list formatting improved.

Minor observations:

  • Rules_13: carte Scénario capitalized mid-sentence; all other rules use lowercase carte scénario. Cosmetic inconsistency.
  • BOM () added to CSV header. Harmless for Excel/standard parsers, but strict parsers may need updating.
  • Prompt files are well-scoped and correctly instruct to preserve game mechanics and terminology. Config entry has Enabled = false.

Security: Clean. No tokens, credentials, or secrets.

File types: CSV data + C# config + 2 text prompts. No executable code logic changes.

Verdict: Solid mechanical clarity pass. All changes are genuine improvements with no game mechanics altered and no translation drift.

@jsboige
jsboige merged commit ac3e2c5 into master May 28, 2026
3 checks passed
@jsboige
jsboige deleted the data/rules-fr-clarity branch May 28, 2026 14:18
jsboige added a commit that referenced this pull request May 28, 2026
…ules) (#372)

Smoke test of "Rules cascade EN-only gpt-5.5" task introduced in PR #371,
validating the max-context narrow-edit approach (FR readonly anchor,
EN editable target).

ChunkSize=8, TakeChunkNb=1 → processed rules 1-8, LLM made 7 surgical edits
on rule "L'École des Menteurs" (rule 1 of the deck), aligning EN to the
refined FR (PR #366):

- Components/Game overview/Setup/Round sequence/The count: section headers
  rephrased to match FR clarity (was: Game material/Game summary/Installation/
  Course of a round/The points count)
- "smooth talker" → "embromador" (proper game term, consistent with FR)
- Added missing emoji codes in win/tie conditions (🥇👇 🥈👇 ✅🎭 etc.)
- Filled in 8 missing "Draws" subsection bullets (winner draws 0, jurors
  draw 1, embromador's-card-finders draw 1 extra...)
- Filled in 6 missing variant bullets (skit interventions, drawer-as-winner,
  multi-card embromador, final-scenario rule)
- Title alignment: "The school of liars" → "Argumentative mixology bingo"
  (matches FR "Le Bingo mixologie argumentative" — was a leftover from a
  prior rule variant)

API cost: ~$0.50, duration 36s, 0 errors, 0 quota issues.
Validates EN-only cascade pattern for follow-up smoke tests on Scenarii/
Virtues/Fallacies.

Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
jsboige added a commit that referenced this pull request May 29, 2026
…6.1.1.2)

Cell-by-cell triptych review of 6 desc_fr cells modified by the Phase 4
FR clarity pass (PRs #366-#369). Verdict: 3 baseline restores, 3 Phase 4
keeps (no-op), 0 AI proposals accepted.

REVERT to baseline f38cfed (Phase 4 collapsed scope toward one child):
  - PK 7.3       Ad hominem       : restore "Vous visez votre adversaire
                                    lui-même..." (covers 3 children, not
                                    just Attaque personnelle)
  - PK 3.1.3     Transfert illicite : restore "...ou l'inverse" (preserves
                                    composition⇄division duality)
  - PK 6.1.1.2   Factoïde         : restore "faits largement acceptés mais
                                    inexacts" (covers Désinformation/Rumeur/
                                    Mythe/Mème, not just Rumeur)

KEEP Phase 4 (genuine improvements):
  - PK 4.2.3     Erreur de modalité : accessible vocabulary
  - PK 4.3.1.1   Prémisses non pertinentes : fixes typo prémices→prémisses
  - PK 3.3       Conclusion math invalide : singulier + concis

Methodology: tmp/runs/triptych.py displays baseline / Phase 4 master / HEAD
side-by-side per PK with word counts and direct-children labels for scope
verification. Validated by jsboige ("OK pour la répartition, tu peux rester
sur cette ligne pour le reste") — to be extended to full Phase 4 Fallacies
audit (~1650 cells / ~955 PKs) via cluster review.

Surgical binary patch via tmp/runs/revert_cells.py: 3 insertions / 3
deletions. CRLF + quoting preserved. No collateral.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
jsboige added a commit that referenced this pull request May 29, 2026
…6.1.1.2) (#387)

Cell-by-cell triptych review of 6 desc_fr cells modified by the Phase 4
FR clarity pass (PRs #366-#369). Verdict: 3 baseline restores, 3 Phase 4
keeps (no-op), 0 AI proposals accepted.

REVERT to baseline f38cfed (Phase 4 collapsed scope toward one child):
  - PK 7.3       Ad hominem       : restore "Vous visez votre adversaire
                                    lui-même..." (covers 3 children, not
                                    just Attaque personnelle)
  - PK 3.1.3     Transfert illicite : restore "...ou l'inverse" (preserves
                                    composition⇄division duality)
  - PK 6.1.1.2   Factoïde         : restore "faits largement acceptés mais
                                    inexacts" (covers Désinformation/Rumeur/
                                    Mythe/Mème, not just Rumeur)

KEEP Phase 4 (genuine improvements):
  - PK 4.2.3     Erreur de modalité : accessible vocabulary
  - PK 4.3.1.1   Prémisses non pertinentes : fixes typo prémices→prémisses
  - PK 3.3       Conclusion math invalide : singulier + concis

Methodology: tmp/runs/triptych.py displays baseline / Phase 4 master / HEAD
side-by-side per PK with word counts and direct-children labels for scope
verification. Validated by jsboige ("OK pour la répartition, tu peux rester
sur cette ligne pour le reste") — to be extended to full Phase 4 Fallacies
audit (~1650 cells / ~955 PKs) via cluster review.

Surgical binary patch via tmp/runs/revert_cells.py: 3 insertions / 3
deletions. CRLF + quoting preserved. No collateral.

Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
jsboige added a commit that referenced this pull request Jul 2, 2026
… PT/RU + glossaire verrouillé (#633) (#640)

Corrige les 23 findings HIGH du dataset Rules (audit ai-01 2026-07-02) :
vieille MT ~2021 jamais remplacée (Text_pt/Text_ru) + contamination LLM
« embromador » (mot PT) injectée dans FR/EN/ES par les passes #366/#372/#376.

Cards/Rules/Argumentum Rules - Cards.csv (15 records) :
- Text_pt : retraduction intégrale depuis FR (gpt-5.5 Responses API)
- Text_ru : Rules_07 titre (Бинго аргументативной миксологии), Rules_12
  formatage headers, Rules_15 retraduction (header vide + phrase FR incrustée),
  « Ход партии » → « Ход раунда » ×3 (cohérence раунд)
- Text_en : Rules_09 « The Last Smooth Talker » (beau parleur), Rules_10
  « rounds » (manche ≠ sleeves)
- Purge embromador FR→baratineur / EN→smooth talker / ES→embaucador (11 cells,
  PT conservé — embromador = canon PT du rôle baratineur)

Cards/Rules/Argumentum Rules - Cards Print and Play.csv (6 records) :
- Text_pt + Text_ru : retraduction intégrale (PP divergeait silencieusement du
  principal ; couverture #306 reportée : « A escola dos mentirosos »)

Taxonomy (2 cellules MEDIUM) :
- Fallacies PK615 example_ar : russe (copie example_ru) → arabe MSA
- Virtues pk71 description_ru : fuite française → russe

Prévention récidive — glossaire verrouillé dans les prompts DatasetUpdater :
- PromptRulesFrClarityUser.txt : « embromador » listé à tort comme terme FR →
  corrigé en canonique « baratineur » + directive anti-contamination étrangère
- PromptRulesCascadeDriftUser.txt : ligne « terminologie incohérente
  (embromador, baratineur) » induisait l'harmonisation → remplacée par le
  glossaire canonique multi-langue (piocheur/baratineur/manche/pli/atout/saynète)

Scanner (scan_translations.py) : 22 findings → 7 (les 7 résiduels = faux
positifs documentés : noms latins canoniques Credo quia absurdum, emprunts
Gish Gallop/Creepypasta/Whataboutism, mnémoniques Celarent/Camestres, nom
propre Johnny Johnny). 0 HIGH Rules, 0 MEDIUM taxonomy résiduel.

NOTE : GDrive (source autoritative, item #6 du scope) NON mis à jour — upload
gated jsboige. Sans sync GDrive, le prochain pull baseline réimportera le
garbage (cause racine du clobber #211 par 7fae40c).

Co-authored-by: Your <your.email@example.com>
Co-authored-by: Claude-Code <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants