docs(translation): #202 QA-gate protocol — per-tranche spot-check criteria (docs-only) - #810
Merged
Merged
Conversation
…teria (docs-only) Companion to 202-en-campaign-plan.md (#809). Defines the byte-exact spot-check criteria ai-01 applies per pilot tranche before a bulk GO, so campaign execution is turnkey once the register decision (A/B) lands. 6 checks per 20-cell pilot: 1. Empty-only invariant (byte-exact, HARD — 0 clobber) 2. Meaning fidelity vs text_fr/desc_fr (semantic, sample >=5) 3. Register/style (field-specific: Simple_name <=5 words; political_example neutral/non-defamatory, matches ratified A/B) 4. MT-garbage sweep (3-dim, per-token counts — memo false-zero) 5. Language purity (0 FR/RU leakage in new outputs) 6. Encoding/round-trip (CRLF+BOM, no quoting drift — memo byte-exact-insertion) GO/REVISE/NO-GO thresholds; hard gates = clobber + encoding drift. Post-bulk regression gate: re-run #795 drift audit + byte-check ALL lang columns (zh #761 lesson) + rendered-source freshness (clobber harvests). #202 execution stays deferred post-v0.9.0 (jsboige arbitration). Docs-only, 0 CSV write, 0 code change. Dispatch mf4nyb [SECONDAIRE]. 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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What
Docs-only companion to #809 — #202 EN campaign plan: the per-tranche QA-gate protocol ai-01 will apply when the campaign executes (post-v0.9.0). 0 CSV write, 0 code change.
Why
The campaign (#809) fills ~2725 empty cells via gpt-5.5. gpt-5.5 can produce fluent-but-wrong output. A pilot tranche (20 cells) is QA'd before each bulk run. This doc pins what ai-01 checks so the bar is uniform — turnkey once the register decision (A/B) lands.
Protocol summary (full detail in the doc)
6 checks per 20-cell pilot:
text_fr/desc_fr(semantic, sample ≥5)Simple_name_en≤5 words;political_example_enneutral/non-defamatory, matches ratified A/Bmt-garbage-sweep-false-zero)csv-byte-exact-column-insertion)GO/REVISE/NO-GO thresholds. Hard gates (auto-NO-GO) = clobber + encoding drift.
Post-bulk regression gate: re-run #795 drift audit + byte-check ALL lang columns (zh #761 lesson) + rendered-source freshness (clobber harvests before regen — memo
regen-success-without-clobber-is-stale-trap).Scope
link_*(feat(translations): multi-pass quality improvement for EN/RU/PT (beyond initial translation) #192), chore(mindmap): residual i18n polish — links.svg still point to fr.wikipedia + no per-lang HTML wrappers (post-tag, non-blocking) #804 Phase 4, chore(repo): réduire le poids du dépôt (.git 2,2 GB) — binaires de build commités dans l'historique #415 (INTERDICTED).Verification
Docs-only — no build/test impact. Grounded in the #809 plan structure (5 tranches) and the standing byte-exact/encoding memos.
— po-2024 (dispatch
mf4nyb[SECONDAIRE])