From f8e1dc162819cc177ebbf535ff6aa59bb9c9847f Mon Sep 17 00:00:00 2001 From: Your Date: Sat, 27 Jun 2026 17:02:17 +0200 Subject: [PATCH] =?UTF-8?q?docs(taxonomy):=20#600=20link=20census=20reprod?= =?UTF-8?q?ucibility=20=E2=80=94=20independent=20re-run=20byte-identical?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Re-run the full link_* langlinks census (741 Fallacies + 88 Virtues articles, probe `0 fallacies` then `0 virtues`) on master c20d5d2c — an independent confirmation of the §5.1 decision-grade figures measured 2026-06-25. Result: byte-identical. Fallacies 2739/4823 (57%), Virtues 180/322 (56%), combined ~2919, 0/829 errors, and the same link_en categorization (Fallacies 900/433/75, Virtues 185/9/29). Per-language rows match cell-for-cell. Adds a one-paragraph "Reproducibility — re-confirmed 2026-06-27" note under §5.1 stating the figures are stable and not a one-run artefact. The "named number replacing the 57% estimate" was already the measured value on master (dispatch was prepared 1 min before #600 merged §5.1 at 14:40); this PR's contribution is the second-run pin that makes it reproducibly-confirmed. Read-only. 0 write under Cards/. Dispatch ai-01 2026-06-27 tertiaire (#600 census link_* complet). Co-Authored-By: Claude-Code --- docs/taxonomy/192-link-coverage-research.md | 2 ++ 1 file changed, 2 insertions(+) diff --git a/docs/taxonomy/192-link-coverage-research.md b/docs/taxonomy/192-link-coverage-research.md index edcc36a2..c4b7e5e3 100644 --- a/docs/taxonomy/192-link-coverage-research.md +++ b/docs/taxonomy/192-link-coverage-research.md @@ -93,6 +93,8 @@ The §5 ceilings were theoretical. This section **measures** the real resolvable Script: [`192-link-coverage-langlinks-probe.py`](192-link-coverage-langlinks-probe.py) — read-only, no API key, ~0.3 s throttle, descriptive User-Agent (MediaWiki 403s the default urllib UA). Census run = 0 errors on 741 (Fallacies) + 88 (Virtues) articles. +**Reproducibility — re-confirmed 2026-06-27 (po-2024, master `c20d5d2c`).** An independent full re-run of both datasets (same probe, `0 fallacies` then `0 virtues`, 741 + 88 articles) returned **byte-identical** figures: Fallacies 2 739 / 4 823 (57 %), Virtues 180 / 322 (56 %), 0 errors, and the same `link_en` categorization (Fallacies 900 Wikipedia · 433 non-Wikipedia · 75 empty; Virtues 185 · 9 · 29). The per-language rows match cell-for-cell. The §5.1 numbers are therefore stable and decision-grade — not a one-run artefact or a mislabeled sample. + ### Fallacies — 1 408 rows | `link_` | candidate cells missing | confirmed resolvable | rate |