Skip to content

research: add supporting and dissenting evidence contact for six Rights Article Labs - #19

Merged
erinepshovel-code merged 3 commits into
mainfrom
claude/textbook-chapters-needed-jd2nht
Jul 28, 2026
Merged

erinepshovel-code merged 3 commits into
mainfrom
claude/textbook-chapters-needed-jd2nht

Conversation

@erinepshovel-code

@erinepshovel-code erinepshovel-code commented Jul 28, 2026

Copy link
Copy Markdown
Contributor

Two changes on this branch:

1. Article Lab research contact (sources.yml + claims.yml)

Adds reviewed sources and bounded claims to the Article Lab research ledgers. Before this change, Articles One, Five, Six, Seven, and Eight had no dissent contact and Article Four had only a single "mixed" claim; the lab's method requires missing support/dissent to surface as documented gaps, so this fills them with real, verified literature.

New dissent contact

  • Article One — Pattaro et al. 2022 (J. Social Policy, scoping review): benefit sanctions produce worse jobs, hardship, health harms, poorer child wellbeing → risk reading of the exclusion clause.
  • Article Four — WHO 2002 World report on violence and health: violence as preventable public-health harm → weighs against the "honestly violent" footnote.
  • Article Five — Icenogle et al. 2019 (Law & Human Behavior): the cognitive/psychosocial "maturity gap" → empirical basis for age-based lines the article replaces.
  • Article Seven — Sunstein 2002 (J. Political Philosophy): group polarization → definitional juries may diverge, not converge.
  • Article Eight — Harcourt 1999 (J. Crim. L. & Criminology): the harm principle's indeterminacy → "harms none" is not self-executing.

New support contact

  • Article One — Fehr & Gächter 2000 (AER): costly punishment of free riders sustains cooperation.
  • Article Four — Goodwin & Maru 2017 (Hague J. on the Rule of Law): legal-empowerment evidence for teaching law to laypeople.

All bibliographic metadata (DOIs, volumes, pages) verified against publisher records. Every source carries reviewed_on: 2026-07-28, relevance, and an explicit boundary; every claim carries a stance, status, and limitation per docs/research-method.md. Stance classification is editorial review, as the method requires.

Merge note: main's research deepening (#16) landed concurrently and independently added the same Stephan & Chenoweth 2008 source with an equivalent article-six-nonviolent-power dissent claim. The merge keeps main's versions of those two records and this branch's seven other sources and seven other claims — no duplicate ids, no duplicate DOIs.

2. Canon verifier: footnote-marker normalization (scripts/verify-article-canon.mjs)

The article pages quote canon with superscript footnote markers (¹ ² ³) while the generated canon.json uses bracketed [n] markers; the verifier treated this presentation difference as drift. It now strips both marker notations from both sides before comparison, stops canonical-body extraction at notes blocks and --- separators, and reports all drift findings in one run instead of throwing on the first.

Deliberately not fixed: with markers normalized, 17 genuine word-level drift findings remain across Articles Two–Eight (excerpt drift on Three, Five, Six, Seven, Eight; canon notes missing from several pages). The pages track the upstream canon at wayseer00/main:canon/INTERDEPENDENT_WAY.txt, which this build environment cannot reach (403 — outside session scope), so refresh:canon falls back to the older local mirror. These findings stay visible for a maintainer-side canon refresh; per maintainer decision, they are not papered over here.

Testing: npm test — 20/20 pass after merging main (includes the research-ledger validation and main's new math/textbook suites). npm run build / test:generated still fail on the 17 remaining genuine canon-drift findings above (pre-existing; the verifier now reports them completely and accurately).

🤖 Generated with Claude Code

https://claude.ai/code/session_01E93Rxmrs8bXHtWjD7z94bi

claude added 3 commits July 28, 2026 02:59
…ts Article Labs

Adds eight reviewed sources and eight bounded claims to the Article Lab research
ledgers, filling the dissent gaps on Articles One, Four, Five, Six, Seven, and
Eight and deepening support on One and Four. Every record carries an explicit
evidence boundary per docs/research-method.md; stances are editorial review.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01E93Rxmrs8bXHtWjD7z94bi
…rifier

Superscript digits and bracketed [n] markers are presentation, not canon words;
strip both from the canonical body and the page excerpt before comparison, stop
body extraction at notes blocks and separators, and report every drift finding
in one run instead of throwing on the first. Word-level canon drift still fails:
17 findings remain visible across Articles Two-Eight pending a canon refresh
from wayseer00/main, which this environment cannot reach.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01E93Rxmrs8bXHtWjD7z94bi
Main's research deepening (#16) and this branch both extended the ledgers.
Resolution keeps main's records and appends only this branch's non-duplicative
additions: the shared Stephan & Chenoweth 2008 source and the equivalent
article-six-nonviolent-power dissent claim keep main's versions; the branch's
seven other sources and seven other claims are retained.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01E93Rxmrs8bXHtWjD7z94bi
@erinepshovel-code
erinepshovel-code marked this pull request as ready for review July 28, 2026 19:40
@erinepshovel-code
erinepshovel-code merged commit 63b64dd into main Jul 28, 2026
5 checks passed
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