Deepen Article Lab research evidence

[erinepshovel-code]

Purpose

Continues the balanced Article Lab research program after PR #15 by adding systematic reviews, meta-analysis, replication disputes, operational evidence, and explicit evidentiary limits across all eight Rights Articles.

This tranche tests bounded components of each Article rather than treating a paper as assenting to or dissenting from an Article as a whole.

What changed

Research evidence

Adds 11 reviewed sources and 9 claim records:

• Article One — mixed: humanitarian cash, vouchers, and in-kind food can each support food security; modality choice remains context- and outcome-dependent.
• Article Two — support: a 184-dataset meta-analysis strengthens the autonomy-support and autonomous-motivation component.
• Article Two — dissent: a caregiver meta-analysis limits any reading that allocates survival duties as though unpaid caregivers have unlimited capacity.
• Article Three — mixed: direct reciprocity can sustain cooperation, while strict act-for-act retaliation is error-sensitive and vulnerable to cross-relationship spillover.
• Article Four — mixed: plain-language redrafting improves legal-contract comprehension but does not eliminate conceptual misunderstanding.
• Article Five — limit: decision-making capacity is ordinarily task-specific and dynamic rather than a sound basis for global status removal.
• Article Six — mixed: group-performance experiments support a collective factor, but replication work disputes whether interaction patterns or member cognition best explain it.
• Article Seven — mixed: standardized professional terminology can support coordination and outcomes, but the causal evidence remains low-certainty and domain-specific.
• Article Eight — support: a 209-trial Cochrane review supports option-explicit, probability-bearing, values-sensitive decision aids.

Durable ledger growth

• adds src/_data/research/ledger-manifest.json
• moves new work into date-bounded source and claim tranche files rather than extending one permanent monolith
• updates the Eleventy data adapter to load and flatten allowlisted manifest-declared YAML ledgers
• rejects duplicate, empty, malformed, or path-escaping ledger declarations
• expands the test gate across all declared ledgers
• raises the executable evidence floor to 32 reviewed sources, 30 bounded claims, and at least three claims per Rights Article while retaining required support and dissent

Evidence and authority boundary

The canonical text remains wayseer00/main:canon/INTERDEPENDENT_WAY.txt. Research cards do not amend or supersede canon.

Each claim records its source type and limitation. Systematic reviews, experiments, formal models, legal or medicolegal analysis, and replication disputes retain different evidentiary force. Conflicting high-quality results are preserved rather than averaged into false certainty.

The manifest is repository-owned and accepts only unique .yml paths beneath src/_data/research; it cannot point the build at arbitrary files or network locations.

Validation

• branch is five commits ahead of merged PR Add balanced Article Lab research #15 with no base divergence
• diff is confined to three new research-ledger files, the research data adapter, and the research-ledger test
• manifest flattening produced 32 sources and 30 claims
• GitHub Actions Verify generated site passed on head f78872d13f9f261aa4929ca461379e2d88519611
• the green gate includes source parsing, provenance and content validation, Eleventy generation, Pagefind, unit and generated-site tests, Chromium browser checks, and accessibility checks

hmmm

This tranche improves depth and preserves disagreement, but it is still not a comprehensive literature review. Further passes should add newer replications, non-English and jurisdictionally diverse evidence, implementation failures, and source-by-source correction or retraction monitoring.