Current-Country Evidence Audit: Cross-Batch Disagreement, Asymmetry, and Unresolved Review

Status

Reviewed with limitations · model-generated source audit · publication blocked · exact audit cutoff 2026-07-23T10:53:59Z.

Purpose

Compare the first evidence-audit burden across Senegal, Trinidad and Tobago, Nepal, Slovakia, and Kiribati without ranking the jurisdictions or treating equal structure as equal conditions.

Scope

The comparison covers sixty preserved 2.0.22 current observations, their renewed source checks, disposition records, contradiction searches, corrections, qualifications, source-language notes, and unanswered human and locally informed review packets.

Executive Summary

• 60 claims received claim-level audit records and contradiction-search fields.
• 43 were corroborated without material wording change.
• 14 were retained with qualification.
• 3 require correction before reuse.
• 0 claims were marked contradicted, superseded, or unresolved in this source-audit pass.
• 0 independent human reviews and 0 locally informed responses are documented.

The absence of a contradiction disposition is not evidence of agreement. It records only that this bounded public-source pass did not establish a defensible contradiction or supersession for those claims.

Evidence Reviewed

• Five original proposal records and sixty original source records from 2.0.22.
• Five audit records, sixty audit source records, five source registers, five unanswered review packets, one correction register, and this cross-batch record.
• Current primary, authoritative, locally authored, local-language, oversight, remedy, regional, technical, and affected-community sources where feasible.
• Eleven preserved information-environment packets were available only as archival question generators; none was silently used to corroborate or amend a country claim.

Findings

Disposition comparison

Evidence disagreement

The audit preserved the clearest located disagreement instead of flattening it: PASTEF's stated non-participation in Senegal's new government is distinct from reporting that several announced ministers were affiliated with PASTEF. The record treats the first as an attributable party position and the second as a cabinet-affiliation question requiring direct party and ministerial evidence.

No other claim met the audit's threshold for a contradiction disposition. Several claims received qualifications because an official route established formal authority but did not establish implementation, field performance, public opinion, remedy outcomes, final certification, complete data coverage, or legal effective dates.

Corrections

Three corrections are append-only:

1. Normalize Senegal's Prime Minister name to the official current spelling and keep alternate spellings as search aliases only.
2. Separate PASTEF's stated cabinet position from reported cabinet affiliations.
3. Replace Kiribati's unverified affordability-objective wording with documented consultation and implementation questions.

Comparative fairness review

Every jurisdiction received twelve claim audits, exactly two under each lens; twelve contradiction searches; twelve source records; twelve human-review questions; twelve locally informed prompts; identical reviewer-attribution, conflict, consent, and publication-scope fields; and the same publication blockers. Comparable structure does not erase differences in public-source availability or justify equal conclusions.

The audit compared adjective intensity, threat framing, rights, oversight, remedy, uncertainty, correction burden, source-language diversity, local authorship, affected-community visibility, and unresolved-review burden. No country score, rank, stability label, threat label, loyalty label, moral label, or alignment score was created.

Evidence asymmetry

Official English or French pages were easier to retrieve than some local-language, subnational, oral, community, court, and implementation records. Transparent institutions can appear more failure-prone because they publish disputes, audits, corrections, and complaints. Opaque or poorly indexed systems may produce fewer visible discrepancies without being more accurate or effective.

Unresolved human and local review

Every supplied review packet remains unanswered. Public-source collection cannot determine whether wording is locally intelligible, whether institutional terminology maps to practice, which communities were excluded, whether a remedy is usable, or whether a reviewer consents to quotation. Those questions remain publication blockers.

Separation of duties

• Deterministic checks test repository structure.
• Model-generated source checks test public-source support and identify discrepancies.
• Independent human reviewers must judge materiality, legal and political interpretation, and publication fitness.
• Locally informed reviewers must test language, implementation, missing perspectives, and permission boundaries.

Decisions or Recommendations

1. Keep all five profiles at reviewed-with-limitations or draft status.
2. Apply the three corrections before reuse, while preserving the original records.
3. Preserve every qualification and resolution test in public and machine outputs.
4. Seek independent and locally informed review through the unanswered packets; never simulate responses.
5. Repeat the source audit whenever a volatile claim changes or before any publication decision.

Risks and Limitations

A passing source-audit gate proves neither correctness nor neutrality. Automated checks, model-generated review, and public-source collection do not establish human comprehension, exhaustive local representation, independent human review, consultation or consent, geopolitical neutrality, legal or scientific validity, native Firefox behavior, native assistive-technology speech output, production-host readiness, field performance, or live search and crawler outcomes.

Validation Performed

The validator checks all sixty claim records, all six lenses in every jurisdiction, exact cutoff, disposition and correction counts, source metadata, local-review packet emptiness, equal burden, reciprocal memory links, public routes, schemas, and prohibited ranking or operational content.

Memory References

• Active audit memory
• Pointer ledger entry
• Original batch memory

Related Durable Documents

• [Balanced-batch selection and comparative-fairness report](current-country-proposals-balanced-batch-selection-and-comparative-fairness.md#findings)
• [Country Context Composer and comparative-fairness report](country-context-composer-and-comparative-fairness-report.md#findings)
• [Information-environment packet custody and reuse boundary](information-environment-research-packet-intake-provenance-and-reuse-boundary.md#findings)
• [World Systems governance, resilience, and evidence integration report](world-systems-governance-resilience-and-evidence-integration-report.md#findings)

Supersession Status

First model-generated source audit of the five 2.0.22 proposals. It supersedes no proposal and authorizes no publication. Later audits must append reciprocal supersession and correction links rather than overwrite this record.