# Synthetic Consent Comprehension Fixtures: Accessibility, Translation, and Fail-Closed Validation

## Status and scope

**Release:** 2.0.28-wip  
**Generated UTC:** 2026-07-24T16:43:00Z  
**Synthetic task definitions:** 12  
**Synthetic response fixtures:** 6  
**Translation drafts:** 3  
**Real participants:** 0  
**Human comprehension results:** 0  

## Findings

The fixture set demonstrates clear, overbroad, contradictory, privacy-leaking, deletion-confused, and translation-uncertain records. Each outcome applies to the record and interface only. No person-level score, pass grade, capacity judgment, susceptibility profile, credibility ranking, reviewer-worth measure, or community-representativeness claim is created.

All nine permissions receive the same structural burden: original wording, plain-language alternative, authorization boundary, non-authorization boundary, legal-review blocker, accessibility-review blocker, one or more comprehension tasks, and disclosure consequences. Differences in issue state follow only from the synthetic answer or field state.

The Spanish, French, and Arabic examples are visibly model-generated drafts for layout and protocol testing. Original English wording remains present, script direction is declared, uncertainty is recorded, and no legal or local equivalence is claimed.

Disclosure fixtures cover included, excluded, unresolved, withdrawn, contradictory, and blocked fields. Private identity and verification fields remain excluded by default. An allowed publication choice does not override missing quotation, paraphrase, attribution, identity, safety, source-verification, adjudication, or scope requirements.

## Accessibility review boundary

Static and managed-browser checks can verify DOM order, labels, keyboard operation, focus visibility, reflow, reduced-motion declarations, right-to-left rendering containers, and no unexpected network traffic. They cannot establish screen-reader speech quality, cognitive comprehension, sign-language access, translation adequacy, usability for representative disabled users, or suitability in low-bandwidth field conditions.

## Fail-closed behavior

No automatic pass is calculated. A record is either clear for demonstration, needs clarification, blocked, withdrawn, or unresolved. The lab shows the relevant issue category and expected concept, preserves the earlier state, and requires an explicit new response before any operational use. Even a clear synthetic demonstration remains non-publishable.

## Memory References

- Active summary: [`.uai/country-context-consent-evaluation.uai`](../../../.uai/country-context-consent-evaluation.uai#current-publication-contract)
- Pointer ledger: [`.uai/long-term-memory.uai`](../../../.uai/long-term-memory.uai#pointer-records)
- Protocol report: [Consent Language and Disclosure Comprehension Evaluation Protocol](country-context-consent-language-and-disclosure-comprehension-evaluation-protocol.md#findings)

## Related Durable Documents

- [Reviewer Submission Readiness Workspace](country-context-reviewer-submission-readiness-workspace-privacy-consent-redaction-and-portable-export.md#findings)
- [Reviewer Evidence Intake, Consent, Adjudication, and Publication Boundary](country-context-reviewer-evidence-intake-consent-adjudication-and-publication-boundary.md#findings)
- [Information-Environment Packet Custody and Reuse Boundary](information-environment-research-packet-intake-provenance-and-reuse-boundary.md#findings)

## Supersession Status

Current for the 2.0.28-wip deterministic fixture set. It supersedes no real participant evidence, consent, locally informed review, accessibility finding, translation approval, legal advice, adjudication, or publication decision.