Status and scope
Release: 2.0.28-wip Generated UTC: 2026-07-24T16:43:00Z Review state: Current implementation protocol with WIP limitations Real participants: 0 Human comprehension results: 0
This protocol defines a controlled, browser-local method for examining consent wording and disclosure previews with synthetic tasks only. It does not recruit, contact, identify, profile, or score reviewers. It does not establish informed consent, legal sufficiency, human comprehension, locally informed review, representative accessibility, or publication approval.
Findings
- Nine independent permission records preserve original wording, a plain-language alternative, what the permission authorizes, what it does not authorize, version rationale, and explicit unresolved legal and accessibility review states.
- Twelve synthetic comprehension tasks test permission boundaries, private/public disclosure, lifecycle rights, and translation uncertainty without producing a numeric pass score.
- Twelve non-numeric rubrics record expected concepts, misunderstanding patterns, issue categories, and an explicit clarification rule.
- Record-level issue classifications replace person-level scoring. They identify authorization overreach, unresolved or contradictory permission, privacy-boundary errors, lifecycle-rights confusion, translation uncertainty, accessibility review needs, and legal-governance questions.
- Spanish, French, and Arabic drafts preserve the English source, translator role, translation method, script direction, uncertainties, and version history. They are model-generated interface fixtures and are not human-reviewed translations.
- Disclosure outcomes make included, excluded, unresolved, withdrawn, contradictory, and blocked fields explicit before any local export.
- Withdrawal, consent change, correction, appeal, deletion request, and retention questions remain separate append-only events and unresolved governance duties.
- The workspace preserves original imported bytes, unknown extensions, and unchanged quarantine, and performs no network calls.
Consent-language inventory
The nine choices remain separate: participation, quotation, paraphrase, attribution, private retention, anonymization, publication, withdrawal contact, and deletion-request handling. No single permission authorizes another. Missing, withdrawn, conditional, or contradictory permission fails closed for the affected use.
Legal and governance questions requiring qualified review
The protocol deliberately does not answer jurisdiction-specific legal questions. Qualified human review is still required for lawful basis, age or capacity rules, notice duties, controller and processor roles, retention and deletion obligations, legal holds, breach response, transfer rules, records management, publication liability, accessibility obligations, translation adequacy, and accountable research and editorial governance.
Synthetic comprehension method
Each task tests one bounded concept. A response is compared with expected concepts and misunderstanding patterns. The result is an issue log, not a person score. A misunderstanding triggers a clarification and a renewed explicit response; it must not be converted into a clinical, intelligence, credibility, susceptibility, loyalty, or capacity assessment.
Accessibility and language protocol
The browser-local lab exposes keyboard-operable controls, source order, visible focus, responsive reflow, reduced-motion compatibility, plain-language wording, script direction, and disclosure status. Structural checks do not substitute for testing with representative screen-reader, low-vision, motor, cognitive, d/Deaf, and multilingual users.
Translation records preserve original wording, translated wording, translator role, method, uncertainty, and version history. A model-generated draft cannot establish legal or cultural equivalence. Ambiguity remains visible and blocks inferred consent.
Disclosure-preview and lifecycle boundary
A disclosure preview identifies public fields, private exclusions, unresolved permission, withdrawn fields, contradictions, and publication blockers. Private identity, contact, precise location, employer, immigration status, protected characteristics, and verification material remain excluded by default.
Withdrawal, consent change, deletion request, correction, appeal, and retention are different processes. The protocol does not promise deletion of every public, archival, legal, backup, checksum, or third-party copy. Any real process must state its limits, reasons, responsible authority, and review route before consent is sought.
Research-governance boundary
Synthetic usability evidence is not consent evidence. Consent evidence is not legal advice. Legal review is not locally informed review. Local review is not representative public opinion. A valid form is not publication approval. Every layer remains separately addressable and blocked until direct evidence exists.
Source and method references
The protocol reuses bounded methodological lessons from the preserved information-environment packets on plain-language trust repair, language and translation, disability and low-bandwidth access, and the separation of public opinion from institutional claims. Their source-authored “Final” labels do not establish repository review or legal authority:
- Media Literacy and Trust Repair
- Cross-Cultural Communication Research
- Global Information Access Research
- Public Opinion Measurement Research
Validation contract
The focused validator checks twelve schemas, every generated record, equal treatment of all nine permissions, non-numeric rubrics, translation provenance, private-field exclusions, append-only history, unresolved legal and accessibility questions, original-byte preservation, route integration, no-network markers, and publication blockers. Managed Chromium checks exercise task selection, issue rendering, disclosure states, translation direction, imports, downloads, keyboard controls, responsive containment, and unexpected network traffic.
Memory References
- Active summary:
.uai/country-context-consent-evaluation.uai - Reviewer workspace:
.uai/country-context-review-workspace.uai - Pointer ledger:
.uai/long-term-memory.uai - Synthetic validation report: [Synthetic Consent Comprehension Fixtures](synthetic-consent-comprehension-fixtures-accessibility-translation-and-fail-closed-validation.md#findings)
Related Durable Documents
- [Synthetic Consent Comprehension Fixtures](synthetic-consent-comprehension-fixtures-accessibility-translation-and-fail-closed-validation.md#findings)
- [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)
Supersession Status
Current for the 2.0.28-wip synthetic consent-language and disclosure-comprehension evaluation layer. It is additive and supersedes no reviewer submission, consent record, audit, correction, source verification, adjudication, recheck, or publication decision.
