---
uai: "1.0"
type: country-context-consent-evaluation
status: active
stable_id: iarpg-country-context-consent-evaluation
project: iarpg.com
release: IARPG-OPS-2
version: "2.0.28-wip"
review_state: current-until-superseded
---

# Country Context Consent Language and Disclosure Comprehension Evaluation

## Current Publication Contract

- Public route: `/country-context/reviews/consent-lab`.
- Machine index: `/country-context/consent-evaluation/index.json`.
- Inventory: 12 schemas, 69 individually addressable records, 9 independent permission-language records, 12 synthetic comprehension tasks, 12 non-numeric rubrics, 6 synthetic response fixtures, 8 record-level issue classifications, 3 model-generated translation drafts, 5 disclosure outcomes, 5 append-only history events, 6 unresolved qualified-review questions, 1 workspace state, and 1 original-byte import envelope.
- Human evidence state: 0 real participants and 0 human comprehension results.
- The lab is browser-local and uses no accounts, analytics, external model, network submission, remote storage, database, upload service, or production study endpoint.
- Original wording, plain-language alternatives, translated drafts, translator role, method, direction, uncertainty, and version history remain separately addressable.
- Participation, quotation, paraphrase, attribution, private retention, anonymization, publication, withdrawal contact, and deletion-request handling remain independent choices.
- No numeric pass score is generated. Issue categories and unresolved states apply to records and interface behavior, not to people.
- Disclosure outcomes expose included, excluded, unresolved, withdrawn, contradictory, and blocked states while excluding private identity and verification material by default.
- Invalid JSON and unknown extensions preserve original bytes, filename, media type, extension, byte count, hash, text where decodable, and unchanged quarantine status.
- Country Context Composer references this protocol additively. Consent-evaluation records may not mutate proposal claims, audit dispositions, reviewer responses, source verification, corrections, adjudications, public candidates, handoffs, or volatile rechecks.

Evidence: [protocol findings](../docs/long-term-memory/reports/country-context-consent-language-and-disclosure-comprehension-evaluation-protocol.md#findings), [synthetic fixture findings](../docs/long-term-memory/reports/synthetic-consent-comprehension-fixtures-accessibility-translation-and-fail-closed-validation.md#findings), [machine index](../country-context/consent-evaluation/index.json), and [public lab](../country-context-consent-lab.php#consent-lab-boundary).

## Analytical Separation

1. Consent-language inventory and plain-language alternatives.
2. Legal and governance questions requiring qualified human review.
3. Synthetic comprehension tasks and non-numeric rubrics.
4. Structural accessibility checks and unresolved representative-user review.
5. Translation provenance and uncertainty.
6. Disclosure-preview comprehension and publication blockers.
7. Withdrawal, consent change, deletion request, correction, appeal, and retention explanations.
8. Research-governance separation among usability evidence, consent evidence, legal review, local review, and publication approval.
9. Append-only machine records for tasks, responses, issues, corrections, and unresolved questions.
10. Publication and promotion gates that remain blocked without direct evidence.

## Safety and Truth Boundary

- Synthetic tasks are not participants, consent, comprehension evidence, accessibility acceptance, legal advice, local consultation, representative public opinion, or publication approval.
- A structurally valid form or clear synthetic answer does not establish voluntariness, capacity, legal sufficiency, language equivalence, cultural fit, or informed human understanding.
- No reviewer-worth, community-credibility, comprehension, susceptibility, capacity, country, morality, legitimacy, stability, threat, loyalty, dangerousness, diagnosis, protected-characteristic, recruitability, or targetability score is authorized.
- The Spanish, French, and Arabic records are model-generated interface drafts. They remain blocked pending qualified human translation, accessibility, local-context, and legal review.
- Native Firefox behavior, native assistive-technology speech output, representative disabled-user comprehension, production-host security, field performance, and live search/crawler outcomes remain unestablished.

## Related Hot Memory

- [Reviewer submission readiness](country-context-review-workspace.uai#current-publication-contract)
- [Reviewer evidence and rechecks](country-context-reviews.uai#current-publication-contract)
- [Country Context Composer](country-context-composer.uai#reviewer-workflow-attachment-contract)
- [Accessibility](accessibility.uai#current-accessibility-contract)
- [International fairness](international-fairness.uai#core-rule)
- [Evidence model](evidence-model.uai#current-doctrine)
- [Pointer ledger](long-term-memory.uai#pointer-records)