Status
Current workflow specification · WIP · exact source-recheck cutoff 2026-07-23T20:01:27Z · no reviewer submissions imported.
Purpose
Define a safe, equal-burden, machine-readable path for accepting real reviewer evidence without inventing identity, expertise, local relationship, language competence, conflicts, consent, claim responses, or adjudication. The method separates public-source corroboration from human endorsement and fails closed when required evidence is absent.
Scope
The method applies to the five audited draft country proposals. It supports reviewer evidence from people with relevant professional, institutional, linguistic, lived, local, or affected-community knowledge. Relevance must be described, not scored. One reviewer cannot represent an entire jurisdiction or community.
Executive Summary
- Five unanswered 2.0.23 review packets are imported into five new intake packets.
- Reviewer-owned fields remain null.
- Attributable, pseudonymous, confidential, and non-public handling states are supported.
- Consent to participate is separate from permission to quote, attribute, retain, or publish.
- Twelve claim-level adjudication slots per jurisdiction remain unperformed.
- Publication fails closed until required provenance, permission, evidence, and adjudication fields are complete.
Evidence Reviewed
- Review workflow index
- Cross-batch record
- Correction and supersession register
- Consent schema
- Reviewer-evidence schema
- Adjudication schema
- Intake packet schema
Findings
Who may review
A reviewer may contribute relevant institutional, professional, legal, linguistic, lived, local, or affected-community knowledge. The workflow records the relationship and scope. It does not create reviewer-worth or community-credibility scores, and it does not infer representativeness from identity, nationality, ethnicity, religion, language, location, occupation, or institutional role.
Required provenance
Every supplied review must record an attributable identity, a consented pseudonym, or a confidential/non-public identity state; a verification route held privately where necessary; relevant qualification or experience; relationship to the jurisdiction, institution, or affected community; language competence and translation support; conflicts and funding; and the date and scope of review. Missing fields remain null.
Consent and publication scope
Consent is granular. The record distinguishes consent to participate from permission to quote, paraphrase, attribute, publish anonymously, retain privately, or publish the response. None may be inferred from another. Safety restrictions override publication convenience.
Claim-level evidence
Each claim response records the response, rationale, supporting source or experience, confidence, requested wording, and any requested correction or withdrawal. Lived experience is recorded as bounded evidence, not representative public opinion. Public sources are checked separately from reviewer testimony.
Source verification
A model may validate record structure, compare supplied text with preserved sources, identify inconsistencies, and draft questions. It may not fill reviewer-owned fields, infer consent, certify expertise, or label its own comparison as independent or locally informed review.
Editorial adjudication
An editor or adjudicator records evidence considered, decision, rationale, requested wording, publication effect, and appeal or correction route. Adjudication remains separate from the reviewer record. Conflicting reviews are preserved rather than collapsed into a fabricated consensus.
Public status vocabulary
- Unanswered: no reviewer-owned response is supplied.
- Submitted: a record exists but provenance, permissions, or evidence may still require verification.
- Verified-source: cited public sources were checked; this does not verify lived experience or endorse the conclusion.
- Adjudicated: an accountable editor recorded a decision and rationale.
- Withdrawn: the reviewer requested withdrawal within the recorded scope; published history is not silently erased.
- Superseded: a later record replaces a prior status or wording while preserving history.
- Confidential: the response is retained under a non-public permission boundary.
- Publishable: all required fields, consent, safety, provenance, evidence, and adjudication gates pass within the recorded scope.
Fail-closed publication gate
A record is not publishable if identity or anonymity choice, relevant relationship, language/translation, conflicts/funding, consent scope, claim response, supporting evidence or experience, source verification, adjudication, or safety review is incomplete. A passing structural gate proves only that a record is reviewable.
Withdrawal, correction, and deletion
Withdrawal, correction, and deletion requests receive dated append-only records. Private personal data may be removed from active systems according to the recorded request and legal obligations, while any prior public statement receives a visible withdrawal or correction notice rather than silent historical erasure.
Fairness and safety
All five jurisdictions receive the same intake, disclosure, consent, adjudication, appeal, correction, and publication burdens. Differences in evidence or reviewer coverage must be explained without rankings, scores, protected-characteristic inferences, threat framing, or permanent cultural traits.
Decisions or Recommendations
- Accept only real supplied reviewer evidence; never infer reviewer-owned fields.
- Keep private verification data out of public records.
- Treat source verification, reviewer testimony, and adjudication as separate layers.
- Preserve disagreement and bounded scope.
- Keep every profile publication-blocked until the fail-closed gate passes and the approval scope is explicit.
Risks and Limitations
One or more supplied reviews would not establish exhaustive local representation, representative public opinion, universal consent, legal correctness, scientific validity, geopolitical neutrality, or publication fitness for every audience. Confidentiality and withdrawal requests may constrain what can be published or retained. Automated validation cannot determine whether a reviewer understood every question or whether all affected communities were reached.
Validation Performed
Local validation checks schema conformance, null preservation, supported safety states, consent separation, equal burden, append-only history, route integrity, memory links, and fail-closed publication effects. It does not validate human identity, expertise, consent authenticity, lived experience, or factual truth.
Memory References
Related Durable Documents
- [Cross-batch reviewer workflow report](current-country-review-workflow-cross-batch-fairness-coverage-disagreement-and-unresolved-review.md#findings)
- [Country Context Composer report](country-context-composer-and-comparative-fairness-report.md#findings)
- [World Systems governance and evidence report](world-systems-governance-resilience-and-evidence-integration-report.md#findings)
- [Evidence audit cross-batch report](current-country-evidence-audit-cross-batch-disagreement-asymmetry-and-unresolved-review.md#findings)
Supersession Status
This workflow supplements the 2.0.23 unanswered review packets. It does not convert them into completed review and does not supersede their identities or blank response history.
