Skip to content
The Second Pair of Eyes markPor DerechoInitiative in formation

Method · source discipline · human gates

Method under development

The Second Pair of Eyes.

The method exists to make difficult legal-record decisions more traceable before harm becomes irreversible. It is not an accusation engine, appeal body, disciplinary authority or automated decision system.

Stable method

Six checks before a decision hardens.

Each check must identify the source, the uncertainty and the next human decision. The method fails if it only makes an allegation look more sophisticated.

SourceWhich document, version and evidence state supports the proposition.
AuthorityWho could act, decide, certify or represent at the relevant date.
PerimeterWhich assets, people, time periods and proceedings are actually reached.
ContradictionWhat material evidence points elsewhere or weakens the proposed view.
ConsequenceWho gains, who loses and what practical effect becomes embedded.
ReversibilityWhether a later remedy could genuinely repair the harm.

AI boundary

What AI may and may not do.

The method can be AI-assisted only when the system preserves human authority, source traceability and correction.

Permitted assistance

  • Ingest and classify documents.
  • Extract dates, entities, amounts, authorities and proceedings.
  • Find source-backed answers and contradictions.
  • Build chronology, gaps and review questions.
  • Route items to human review and preserve audit trail.

Prohibited outputs

  • No automated decision.
  • No guilt score, credibility score or person-risk score.
  • No sanction, closure, professional finding or legal-liability decision.
  • No inference presented as established fact.
  • No institutional adoption claim without express written confirmation.

From founding stress test to transferable method

The live matter explains why the method exists. It does not control the method.

Sun Park remains the founding stress test. The Second Pair of Eyes must be capable of being tested on synthetic and contrary scenarios where the concern is weakened, corrected or closed. That is why partner and institutional conversations should begin with a synthetic pilot, not confidential live evidence.

Open the synthetic pilot packTechnical partner pathway

From six questions to a review record

The operating standard dated 12 September 2026 adds an immutable source snapshot, event and knowledge timelines, contrary evidence, a named human decision, and correction and handover records. Existing founder-related applications are internal experiments. The first independent review uses wholly fictional records; neither those experiments nor a successful synthetic test establishes institutional adoption.

Public method and current operating status (12 September)Review protocolStatus and interests