Work in progress IARPG-OPS-2 2.0.30-wip

Global publication search

Search the operations design system

Type to search standards, missions, roles, authorities, reports, sources, pages, schemas, and UAI memory.

Direct answer

It evaluates the structure of wording and disclosure logic—not people. The lab keeps all nine permissions independent, shows what each choice does and does not authorize, exposes unresolved translation and legal questions, and records synthetic misunderstandings as issue categories rather than numeric scores.

Evaluation material
Synthetic task definitions only
Scoring
No numeric pass or person score
Storage
Browser memory + explicit downloads
Publication state
Blocked

How to use this page

A form can be structurally clear while still being legally inadequate, inaccessible to representative users, poorly translated, or unsuitable for a specific community. Each evidence type has a separate owner and gate.

  1. 01
    Inventory the wordsPreserve original wording and a versioned plain-language alternative.
  2. 02
    Test bounded conceptsUse synthetic tasks and issue categories, never a reviewer-worth or capacity score.
  3. 03
    Expose disclosureShow included, excluded, unresolved, withdrawn, contradictory, and blocked fields.
  4. 04
    Route qualified reviewKeep legal, translation, accessibility, governance, and local review questions unresolved until direct evidence exists.
  5. 05
    Preserve historyAppend corrections, consent changes, withdrawal, deletion requests, and appeals without silent rewriting.
Open the reviewer-submission readiness workspace

Promotion boundary

Only direct, properly governed human and legal evidence can change those states. Synthetic tasks and automated checks are preparation tools, not proof.

USABILITY

Observe interaction separately

Keyboard, reflow, labels, and task flow can be tested without claiming informed consent or representative accessibility.

CONSENT

Require an explicit bounded choice

Do not infer permission from participation, silence, a related choice, or editorial usefulness.

GOVERNANCE

Name accountable human roles

Legal review, translation approval, study governance, adjudication, incident response, and publication approval remain separate duties.

Connected tools and standards

Explore the wider AI ecosystem.