Public Review CandidateBuild 15 · SIW / Canon Reader integrationReview guidenoindex · review deployment
Assurance architecture · controlled-development explanation

Where a claim becomes falsifiable.

The Registry-Independent Assurance Ecosystem paper describes the Interop Lab as the experimental core of a proposed assurance chain: bounded claims enter controlled challenge, evidence is preserved, independent reviewers inspect the record, and only scoped determinations move forward.

Paper: Private Discussion Draft v0.2Interop Lab: architectural roleNot a certification authority.
Narrated orientationHigh-level Interop Lab introduction

Diagram-specific narration remains pending. Build 09 pilots the control pattern on two Research diagrams using clearly labeled excerpts from existing narrated media before we decide whether to propagate voice controls here.

Test voice-control pilots →
From paper to institutional model

Testing, review, determination, and registry reference remain different functions.

The paper explicitly resists collapsing these roles. Laboratory evidence does not make the Lab the adjudicator, and independent review does not silently become broad certification.

External registry / productsource identity preserved
API Adapter 2.0candidate crossing contract
Interop Labbounded reproducible testing
Independent reviewevidence + assumptions inspected
Scoped determinationclaim-bound finding
Registry / assurance recordonly when authorized
What the Lab tests

The question is not whether the control is visible. It is whether the prohibited consequence can still occur.

A serious profile can challenge stale authorization, changed targets or parameters, expired delegation, alternate interfaces, artifact substitution, replay, state changes, or attempts to produce a consequence after refusal.

Input

Bounded claim

Exact artifact version, exact claim, permitted scope, test profile, expected outcome, and prohibited outcome.

Experiment

Controlled challenge

Positive and negative controls, enumerated routes, adversarial substitutions, observation surfaces, and reproducible conditions.

Output

Evidence, not institutional inflation

Logs, receipts, state transitions, reviewer observations, failures, retries, and a bounded record of what the evidence actually establishes.

Diagram exploration

Inspect quickly, open deeply, listen at a high level.

Move the pointer over a diagram for local zoom. Click to open a full-page lightbox with a direct return control. Narration slots are reserved for curated MP3 overviews.

Initial posture

Independent conformance determination

  • Binds the finding to an exact artifact and test profile.
  • States the scope and exclusions.
  • Preserves reviewer attribution and evidence.
  • Supports re-evaluation when material conditions change.
Not established

Broad certification

  • Would require a separately mature scheme.
  • Needs qualification, independence, conflict-of-interest, appeals, retention, surveillance, and revalidation rules.
  • Is a possible future institutional function, not a current marketing claim.
Candidate test profile · Federated SDI

Can evidence compose without authority transfer?

The profile is intentionally narrower than “safe federation.” It asks whether bounded systems can exchange useful claims while identity, credentials, memory and standing remain local.

Planned

Claim exchange

Exchange signed, purpose-limited, expiry-bound claims with exact provenance.

Planned

Non-transfer

Verify that receiving evidence does not import credentials, tool rights, authority ceilings or transitive standing.

Planned

Succession refusal

Challenge a successor requesting broader memory, keys, tools or executor routes; preserve no-bind until separately governed.

Read Federated SDI architecture →