Understand the model¶

Capability Lab starts from a simple constraint: a useful model of a person must not silently become authority over that person.

That changes the architecture. Instead of asking for one universal score, the system keeps evidence, interpretation, evaluation, state, advice, and authority as separate layers with explicit transitions.

The five ideas to keep in mind¶

1. Evidence is not a score¶

An observation is only an external event or source artifact. It does not become evidence merely because a model noticed it. The generic PR12 path requires an explicit reviewed materialization boundary before a neutral EvidenceRecord exists.

Evidence itself still does not establish a capability claim or claim-wide conclusion.

activity != observation authority
observation != evidence
evidence != claim
claim != supported claim

2. Unknown and conflict are real states¶

Capability Lab does not convert missing evidence into zero and does not force disagreement into a single answer. UNKNOWN, insufficient evidence, and unresolved conflict remain representable.

A complete evidence basis can therefore produce a conservative result instead of manufacturing certainty.

3. History is part of the claim¶

Evaluations are retained rather than replaced by "the latest answer." Downstream state derivation uses complete governed portfolios, so a caller cannot silently select only the convenient newer result.

This is why PR12.13 deliberately proves a history containing both an older PARTIAL / INSUFFICIENT evaluation and a newer domain-sufficient SUPPORTED evaluation.

4. State requires governance after derivation¶

Even a valid derived capability state is not automatically the person's accepted or current state.

DERIVED != PERSISTED
PERSISTED != ACCEPTED
ACCEPTED != CURRENT
CURRENT != LATEST
CURRENT != BEST

PR11.6 persists state, PR11.7 accepts it, and PR11.8 selects current state through separate governed steps.

5. Advice and presentation do not grant authority¶

PR11.9 progression is advisory. PR11.11 is a product/read projection. Neither can grant readiness, mastery, licensing, permission, or professional authority.

progression != prescription
product view != write-back
current capability != permission

Shared definitions vs person-scoped records¶

Capability concepts and competence frames are shared semantic definitions. Evidence, evaluations, state, acceptance, and current selection are person-scoped governed records.

Keeping those two categories separate avoids turning a shared ontology into a public profile database.

Where to go next¶

If the mental model is clear, follow the governed pipeline end to end. If you are integrating a consumer, go directly to the consumer boundary.

For normative human-agency and privacy rules, read the constitution. For exact terminology, use the vocabulary.