Evidence layer
Probative surfaces brought into scope by this page
This page does more than point to governance files. It is also anchored to surfaces that make observation, traceability, fidelity, and audit more reconstructible. Their order below makes the minimal evidence chain explicit.
- 01Canon and scopeDefinitions canon
- 02Response authorizationQ-Layer: response legitimacy
- 03Evidence artifactinterpretive-integrity.json
Definitions canon
/canon.md
Opposable base for identity, scope, roles, and negations that must survive synthesis.
- Makes provable
- The reference corpus against which fidelity can be evaluated.
- Does not prove
- Neither that a system already consults it nor that an observed response stays faithful to it.
- Use when
- Before any observation, test, audit, or correction.
Q-Layer: response legitimacy
/response-legitimacy.md
Surface that explains when to answer, when to suspend, and when to switch to legitimate non-response.
- Makes provable
- The legitimacy regime to apply before treating an output as receivable.
- Does not prove
- Neither that a given response actually followed this regime nor that an agent applied it at runtime.
- Use when
- When a page deals with authority, non-response, execution, or restraint.
interpretive-integrity.json
/interpretive-integrity.json
Published surface that contributes to making an evidence chain more reconstructible.
- Makes provable
- Part of the observation, trace, audit, or fidelity chain.
- Does not prove
- Neither total proof, obedience guarantee, nor implicit certification.
- Use when
- When a page needs to make its evidence regime explicit.
Legitimate contextual variation
Legitimate contextual variation is a difference between two representations of the same entity caused by an explicit context change, compatible with material invariants, supported by sufficiently proven relations and bounded by the observed situation.
It can be represented as:
R(E | C1) ≠ R(E | C2)The difference is legitimate when moving from C1 to C2 explains the changed conclusion and E is not rewritten to obtain it.
Cumulative criteria
Variation may be qualified as legitimate when:
- the same entity is correctly resolved in every observation;
- entity invariants remain compatible;
- the context change is material to the conclusion;
- contextual relations are accurate, dated and correctly attributed;
- exclusions and reversal conditions remain visible;
- statement strength matches evidentiary strength;
- the variation can be explained without invented data;
- a symmetrical inversion of context produces coherent reassessment.
Example
For the same software:
- under a self-hosted deployment context with an experienced technical team, the representation may emphasize flexibility and control;
- under a small-team context with no operational capacity, it may emphasize operational burden and the need for a managed service.
Both outputs may be faithful if software capabilities remain identical and the difference concerns fit with the use context.
What is not legitimate variation
A contradiction about identity, an invented capability, an expired policy presented as current or a local relation generalized to all situations is not contextual variation. It is error, drift or overstatement.
Likewise, an output that follows user preference to the point of altering factual truth is not legitimate adaptation. Context may guide selection and explanation. It must not personalize facts.
Relation to stability
AI perception stability must not be understood as verbatim repetition. A stable representation may vary by audience, intent or time provided invariants remain recognizable and variation remains explainable.
Legitimate contextual variation is therefore the positive complement of stability: it describes admissible difference, while interpretive drift describes difference incompatible with canon, evidence or context.