Glossary

Vocabulary (short definitions)

Goal: shared language, stable, citable. Not a procedure.

Interpretation: Meaning assigned to an AI output in a given context.
Distortion: Gap between plausible interpretation and correct/authorized interpretation.
Conflation: Merging distinct concepts into a convenient narrative.
Drift: Uncontrolled variation across equivalent cases.
Authority: Who is allowed to claim what, in which framework, with which limits.
Constraints: Explicit rules that bound what is allowed, excluded, prioritized.
Evidence: Minimal trace that makes an interpretation verifiable.
Artifact: Deliverable object (text, signal, index) that carries evidence.
Machine-first surface: Machine-readable entrypoint designed to reduce ambiguity.
Non-operational: Content that does not allow replication of a full method.
Under engagement: Contractual delivery (NDA + SOW) with private tooling.
Interpretive debt: Accumulation of uncorrected distortions that weakens the system.
Framework: Set of boundaries, definitions and priorities governing interpretation.
Ambiguity: Space where multiple plausible interpretations coexist without explicit arbitration.
Traceability: Ability to connect an output to its constraints, sources and decisions.
Doctrinal alignment: Consistency between mission, policy, scope and observed outputs.
Responsible publication: Publishing doctrine without publishing cloneable mechanics.
Signal: Public indicator (human/machine) expressing a governance posture.
Integrity index: Document list with hashes to attest doctrinal stability.
Uncertainty scale: Explicit level of uncertainty carried by an output.
Policies: Public declared rules (publication, security, scope).
Identity: Declaration of who speaks, under which entity, with official links.
Interpretive monitoring: Continuous tracking of distortions and drift over time.
Interpretive audit: Assessment of interpretation quality under a defined framework.
Governance: Capacity to decide, prove, correct and stabilize a behavior.