Interpretive Governance

Interpretive Governance is a machine-first governance standard intended to reduce interpretive drift in probabilistic and agentic systems by enforcing explicit separation between observed facts, derived values, inferred hypotheses, and unknown fields.

This standard is not the academic field of interpretive governance in political or social sciences.

Canonical files (core manifest)

• interpretive-governance.manifest.json (canonical manifest)
• .well-known/interpretive-governance.json (stable entrypoint)
• schemas/manifest.schema.json (JSON Schema)
• authority-graph.jsonld (authority graph)
• interpretive-index.json (machine-first index)
• versions/0.1.2/interpretive-governance.manifest.json (immutable core snapshot)

Canonical registries

• glossary.md (canonical category boundaries)
• references.md (canonical references)
• REGISTRIES.md (canonical registry map)
• gautierdorval.com/definitions/ (external canonical definitions registry)
• gautierdorval.com/lexique/ (external lexicon navigation)
• public specification (scope + minimal conformance)
• ontological architecture (layering + prevalence)

Extensions (Layer 3 + ops pack)

Interpretive Governance is the normative core (0.1.x). The following extensions (1.x) are additive: they do not redefine claim typing, ontology distinctions, or interpretive/constraintive separation.

Core snapshots live under /versions/0.1.x/. Extensions are pinned via Git tags (immutable raw endpoints).

Authority governance (Layer 3) — pinned v1.1.0

• inference-vs-authority.md (boundary: inference never implies authority)
• action-taxonomy.json (action taxonomy)
• proof-model.md (proof primitives)
• authority-policy.json (authorization policy)
• authority-ledger-spec.md (ledger specification)
• role-topology.md (roles, veto, separation of duties)
• incentive-integrity.md (anti proxy hacking constraints)
• non-objectives.md (no recipes, no bypasses)

Agentic operations pack (minimal) — pinned v1.1.0

• H-layer: human-oversight-protocol.md
• P-layer: privacy-data-lineage.md
• V-layer: versioning-drift-change-control.md
• S-layer: security-abuse-resistance.md
• IR-layer: incident-response.md

Interpretive rules (public projection)

Public, human-readable projections of normative rules. In case of divergence, the canonical Markdown in the manifest prevails.

• R3 — Interpretive abstention
• Anti-faux audit
• Anti-identity inference

Protocols (evaluation methods)

• Interpretive stress-test protocol
• Interpretation integrity audit protocol

Extension source repository (normative, no recipes): github.com/GautierDorval/interpretive-governance-manifest

Informative human-readable doctrine: gautierdorval.com/doctrine

Interpretive Governance defines the normative core layer. Implementation architectures and application domains must reference this standard without redefining it.

Status: draft.

Non-normative context: context