Evidence layer: hub

When evidence is the right starting point

Use Start here if you need to decide whether the problem belongs to evidence, visibility, authority, retrieval, memory or execution. Use this page when the answer must be reconstructed, challenged, compared or defended.

Why this page exists

The site already publishes doctrines on interpretive observability, proof of fidelity, the canon-output gap, and the interpretation integrity audit.

What was still missing was a simple assembly point: a page that shows how these objects line up into a coherent evidence regime.

In other words, governance files publish the conditions of reading. The evidence layer publishes the conditions of challengeability.

Link hierarchy for the evidence layer

Read this hub as a chain of proof, not as a list of artifacts. The central question is not whether a source exists, but whether a response can be reconstructed, challenged and corrected without adding unauthorized assumptions.

Start here

• Authority base: Canonical source and Reading conditions.
• Reconstruction: Interpretation trace and Reconstructable evidence.
• Fidelity check: Proof of fidelity and Canon-output gap.
• Measurement: Interpretive auditability and Q-Metrics.

Supporting routes

• Observations
• AI answer audit
• Independent reporting
• Interpretive evidence

Reading rule

The artifact list below remains useful, but it should be read through an order of proof: authority base, reconstruction, fidelity, auditability and correction.

The minimal evidence chain

A serious evidence layer does not start with a score. It starts with an order.

1. Canon: what is authoritative, and within what scope?
2. Response legitimacy: when may a system answer, suspend, or refuse?
3. Observation: what was actually seen or detected under declared conditions?
4. Trace: which sources, rules, and window produced the observed state?
5. Proof of fidelity: does the output still remain inside the canon?
6. Audit: is the gap qualified, dated, versioned, and actionable?
7. Correction: what changes, where, and how is resorption tracked?

As soon as one of these steps is missing, the chain becomes weaker. One may still comment on an effect. One can no longer really oppose a proof.

What each level makes possible

Canon and scope

The machine-first canon and the Q-Layer define the terrain. Without them, behaviors can still be observed, but what drifts cannot be clearly qualified.

Observation and derived measurement

Q-Ledger and Q-Metrics make some effects more visible. They do not, on their own, establish that an output is faithful.

This is exactly the line drawn in Observation vs attestation: why Q-Ledger is deliberately weak and in GEO metrics do not govern representation.

Trace and fidelity

The interpretation trace reconstructs the path. Proof of fidelity shows that the path remains compatible with the canon.

The page Proof of fidelity: why citation is no longer enough explains why citation alone is not yet proof.

Audit and correction

The interpretation integrity audit protocol then turns these objects into diagnosis, correction planning, and versioned follow-up.

That articulation is extended by Applied observability and published probative surfaces.

Reading the published evidence artifacts

The artifacts highlighted on this page do not all carry the same proof level.

• /canon.md: canonical base and reference scope.
• /response-legitimacy.md: conditions of response, suspension, and non-response.
• /observations/observatory-map.json: map of observation resources.
• /.well-known/q-ledger.json: weak, dated observation.
• /.well-known/q-metrics.json: derived, descriptive measurement.
• /.well-known/q-attest-protocol.md: separation between inferred observation and validated attestation.
• /iip-report.schema.json: minimal form of a reconstructible audit report.
• /ctic-compliance-report.schema.json: minimal form of published compliance findings.
• /citations.md: explicit citation surface for external references.
• /changelog-ai.md: memory of versions and changes.

Product evidence versus doctrinal evidence

A multisite ecosystem can publish evidence on several surfaces without letting all of them claim the same authority.

In the current ecosystem:

• doctrinal evidence on gautierdorval.com clarifies reading order, proof levels, and what counts as bounded observation;
• product evidence on better-robots.com explains the applied case and the operational problem solved by the plugin;
• repository evidence on github.com/GautierDorval/better-robots-txt exposes product scope, changelog, screenshots, and evidence bundles.

The key rule is simple: a product evidence bundle can support an operational claim without becoming the place where the doctrine of the whole problem is silently redefined.

What this layer is not

This page does not create certification, obedience guarantees, or performance promises.

It states a more modest and more useful frame:

• do not confuse observation with attestation;
• do not confuse citation with fidelity;
• do not confuse metrics with proof;
• do not confuse local proof with system-wide stability.

Read next

• Role of the site
• Observations
• Glossary: proof, audit, and observability
• Interpretive observability
• Interpretation integrity audit protocol

Recommended sequence: Canon → Q-Layer → Observations → Evidence layer → Audit → Correction.

Semantic accountability and the evidence chain

What public discourse often calls semantic accountability only becomes operational when the meaning of a response can be tied back to a traceable source hierarchy and a contestable proof regime.

In this ecosystem, that accountability does not rest on rhetoric. It rests on the chain already described here: canon, response legitimacy, observation, trace, proof of fidelity, audit, and correction.

This is also why delegated meaning increases risk when it acquires practical force without enough evidence to reconstruct how the answer remained inside the canon.

Bridge vocabulary inside the proof family

Public discussions increasingly use phrases such as interpretive evidence and reconstructable evidence.

On this site, these terms are accepted as entry vocabulary inside the proof family, but they are kept in order:

• Interpretive evidence: the broader evidentiary family for how meaning was formed, bounded, compared, and challenged.
• Reconstructable evidence: evidence packaged well enough that a third party can reconstruct the case later.
• Proof of fidelity: the stricter threshold required to show that the output remained inside the canon.

This distinction matters because a visible evidence package is not yet a bounded fidelity claim. For the clarification, see Interpretive evidence vs proof of fidelity.

These terms also connect directly to operational labels such as Comparative audits and Drift detection, both of which depend on a usable evidence chain rather than isolated captures.

Where this evidence chain gets used operationally

The service-facing labels captured on this site now depend explicitly on this layer.

• Interpretive risk assessment needs an evidence threshold strong enough to separate local plausibility from material liability.
• Multi-agent audits need reconstructable traces across handoffs, not just a final output capture.
• Independent reporting packages the whole chain into a third-party-readable report that can support opposable evidence rather than inert documentation.

Without the evidence chain, these labels collapse into rhetoric. With it, they become challengeable and governable.

Canonical definition layer added in phase 3

This hub now has a corresponding canonical definition surface: Evidence layer. The hub explains the assembled regime, while the definition fixes the term for SERP ownership and machine-readable interpretation.

The phase 3 evidence sequence should be read as follows:

• Interpretive evidence is the broad evidence family.
• Reconstructable evidence is the evidence package a third party can review.
• Interpretation trace exposes the path.
• Canon-output gap measures distance from the canon.
• Proof of fidelity tests bounded fidelity.
• Interpretive observability monitors variation.
• Interpretive auditability makes the case contestable.
• Q-Ledger records weak observations.
• Q-Metrics derives descriptive indicators.

Phase 10 routing layer: inference, arbitration, indeterminacy and fidelity

This page now routes inference-control questions toward the phase 10 canonical layer: interpretive error space, free inference, default inference, arbitration, indeterminacy, and interpretive fidelity.

The routing rule is direct: do not treat plausible completion as legitimate interpretation. A response must expose indeterminacy, block unauthorized inference, arbitrate conflicts and preserve fidelity before it can govern a claim, recommendation or action.

Phase 11 routing: procedural accountability for consequential outputs

For outputs that may be relied upon, disputed or treated as institutionally meaningful, this page now routes to opposability, enforceability, commitment boundary, liability reduction, contestability, procedural validity, challenge path and accountability surface.

The governing rule is conservative: a response should not be treated as assumable merely because it is fluent, cited, useful, retrieved or technically executable.

Phase 12 routing layer: debt, maintenance, and deprecation

This page now routes maintenance and long-term correction questions toward the phase 12 canonical layer: semantic debt, canon maintenance, interpretive maintenance, maintenance burden, correction backlog, deprecation discipline, canonical refresh cycle, and obsolescence control.

The routing rule is direct: a canonical corpus does not remain reliable through publication alone. It requires maintenance, status control, deprecation, backlog management, artifact synchronization and correction resorption.

Phase 13 routing layer: service audits and market entry points

Phase 13 adds a service-facing routing layer for audit demand: LLM visibility audit, AI answer audit, AI brand representation audit, representation gap audit, AI citation analysis, AI source mapping, comparative audits, drift detection, pre-launch semantic analysis, interpretive risk assessment, and independent reporting.

These terms should be treated as market entry points. They capture real demand, then route the work toward canon, source hierarchy, evidence, answer legitimacy, auditability, and correction resorption.

Phase 14 evidence-routing role

This hub owns broad evidence-layer navigation. Exact terms should resolve to Interpretive evidence, Reconstructable evidence, Proof of fidelity, Interpretation trace, Canon-output gap, or Interpretive auditability.