Private Evidence And Verified Propositions¶
Status And Decision¶
Design note and future profile direction. This note clarifies the evidence model; it does not introduce an event kind, wire-format object, proof system, or implemented private-state verifier. Existing Core Record Ruleset 1.0 and Nostr event-validation requirements continue to apply.
Independent verification does not imply full disclosure. Consequential State is independently derived from sufficient verifiable evidence under identified rules. A future profile may permit a required proposition to be established through disclosed evidence or an accepted privacy-preserving proof.
The control layer remains central: control consequences follow from evaluating evidence under the applicable control rules. Recognition and external effect remain separate decisions. No mint, bearer credential, or authoritative spent-state database is introduced by this refinement.
External History And Exact Bytes¶
Artifacts carry content. Evidence carries history. Rules derive consequence.
In OpenETR, the proof is outside the artifact. A Digital Artifact need not contain OpenETR metadata, controller information, or its own history. External Evidence Graph records refers to the artifact by digest; the artifact need not refer back to that evidence.
Byte identity is the concept; digest is the mechanism. In the current binding,
artifact identification checks SHA-256(candidate_bytes) == object_digest.
This establishes exact-byte correspondence under the hash assumptions, not
semantic meaning, authority, or recognition. Identification does not require
parsing a PDF, image, credential, or future format.
The base protocol performs no implicit normalization, metadata stripping, or
payload extraction. A profile must identify the exact bytes it treats as the
artifact. A transformed representation is a different artifact if its bytes
differ; relationships such as derived-from, representation-of, replaces,
or supersedes require evidence and defined profile semantics. These examples
do not register new base actions or tags.
Evidence And Disclosure¶
Verifiable evidence is evidence sufficient for an independent verifier to establish the propositions required by the applicable rules. The underlying information need not necessarily be fully disclosed.
Private source material alone is not independently verifiable by a party that cannot inspect it. That party needs an accepted proof or other evidence with a defined scope. Encryption controls access; a hash binds bytes; neither alone proves that hidden content satisfies a mandate, eligibility, or transfer rule.
Three access questions are distinct: access to the artifact, access to the underlying supporting evidence, and access to the material needed to verify a proof. Independent verification need not mean public publication. Authorized verifiers may exchange evidence privately, subject to availability and policy.
flowchart TD
A[Disclosed evidence] --> V[Evidence verification under an identified profile]
B[Private supporting evidence] --> P[Cryptographic proof]
P --> V
V --> Q[Scoped verified propositions]
Q --> R[Identified transition rules]
R --> S[Consequential State]
S --> C[Contextual Recognition]
C --> E[External Effect]
Verified Propositions¶
A verified proposition is a scoped conclusion established by checking evidence with an identified verification procedure. It is a conceptual interface between evidence verification and transition evaluation, not a new wire-format object or a universally true claim.
A future adapter should preserve the proposition, artifact and event binding,
subject and action, relevant time or graph cut, evidence references, procedure
and version, public inputs, assumptions, and outcome. A bare true supplied
by an application is not enough to reproduce the verification.
| Required proposition | Possible mechanism under an explicit profile | Limit |
|---|---|---|
| A key authorized this action | Disclosed signed action or proof bound to the action and key | Key attribution does not establish institutional authority |
| A qualifying mandate covered the action at the relevant time | Disclosed signed mandate or proof of qualifying scope and validity | Issuer authority, revocation scope, and temporal assurance still need rules |
| The recipient accepted this transfer | Signed acceptance or an accepted proof of that acceptance | A generic credential-possession proof does not establish consent to this transfer |
| No blocking encumbrance exists in the required scope | Evidence or a non-membership proof over a defined, authenticated state commitment | Absence from a partial relay response cannot establish global absence |
These are illustrative predicates, not additions to the baseline transfer rules. In particular, this note does not make recipient acceptance mandatory in profiles that currently use another transfer model.
Rules define what must be proven, not necessarily what must be revealed. The ruleset identifies the required propositions and acceptable evidence profile. The profile specifies how to establish them, including trust roots, proof verification, context binding, freshness, and failure handling. Two mechanisms are interchangeable only if they establish the required proposition with the assurance that the rules require. An institutional attestation proves that an institution made a statement; accepting its substance introduces that institution as a trust dependency.
Missing material, unsupported proof mechanisms, failed proof checks, and a verified negative proposition are distinct outcomes. A future verifier must not silently treat the first three as successful proof or as proof that the underlying proposition is false. If a required proposition remains unresolved, report insufficient evidence or an indeterminate result under the rules.
Durability And Historical Derivation¶
Durability of verification does not require durability of disclosure. It does require preservation of sufficient verification material and assumptions. A historical verification package may need:
- the proof, public inputs, commitments, and binding to the artifact and event;
- the proof-system, circuit, verification-key, and parameter versions;
- the evidence profile and transition ruleset identifiers and versions;
- issuer, mandate, registry, or revocation evidence applicable to that scope;
- Temporal Proofs, where required, and the provenance of relevant checkpoints;
- retrieval scope, known conflicts, and unresolved dependencies.
A mutable URL to a verifier service is not a durable substitute for that package. A signed service verdict is evidence of the service's assertion; it does not let a recipient independently repeat the underlying proof check.
Retain the existing conceptual derive_state(evidence, rules, at_event=E)
boundary. A proof generated today about an earlier event does not by itself
prove that the proof existed then. Time-based claims require explicit temporal
semantics and evidence; signed event timestamps do not supply trusted time or
a total order. Future checks remain subject to availability and the continuing
adequacy of cryptographic assumptions.
Determinacy And Coverage¶
Multiple authentic assertions may concern the same digest. OpenETR does not require uniqueness of assertion. It aims for reproducible consequence given the same evidence, identified rules and versions, evaluation parameters, and verification dependencies. A deterministic result may be conflict, ambiguity, or insufficient evidence rather than one controller.
Different disclosed subsets may produce different results. Each result must identify its evidence scope; no hidden or omitted history is presumed absent. A non-membership proof establishes absence only within its committed set. Rules must separately establish why that set has the required coverage, authority, and freshness. Privacy-preserving proofs do not solve equivocation or global completeness on their own.
Threat And Privacy Considerations¶
| Risk | Required design response for a future profile |
|---|---|
| Replaying a proof for another artifact, transfer, or ruleset | Bind the statement to the relevant digest, event, action, participants, and profile; define challenge or freshness requirements where needed |
| Hiding a revocation, conflicting branch, or encumbrance | Specify retrieval coverage, authenticated checkpoints, and conflict rules; preserve unknown completeness |
| Proving the wrong statement with a valid proof | Identify the exact predicate, circuit, public inputs, and verification parameters; review their correspondence to the rule |
| Substituting untrusted verification keys or setup material | Authenticate and version dependencies; state setup and trust assumptions |
| Losing historical proof dependencies | Archive the verification package and identify unavailable dependencies in results |
| Correlating private activity | Assess exposed digests, public keys, graph links, timestamps, proof inputs, and repeated presentations |
| Guessing sensitive artifact contents | Treat an unsalted artifact digest as an equality handle, not a hiding commitment; predictable candidate files can be hashed and tested |
| Disclosing private records through storage or logs | Minimize published content and enforce access and retention policy in the hosting application |
Hiding a witness does not hide the public Nostr event graph. The current event shape exposes signing keys and graph references to event recipients. Private controller continuity therefore needs a separately reviewed profile and may need a new representation; it cannot be promised by attaching a proof to an otherwise identifying event. A privacy commitment, if introduced by a profile, must be distinguished from the unchanged exact-byte artifact digest.
Implementation Direction¶
Keep ZK and selective-disclosure mechanisms optional. First specify a bounded profile, for example mandate-scope verification, and the evidence-to-proposition interface. Review binding, revocation, completeness, and historical replay; then implement an adapter and independent verification tests. The library, REST API, web app, and CLI should use the same component result when supported.
This note does not authorize current software to bypass required signed events, accept opaque proofs as validated evidence, or claim private controller support. It refines the architectural extension point while preserving the Evidence Graph and control model.