Evidence Graph Design Note¶
Status¶
Draft canonical terminology and boundary note for OpenETR.
Updated 29 October 2026: Evidence Graph replaces Digital Controllable Record (DCR) as the primary model term. This note retains its existing filename so published links continue to work.
Definition¶
An Evidence Graph is a single end-verifiable record or a graph of related end-verifiable records through which consequential state concerning a Digital Artifact can be established and transitioned.
An Evidence Graph is not a file. The file or canonical content is the Digital Artifact. The Evidence Graph is the signed evidence structure that concerns it.
DIGITAL ARTIFACT
uniquely identifiable content
|
v
EVIDENCE GRAPH
single record or record graph spanning the evidenced lifecycle
Evidence Graph + IDENTIFIED RULESET
|
| validation and state transition rules
v
CONSEQUENTIAL STATE -> RECOGNITION -> EFFECT
DIGITAL ARTIFACT + ESTABLISHED CONSEQUENTIAL STATE
|
v
DIGITAL ORIGINAL
The canonical definitions are:
- A Digital Artifact is persistent digital content with a unique content identity, normally established by a cryptographic digest.
- An Evidence Graph is the signed evidence structure from which defined rules derive consequential state concerning a Digital Artifact after the evidence is validated.
- A Digital Original is a Digital Artifact with consequential state.
Content makes an artifact identifiable. Consequential state makes it an original.
Structure¶
An Evidence Graph may contain one record or a directed graph of records expressing:
- anchoring or issuance assertions;
- control and transfer;
- encumbrance and discharge;
- provenance or attestation;
- delegation;
- surrender, redemption, or termination; or
- another protocol-defined consequential relationship.
The Evidence Graph is evidence. A conforming implementation evaluates the record or graph as a whole under an identified, versioned ruleset. Consequential state is the result of that evaluation, not another record beside the Evidence Graph. An application may cache the result, but the cache is not the authority.
Every displayed consequential state should therefore be traceable to the Evidence Graph record or records that produced it.
The Evidence Graph records the consequential actions. State transition rules determine their consequences.
Consequential Action
|
v
End-verifiable event
|
v
Evidence Graph record or graph
|
v
Identified, versioned ruleset
|
v
Consequential State
Claims And Recognition¶
A provenance claim (“I created this artifact”) is distinct from an authority claim (“I recognize or certify this artifact for purpose X”). OpenETR can show who signed either claim and which artifact it concerns. Whether the signer has recognized standing or authority is a recognition-layer question.
Cryptographic validity does not itself create legal or institutional effect:
EVIDENCE GRAPH + IDENTIFIED RULESET
|
| validation
v
CONSEQUENTIAL STATE (result)
|
v
RECOGNITION
|
v
EFFECT
Legal Boundary¶
Evidence Graph is an OpenETR protocol construct. It is not a synonym for the Controllable Electronic Record defined by UCC Article 12. An Evidence Graph may qualify as a CER, electronic transferable record, negotiable cargo document, or another legally recognized record where the applicable legal requirements are satisfied.
Domain profiles may define the additional rules needed for bills of lading, warehouse receipts, or other instruments. They should not encode legal classification into the generic Evidence Graph merely by naming it.
Nostr Binding¶
Nostr provides signed events, identifiers, references, keys, relay transport, and redundant distribution. OpenETR defines event semantics, authorization, transition rules, conflict handling, and state derivation.
Nostr carries the events. OpenETR determines their consequences.
Relays preserve and transport Evidence Graph records. They do not determine OpenETR state.
Application Boundary¶
Implementations should separate:
ARTIFACT digest + content
Evidence Graph signed end-verifiable records or graph
STATE PROJECTION state derived from validated Evidence Graph records under defined rules
The state projection should be produced through one versioned derivation boundary conceptually equivalent to:
evaluate_evidence_graph(artifact_id, evidence_records, policy_id) -> ConsequentialState
If the answer to “Why is this state true?” is only “because the application database says so,” the implementation is not aligned with Consequential State Architecture.
Copying And Originality¶
Byte-identical files with the same digest represent the same Digital Artifact. Copying those bytes does not create another Evidence Graph and does not independently reproduce consequential state.
A copy can reproduce content. It cannot independently reproduce consequential state.
Governing Rules¶
- Consequential state should be derived from end-verifiable evidence according to defined rules, not merely asserted by applications, databases or blockchains.
- Applications derive consequential state; they do not own it.
- Recognition determines the effect given to consequential state.
- System failure need not become state failure.
- Digital Originals should outlive applications.
Terminology Migration¶
Evidence Graph replaces Digital Controllable Record (DCR), not Digital Artifact. The artifact remains the byte-identified content; the graph is the signed evidence concerning it. A single anchor can form a one-node candidate graph. Later events add evidence, not necessarily a control transition. The Control Graph remains the subset used by a ruleset to derive control consequences.
This is a terminology migration, not a wire-format or runtime API migration. Existing event kinds, tags, signed bytes, event IDs, and protocol behaviour remain unchanged. Legacy filenames and published URLs are retained for compatibility. The conceptual function above is illustrative, not an API name.
The minimum defined baseline is OpenETR Core Record Ruleset 1.0. It derives anchored state and a Publisher Position from an Anchor Event and qualifying Publisher Notices. Transfer, Current Controller, encumbrance, discharge, and similar state require an additional ruleset.