RFC 0002 — Evidence Model (E+ / E−)
Status: Draft Author: Martín Nicolás Sánchez Morales Created: 2026
Summary
This RFC formalizes the evidence model used by Janus Core Lite.
The model distinguishes between explicit recorded evidence (E+) and verified omission evidence (E−).
This distinction allows governance systems to reason not only from recorded facts but also from verified absences of expected records.
Motivation
Traditional systems treat missing records as undefined states.
In governance contexts, however, the absence of an expected record may itself represent a meaningful signal.
Janus introduces a formal distinction between:
- explicit evidence (E+)
- omission evidence (E−)
This allows systems to detect procedural failures, structural gaps, and governance breakdowns.
Specification
Explicit evidence (E+)
E+ represents evidence explicitly recorded in a system of record.
A claim supported by E+ references a recorded artifact that can be retrieved and inspected.
Omission evidence (E−)
E− represents evidence derived from a verified absence of an expected record.
E− requires:
- a defined expectation
- a deterministic search scope
- a reproducible observation that the expected record is missing
- referenceable schema context from SCHEMA_LOG
Evidence is evaluated by governance flows. Evaluation may produce governance events recorded in AUDIT_LOG.
Relationship with governance events
Omission evidence may generate the governance event:
OMISSION_DETECTED
Interpretation of evidence may require human evaluation, which is recorded through the event:
HUMAN_DECISION
Rationale
Governance systems must account for both presence and absence of records.
Without omission evidence, structural failures may remain invisible.
The E+ / E− model enables systems to reason about procedural compliance, missing actions, and institutional responsibility.
Backwards compatibility
Not applicable (initial evidence model specification).
Security considerations
Incorrect omission detection may result from:
- incomplete search scope
- inconsistent schema interpretation
- corrupted or incomplete logs
Implementations must ensure deterministic search scopes and schema consistency when deriving E−.
License
MIT License