Signal Enrichment (Optional)
This section describes a possible Signal Enrichment Layer. It is not part of the current authoritative engine documentation (see ENGINE-DOCUMENTATION.md). Implementations may or may not include enrichment.
BaseCred may introduce a Signal Enrichment Layer between normalized signals and rule evaluation. The purpose would be not to create new scores, but to extract additional meaning from:
- Agreement between independent systems
- Disagreement between independent systems
- Missing or asymmetric signals
Semantic Roles of Providers
| Provider | Primary Authority |
|---|---|
| Ethos | Long-term trust & credibility |
| Talent Protocol | Ability & skill (builder/creator) |
| Neynar | Social behavior & spam risk |
BaseCred never treats any single provider as complete.
Agreement-Based Enrichment
When multiple providers agree positively, confidence can be increased (e.g. in explanation or eligibility for full ALLOW vs probation).
Disagreement-Based Enrichment
Disagreement can be treated as signal, not noise (e.g. high ability + low trust → ALLOW_WITH_LIMITS).
Enrichment Output
The enrichment layer would produce contextual signal interpretations consumed by the Rule Engine. No raw scores or provider-specific values would be exposed.