Signal Enrichment
BaseCred introduces a Signal Enrichment Layer between normalized signals and rule evaluation.
The purpose is 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, BaseCred increases decision confidence.
Example:
- Ethos = NEUTRAL
- Talent.Builder = EXPERT
- Neynar = VERY_HIGH
Interpretation: Capable and well-behaved actor with low overall risk.
This affects:
- Decision confidence
- Explanation clarity
- Eligibility for full ALLOW vs probation
Disagreement-Based Enrichment
Disagreement is treated as signal, not noise.
Examples:
- High ability + low trust → ALLOW_WITH_LIMITS
- High social trust + low ability → context-dependent
This enables nuanced outcomes rather than binary rejection.
Enrichment Output
The enrichment layer produces contextual signal interpretations consumed by the Rule Engine.
No raw scores or provider-specific values are exposed.