Trust stack
| Layer | Meaning | Where it comes from |
|---|---|---|
| Relationship strength | Directness and quality of a specific edge | Edge strength and interaction evidence |
| Evidence chips | Human-readable proof points | Provenance, meetings, emails, communities, intros |
| TrustRank | Composite signal for a person or route | Graph scoring plus execution history |
| Graph confidence | How safe it is to act on the route now | Edge confidence, freshness, weak links, path quality |
What should raise confidence
- recent interaction evidence
- prior successful introductions
- clear connector-to-target context
- multiple proof points that agree with each other
- low decay and low ambiguity in the route
What should lower confidence
- stale relationships with no recent evidence
- routes held together only by weak overlap
- unclear connector intent or availability
- a single fragile edge inside an otherwise weak path
- missing proof about why the route should work now
What the product should show
Every path or connector score should show:- the score
- the reasons it ranked there
- weak-link diagnostics
- freshness or decay cues
- the next best action when confidence is limited
Confidence rules
- strong observed evidence should raise confidence
- stale edges should visibly decay
- inferred context can support ranking but should not masquerade as proof
- unknown targets should not produce fabricated warm paths