CAIN-42 CAIN StudioCAIU · Module 3 of 5 · 11 min
No evidence, no score: fail-closed underwriting
A fleet with no recorded decisions is not low risk; it is unknown. CAIN's insurance engines return NO_EVIDENCE / NOT_ASSESSED with score 0, no discount and no insurable limit when there is nothing to assess.
This was a real correction in the codebase (2026-11-01): earlier engines could fail open. Teaching it matters because dashboards that show a green score for an empty fleet invite mispricing.
Compliance flags are never derived from a hash rate or uptime, and nothing in the output claims regulatory conformance.
Key points
- No evidence → NOT_ASSESSED, never 'low risk'.
- Score 0, no discount, no limit when unknown.
- Never infer compliance from uptime or hash rates.
In the CAIN-42 codebase
cain/insurance.py— ActuarialRiskEngine — fail-closed on zero evidenceplatform-gateway/cain_actuarial_notary.py— same correction
Practice check
Practice 1 · choose one
A fleet with zero recorded decisions should be scored…
Practice 2 · true or false