Your system can be correct
and still be illegitimate.
Decision Integrity is the governance condition under which a system remains the same legitimate decision-bearing system over time. It is not identical with correctness, compliance, monitoring, or model quality. It is the condition that existing standards do not verify.
Not: is the system working?
But: is it still allowed to?
A system can remain behaviorally correct, statistically stable, and formally compliant — while the decision it is executing is no longer the decision that was approved. The mandate has changed. The authorization basis has expired. The assumptions that justified operation no longer hold.
No existing standard asks this question. Compliance asks whether rules are followed. Monitoring asks whether outputs are anomalous. Safety asks whether harm can be prevented. None ask whether the decision currently being executed is still the same authorized decision it was at approval.
This is the governance gap. This is what Decision Integrity closes.
Identity
The defining attributes of a decision — its authorization basis, scope, and assumption set — must remain intact throughout execution. When identity changes, continuation is no longer the same authorized act.
Continuity
Legitimacy must persist during execution, not only at approval time. The moment of authorization is a point in time. Execution is a continuous process. Decision Integrity bridges the gap.
Witnessability
Claims about identity and drift must be anchored in witnesses — auditable references that substantiate governance objects. Without witnesses, governance remains narrative rather than verifiable.
Re-legitimation
When identity-relevant conditions change, continuation requires an explicit governance act — not silent adaptation. Re-legitimation is the mechanism that keeps execution legitimate as conditions evolve.