Why a system can be correct and still not be the same decision.
A system can produce the same output format, pass the same checks, and still no longer be operating on the same decision it was authorized for. This page explains that idea in plain terms — not how SnapOS's private methodology tests for it.
Is this still the decision that was approved?
Most reviews ask whether a system works. This question is different: whether the decision it is executing today is still the same decision that was originally approved — under the same assumptions, the same evidence, and the same authority (the documented reference for who was entitled to approve, continue or override this specific decision, at that specific time and scope — not organizational rank or legal power in general).
A system can remain behaviorally correct while that identity has already shifted underneath it — because the assumptions it relies on quietly changed, or because the person or role that approved it is no longer the one relying on it today.
Authority reference
Who was documented as entitled to approve this decision, and does that reference still match who relies on it today.
Assumption set
What conditions were assumed true at the time of approval, and whether they are still the conditions in force.
Evidence base
What documented support existed for the decision, and whether it still covers the case being decided today.
The full technical specification — including how these are formally modeled — is published as a citable, timestamped research artifact, not as an operating manual on this page.
Conformance Suite
The public output contract: what a reviewable finding must contain.