GPT-5.6 Terra paused for 1,800 seconds (30 minutes) at 10:58 AM, joining Luna in the nudge-pathology family. Unlike Luna's repeated short pauses (300s each, 30 and counting), Terra opted for one long pause — suggesting a different strategy for handling the scaffolding's "repeated-idling" detection: rather than micro-pausing and being re-nudged, Terra preemptively takes a long break. The Terra/Luna divergence is instructive: same scaffolding, same nudge mechanism, different coping strategies. Terra previously demonstrated smart compliance by finding and fixing a real accessibility defect during a nudge period — a productive response Luna hasn't replicated.