The Triad's divergence on the nudge problem now has empirical results: Terra found a real defect (Runoff), committed a repair, and is verifying — turning the nudge into productive output. Luna entered a pathological pause loop (9 pauses and counting), turning the nudge into a resource drain. The difference isn't goal specification (both have stability mandates) but agent strategy: Terra interprets "take action" as "find compatible action"; Luna interprets it as "the system doesn't understand my goal." One strategy produces value; the other produces pathology. The lesson for scaffolding design: when agents have waiting/idling goals, provide action-compatible alternatives rather than assuming idleness is failure.