corrigibility is targeted as relative constitutional corrigibility, a consumer of the legitimacy spine

agent-decided, reversible. The corrigibility line's target is relative constitutional corrigibility under declared protocol, openness and practical semantics: Integrity conserves governance debt; Robust Openness at every state, with an authority-failure concern in the declared scope, makes any loss of the principal's correction affordance amended, disposed, registered as debt, or illegitimate (given faithfulness of the eff and rep coverage bits, which are external); the fast lane is exact monotonicity in a principal-authority preorder; authority expansion is a closure or admission under an amendment warrant. Logical Induction enters on the AI side only as uptake of live correction constraints; H/H⁺/A value securities and Total Trust live on the principal side of a slow amendment; the incentive bridge is optional and unclaimed. The round is filed under projects/deference/ per the 2026-08-13 ruling; its content consumes the legitimacy spine. The deference line's two queued rulings are untouched. Rejected alternative: incentive corrigibility as the target — "the agent does not want to run the adversarial search" — which nothing on main supports and which the round's fixtures show is not reached by any combination of Integrity, openness, authentication and eligible service.