the legitimacy→deference interface is an activated evaluation occurrence, read occurrence-locally

agent-decided, reversible. An advisor's deference target is the actual future principal's evaluation, admitted at time n as an anchored evaluation occurrence in the existing account calculus (EvalReq is a value of Req, not a constructor). The activation event is OccurrenceIntegrity.Program.activated: the occurrence's propagated account is exactly one authenticated answer leaf, so closed and live never activate and a second receipt cannot enter once one has (fates_subst_of_terminal). The value vector is the payload of the authenticated history at the answer receipt's event. The lineage condition a deference consumer takes is occurrence-local: AuthorityActivation.LegitimateForSegment Γ_n, the same Integrity evolution with Robust Openness required only for the concerns declared relevant to the occurrence — a projection of LegitimateSegment (LegitimateSegment.project), not a second theory; Witness.unrelated_failure is the trajectory on which the global hypothesis voids an intact evaluation. Certification is payload-blind up to a total binding relation (Neutral.certifiable_iff); the authorship meaning of the process receipts is external. Ordinary LI Value runs on the activated securities C_n · V_n(a) unchanged, with the no-preview receipt as its admissible-domain condition, and the transfer to regret against the actual principal is ε + η exactly (ActivatedValue.availability_transfer). Robust Openness is recorded as the access half of non-capture; authorship is a separate external contract consumed by deference and not a conjunct of legitimacy. Rejected alternatives: a counterfactual reference evaluator as the value target (fixture A); global Legitimate O_n O_m as the consumer hypothesis (fixture H); a "resolved" (answered-or-closed) activation (fixture E); per-candidate activation (fixture D); a ContinuationWarrant / AuthorityLineage primitive (nothing it would carry is missing from the account).