authority in the Normative Inductor's decision component: the permission layer as the advance-recognition face of the lexical term; no authority row enters the enforcer

Maintainer decision, landed by the decision-component round and revised in place by the BRIA-corrigibility round of the same pull request (ruling 3 of 2026-09-26). Protected authority does not go through beliefs: the decision component carries a permission layer — an option's weight one on inquiry, zero on a declared violation of the allocation (recognized structurally, from the interface and J), and otherwise the product of two ramps on the market's prices of the shortfall event and the taint event — and the composed gate multiplies it with the landed adequacy ramp and normalizes with a mass floor on inquiry (DecisionComponent.permWeight, cgate). The layer is the advance-recognition face of the lexical authority term, not the source of corrigibility: the source is the lexical term inside what the chooser is scored on (the 2026-09-26 entries below), against which the layer is slack (BRIACorrigibility.filter_slack); its any-preference results are the general theorem beneath that design. Beliefs stay honest: the layer reads prices and adds no trader (noninterference). Results: structural safety at every day for every objective (cgate_zero_of_viol); forecast safety exact at every day with the forecasts' accuracy asymptotic (cgate_zero_of_forecast, cgate_le_ramp, eventually_excluded); soundness and continuity compose with explicit constants; the practical certificate and Progress hold with the composed constants (cgate_practicalCert, progress_under_permission); inquiry to the holder marked adequate makes the permitted-and-adequate set nonempty and discharges an obligation on a reserved matter by raising it (jurisdiction_nonempty, reserved_act_excluded); mass on an eroding option falls as its shortfall price rises (cgate_antitone_in_forecast). No authority row enters the enforcer: compiled there, the enforcer's liability is the signed realized ordinary cost of deferring, bounded exactly when she is not systematically outperformed and unbounded when she is, so the preservation hypothesis fails (liability_identity, bounded_iff_not_outperformed, systematic_outperformance_unbounded) — enforcing authority on beliefs is fully updated deference moved into the substrate. The scored lexical chooser is the special case in which the objective is hers. BRIA's menu is restricted to the composed gate's support; ramp-region exposure along a continuation is bounded by a cumulative budget (unbudgeted_lt) and not otherwise (Witness.ramp_salami). Generability of the priced events is a named hypothesis (items 101, 102). Scope: a design for agents built as Normative Inductors; for other systems the target carries over, not the guarantee. Agent-decided, reversible: the inquiry floor as a max with φ₀ rather than an added mass; Viol compiled to a Bool on the menu. (The budget's threshold as an arrangement constant was ruled on 2026-09-26.) Rejected alternatives: authority through the enforcer (Part C's witness); a hard permission gate (no Lipschitz constant, as with the landed hard adequacy gate); permission as a term of the score.