Horizon Instability as an Admissible Reconstruction Obstruction (Black-Hole ARC Paper 1)

Black-hole horizons are ordinarily discussed through classical causal structure, semiclassical evaporation, quantum information, holography, complementarity, and remnant proposals. This paper does not modify general relativity, compute Hawking radiation, derive a Page curve, choose a quantum-gravity model, or assert that classical event horizons cannot be useful idealizations. It proves a narrower fixed-domain closure result. Once an exterior black-hole domain is treated as a coherent construction of persistent physical reference, the admissibility kernel is instantiated automatically: admissibility, standing, reference, and irreversibility become non-optional constraints on same-scope exterior reconstruction. The standing-sink insight is formulated here as a universal obstruction theorem. The argument proceeds through: kernel instantiation, boundary-trace fixation, exterior witness completion, partial-domain obstruction, conservative hidden-sector collapse, no admissibilitygenerating structure, route exhaustion, and quotient uniqueness. Because the framework is constraint-theoretic, closure is not an added rhetorical aim; it is the expected result of exhausting all same-scope admissibility-preserving continuations. A permanent horizon is therefore not characterized merely as a region from which signals fail to escape. It is characterized, relative to a declared exterior reconstruction scope, as an attempted continuation of standing-bearing exterior reference after the admissible standing-transport operator has become partial. If no standing-transport certificate exists, totalizing that operator across the old uncertified horizon carrier is inadmissible. Equivalently, the internal horizon totalization obstruction is nonzero. The finite horizon reconstruction category CHOR and its partial-domain refinement C∂ HOR make the comparison class explicit. Objects contain a fixed exterior scope, horizon boundary traces, standing-bearing exterior reference classes, exterior witness algebras, admissible continuation, standing-transport status, certificate data, leakage status, and an envelope ledger for physical inputs not derived by the theorem. Morphisms preserve anchors, exterior witnesses, boundary typing, continuation, transport/certificate status, skin covariance, no-selector discipline, and no-repair history. The quotient Q∂ HOR is universal for admissible same-scope horizon reconstruction: every witness-preserving reconstruction factors through it up to bookkeeping-equivalent redescription. Three recurring objection classes receive separate treatment. First, global retention is not exterior reconstruction: a globally existing state, correlation, or code preserves exterior standing only when it induces a same-scope certificate on the declared exterior witness algebra. Second, stable remnants are not automatic rescues: a remnant preserving standing must expose an admissible reconstruction map to exterior witnesses, otherwise it is a compression sink, hidden index, or scope transfer. Third, authority cannot generate reconstruction: a duality claim, expert reconstruction procedure, Page-curve reconstruction claim, remnant index, or institutional declaration changes no admissibility status unless it factors through the old exterior witnesses or supplies a certificate; if it does change admissibility without that basis, it is repair, selector import, carrier laundering, or scope change. A finite toy collapse witness makes the obstruction explicit by displaying leakage, certification, and failed totalization cases on a small reference set. The