Polymorphic Operational Nonclosure: Carrier Transitions from Predictive Residuals to Thermodynamic Control Burden in Finite Observer Architectures

Abstract

Finite observers can fail to close a declared prediction task for several physically distinct reasons: incomplete records, inaccessible boundary data, permanent record-capacity deficits, delayed measurement context, reflexive response policies, or degradation of the resources needed to maintain a record. Earlier versions of the programme treated these obstructions as manifestations of one persistent uncertainty. Adversarial testing rejects that identification. Local positive residuals can disappear exactly after causal pooling; large finite systems can be represented by sufficient statistics; distributed records can close without a centralized copy; and coherent quantum memory can remove a classicalized delayed-context floor. This paper therefore introduces polymorphic operational nonclosure as a narrower replacement. Its central object is a typed, architecture-indexed obstruction profile rather than a conserved scalar uncertainty. A carrier transition is admissible only when the earlier obstruction is shown to close or become subdominant, the later obstruction is independently derived, the target and observer architecture are declared, a causal physical map connects the stages, and hostile controls exclude artificial withholding of available information. Exact information-theoretic results establish target-indexed residuals, pooling gains governed by conditional mutual information, protocol-specific quantum floors, and capacity-indexed target-family bounds. Mesoscopic and control-level tests show that maintaining closure can create a thermodynamic burden. Across broad feedback ensembles, terminal failure is conditional on renewal failing before the control margin vanishes. An explicit reciprocal quantum collision model supplies a finite CPTP regulator in which correction exports energy into a sink, sink loading reduces future correction capacity, and renewal supports recovery. The resulting framework is conditional rather than universal. It derives no observer-independent uncertainty scalar, unavoidable leakage field, spacetime-emergence theorem, additional gravitational stress tensor, or dark-energy contribution.