ACAS: A Sealed Admissibility Instrument for Determining When Measurement Exists

ACAS (“A Sealed Admissibility Instrument for Determining When Measurement Exists”) specifies a deterministic, replayable procedure that classifies whether a proposed measurement is well-posed under sealed admissibility conditions—or terminates with a stable refusal code. The manuscript is explicitly an instrument specification: ACAS is not a theory of meaning, ontology, persuasion, optimization, or governance; it outputs only a bounded computational outcome. ACAS ingests a sealed extraction bundle (target declaration, extraction procedure, boundary specification defining “measured” vs “blinded” conditions, environment descriptor, and admissibility parameters) and emits exactly one Single Output Normal Form (SONF): either ACAS::MEASURED{…} with hash commitments and (when applicable) an IRX value, or ACAS::REFUSED{…} with a single terminal refusal code selected by a total precedence order. The refusal registry is finite, normative, hash-committed, and designed to prevent “repair by narrative” when failures arise from bundle invalidity, non-determinism, leakage, extraction ambiguity, or metric undefinability. The core global consistency statistic is IRX, defined as a probability of vector-rank discrepancy between the measurement object’s rank under measured vs blinded execution conditions. IRX is computed only after sealing, determinism, typing/rank definability, and boundary/leakage obligations pass; otherwise ACAS refuses rather than imputing values. The result is a reviewer-replayable decision surface: measurement exists under the sealed conditions, or it does not (and is refused), with auditability enforced by canonical hashing and replay commitments. Keywords (Zenodo “Keywords” field)admissibility; measurement existence; refusal semantics; reproducibility; sealed extraction bundle; blinded evaluation; leakage detection; determinism; auditability; hash commitments; single output normal form; rank-based invariants; IRX; well-posedness