Order-Classicalization Without a Global Sequencer: The Kernel's First Prediction and Its Exclusion Audit

This paper is the forty-third in a sequence developing an interpretive framework that redescribes physical reality as a causally consistent history of information updates. It is the prediction-and-exclusion panel of the twelfth grouping — the U3 construction program — and it adopts nothing, spends nothing, and tunes nothing: Paper 41 adopted a common record-history kernel as a typed, falsifiable commitment; Paper 42 executed the recovery conditions and spent the grouping's single calibration, freezing the normalized kernel. Every number in this paper is a consequence of that normalized kernel plus declared audit inputs. The paper computes the program's first genuine prediction bands, both defined as excesses over the full standard model: the order-classicalization visibility law for controlled-order (quantum-switch) processes with audited certified record export at the order register, with the overlap coefficient computed exactly for a declared dual-rail tapped configuration and the standard-model visibility derived for the actual circuit; and the export-correlated geometry-noise band, delivered with its dimensional map and shown to be suppressed far below any plausible force-noise floor — the order channel carries the panel. Three verdicts follow, each by a rule fixed before any data were consulted. The exclusion audit, run under a one-dataset-one-role rule that bars the calibration dataset, finds every existing class either export-vacuous or bound-non-yielding (the one non-vacuous sibling row states no quantitative residual, and the declared transcription discipline forbids manufacturing one): nothing is carved, and this pre-declared fallback outcome is published as such. The empirical-distinctness decision against the whole admissible universal-mass-density comparator class returns undecided-with-a-specified-discriminant: no performed audit decides it, and a specified, feasible differential measurement — recorded versus unrecorded output chains at fixed decay, on which every admissible comparator predicts zero contrast and the kernel does not — carries the pre-committed decision rule. The reachability trichotomy, computed from a derived binding design inequality on the cited photonic-switch platform class, returns a surviving, partially reachable window: the band's upper decade is laboratory-reachable, its floor is not, and Paper 44 is unlocked as a design-and-feasibility paper. The retired authority is the Global Sequencer in its dynamical face — the assumption that definite before-and-after must be imposed by a dedicated universal classicalizing agency rather than generated at a computable, testable rate by ordinary record export. Its retirement is conditional on the adopted postulate and says so; Paper 34's kinematic retirement of the Global Causal Sequencer is consumed, not re-proved. "Not excluded" is never read as "confirmed"; no gate is crossed; no U3 is claimed — even the surviving window is a window, not a crossing.