A classification of the (e5) recoverability question and its candidate class, not a certification of any recovery. Paper 61 typed a seat for "recoverability from horizon or static-patch degrees of freedom" and answered nothing — it kept the clause [record]-and-open, registering that the seat "would require a declared algebra of those degrees, a decoding relation, and a provenance/warrant condition; none is supplied here." Paper 71's Strong-Field Completion Charter locked the constructive successor — panel P-D, this paper — under its element-6 hard lock (owner-approved, verbatim): "P-D is locked unless a named decoding-relation candidate exists at its theme time; absent that, P-D publishes the missing-leg finding — an honest scoped/killed landing, never a silently solved gap." This paper consumes that charter unmodified and adjudicates the lock. The adjudication is against a content criterion. A DecRelCand — a named decoding-relation candidate — is any object declaring a map from an algebra of horizon or static-patch degrees of freedom that recovers held records from it (an admissible one, DecRelSpec, adds compatibility with Paper 61's ledger). An audit of the declared challenge set of recovery/decoding-map structures in the ALAC corpus and the cited strong-field literature — general and holographic quantum error correction, entanglement-wedge reconstruction, Hayden-Preskill decoding, islands / quantum extremal surfaces, the static-patch algebra, the corpus accounting-layer decoders, Paper 60's already-barred global-purification decoder, the general recovery formalism, and the Gate-VII rubric — refutes each by a named field: wrong domain, wrong codomain, barred evaporation/holographic physics, or an algebra supplied with no decoding map. The candidate set is therefore empty at this paper's theme time. The (e5) recoverability-relation declared class is uninhabited, and P-D publishes the missing-leg finding: CompletionStatus(recoverability relation) = killed via K0 — the decoding leg registered absent, never solved. Certified is structurally unreachable (the operative leg is absent); scoped is definitionally unreachable (no recoverability-relation instance is constructed). Forgetting the machinery returns Paper 61's registered ledger classification exactly. The leg reopens iff a decoding candidate arrives; the finding never enters P-C2. No endpoint physics; nothing adopted, calibrated, or scored. The pivotal near-miss is the static-patch / de Sitter observer algebra (Chandrasekaran-Longo-Penington-Witten 2023; Chandrasekaran-Penington-Witten 2023; Witten 2022): a genuine declared algebra of the very degrees of freedom the (e5) seat names — but it supplies no decoding map recovering Paper 61's held records, so a leg-1 algebra without a leg-2 map is not a decoding-relation candidate. Possessing an algebra of observables, or a state/evaluation functional on it, is not yet a decoding relation in the present sense. The foil retired is the Global Recoverer — the authority that would grant strong-field recoverability by default, the decoding step assumed rather than exhibited — retired preemptively as a guard against a future overread, since the corpus does not currently overclaim: Paper 05 marks recoverability conjectural, Paper 61 keeps the clause [record]-and-open, and Paper 60 already displayed and barred the one global decoder it named.
A classification of the (e5) recoverability question and its candidate class, not a certification of any recovery. Paper 61 typed a seat for "recoverability from horizon or static-patch degrees of freedom" and answered nothing — it kept the clause [record]-and-open, registering that the seat "would require a declared algebra of those degrees, a decoding relation, and a provenance/warrant condition; none is supplied here." Paper 71's Strong-Field Completion Charter locked the constructive successor — panel P-D, this paper — under its element-6 hard lock (owner-approved, verbatim): "P-D is locked unless a named decoding-relation candidate exists at its theme time; absent that, P-D publishes the missing-leg finding — an honest scoped/killed landing, never a silently solved gap." This paper consumes that charter unmodified and adjudicates the lock. The adjudication is against a content criterion. A DecRelCand — a named decoding-relation candidate — is any object declaring a map from an algebra of horizon or static-patch degrees of freedom that recovers held records from it (an admissible one, DecRelSpec, adds compatibility with Paper 61's ledger). An audit of the declared challenge set of recovery/decoding-map structures in the ALAC corpus and the cited strong-field literature — general and holographic quantum error correction, entanglement-wedge reconstruction, Hayden-Preskill decoding, islands / quantum extremal surfaces, the static-patch algebra, the corpus accounting-layer decoders, Paper 60's already-barred global-purification decoder, the general recovery formalism, and the Gate-VII rubric — refutes each by a named field: wrong domain, wrong codomain, barred evaporation/holographic physics, or an algebra supplied with no decoding map. The candidate set is therefore empty at this paper's theme time. The (e5) recoverability-relation declared class is uninhabited, and P-D publishes the missing-leg finding: CompletionStatus(recoverability relation) = killed via K0 — the decoding leg registered absent, never solved. Certified is structurally unreachable (the operative leg is absent); scoped is definitionally unreachable (no recoverability-relation instance is constructed). Forgetting the machinery returns Paper 61's registered ledger classification exactly. The leg reopens iff a decoding candidate arrives; the finding never enters P-C2. No endpoint physics; nothing adopted, calibrated, or scored. The pivotal near-miss is the static-patch / de Sitter observer algebra (Chandrasekaran-Longo-Penington-Witten 2023; Chandrasekaran-Penington-Witten 2023; Witten 2022): a genuine declared algebra of the very degrees of freedom the (e5) seat names — but it supplies no decoding map recovering Paper 61's held records, so a leg-1 algebra without a leg-2 map is not a decoding-relation candidate. Possessing an algebra of observables, or a state/evaluation functional on it, is not yet a decoding relation in the present sense. The foil retired is the Global Recoverer — the authority that would grant strong-field recoverability by default, the decoding step assumed rather than exhibited — retired preemptively as a guard against a future overread, since the corpus does not currently overclaim: Paper 05 marks recoverability conjectural, Paper 61 keeps the clause [record]-and-open, and Paper 60 already displayed and barred the one global decoder it named.
Paper 55 of this series proved that the certified record-history kernel composes foliation-independently at the displayed orders under D0 — undressed update operators, Kraus operators in mutually commuting spacelike algebras — and relocated the dressed case in both directions: "dressed update operators (beyond D0) — the class-boundary question — not settled here in either direction: physical dressed operators fail exact spacelike commutation (Donnelly & Giddings 2016), so whether they reintroduce ordering dependence is exactly what a dressed construction must decide." Paper 57 built that dressed construction at declared anchor classes. This paper — the centre panel of the Shared-Frontier Program (Papers 57-59), consuming the charter Paper 57 defined — makes it decide at the kernel's own grade: the classical-quantum class the corpus actually runs (geometry classical at tree level), with the quantum-field dressing half of the question excluded at the definition, not at the verdict, and seated at (d5). What is decided is the kernel-grade half of the question; the paper's thesis, annotations, and verdict all carry that qualifier. The audited authority is the Global Integrability Oracle: the guarantor that the world's many dressed local updates compose consistently, assumed rather than earned. The paper replaces the pronouncement with a computation. Delivered: the dressed slab calculus — Paper 55's one-node factorization lifted to the package, with the dressing correction displayed per factor in three typed components (support re-anchoring, collar-bounded; the retarded dressed-response correction; the constraint-tail conditioning at the kernel's classical-quantum grade), complete positivity, adaptedness, and localization proved, with the common-domain fact typed at the channel level (apparatus channel type vs support realization — no canonical local-algebra element identity asserted across anchorings, consistent with Paper 57's sandwich); the residual theorem — the transposition computation extended to the full dressed pair table (sixteen cross-factor pairs, displayed as a 4x4 matrix with an order dictionary, each computed, plus the surface-tied booking schemes whose mid-composition differences cancel at the endpoint) — landing every in-class same-endpoint spacelike transposition residual in the zero cell of the pre-committed typology {zero, pure-convention, order-effective}, strict map identity for every admissible representative, the trichotomy exhaustive and decidable via a fully specified dressing-convention quotient; the representative-equivalence lemma owning the presentation content (the pure-convention cell is inhabited by presentation data, not endpoint residuals); Paper 45's (o4) operator families entering as first-class rows (their composition-facing half decided — no self-adjointness consumed; their measurement-scheme half returned open-named); the stability theorem (Paper 55's Theorems 1-2 persist under dressing at both equality grades); an out-of-class witness displaying the order-effective cell's shape (a quantum-tail toy phase, with the Weyl-convention note: the factor-2 ordering ratio is convention-independent in magnitude); and exact recovery of the undressed calculus with the Model M band unchanged. The verdict is scoped, not certified: conjuncts (d1)-(d4) close at the displayed orders on the declared classes; (d5) — iterated feedback, all-orders composition, the nonperturbative classical-quantum constraint sector, and the quantum-field dressing boundary — is scored open and built elsewhere. FrontierStatus(composition) = scoped at model grade. Paper 55's (t5) closes the kernel-grade half of its dressed-operator row by annotation; the Oracle is unemployed: consistency was computed, not pronounced. Nothing is adopted, calibrated, or scored; no band moves; the substantive integrability-of-nature face is declined, not refuted.
Paper 34 typed quantum-indefinite causal order for controlled-order processes and told the truth about the rest: full quantization of the causal-precedence relation is "flagged, not claimed," and "no microscopic or holographic account of a quantized precedence is supplied." Paper 57 registered the stay-reason: the radical case "needs a dressed order-sector algebra beyond (r1)'s declared classes; a program of its own." But the corpus has no checkable specification of what counts as a candidate for that program's central object. While none exists, a quiet authority works by default — the Global Chronologist: not the Gatekeeper deciding candidate admission, not the 34/43 Sequencer imposing or dynamically selecting an order, but the imagined authority that supplies a canonical chronology/order-sector algebra and composition law from outside the candidate class. This paper types the candidate layer and starves the Chronologist. "An order sector is a type, not a timeline." / "Typed is not constructed." / "Admissible-as-typed is not selected." The instrument is the OrdSecSpec normal form — "Dressed here names the target class of a future algebra; this paper does not construct dressed operators or dressed composition." Seven declared fields on a finite declaration language (Paper 69's discipline consumed as a template, all order-sector objects redefined locally): a source-data grammar O = (E, L, C_ord, R_ord, Gamma_ord) whose type constraint bars external chronology — global time functions, canonical foliation labels, pre-given total orders — while allowing internally declared order-control and order-record data (with a three-scan renaming test, so a global clock smuggled as control data is caught structurally); a sector codomain Sigma_sec = {coh, mix, cond} plus the typed residual outside-ctrl, one-to-one with 34's trichotomy on the declared controlled-order subdomain — general nonseparable processes are never forced into the trichotomy; a composition/conditioning grammar that provides symbols and reduction targets but no composition theorems (no associativity theorem, no spacelike-independence theorem, no measurement model, no all-orders dressed composition result); a reduction map rho^ord onto 34's conditioned record-relative effective order, the canonical chronology appearing only in its codomain; four recovery rows (34's machinery; 57's fixed-order limit; 58's undressed limit; and 34's R-E — radical-precedence remains flagged, not claimed — preserved as a row, not resolved); and skeleton-grade failure criteria. The admission test is decidable — four finite checks with a verdict routing, including the order smuggling sentence: "a candidate fails the admission test if a composition or conditioning clause reads an external chronology label before the reduction map is applied." The declared candidate table — the process-matrix/quantum-switch family, the causal-set row (typed only as a source of order data and order-composition discipline, with every adequacy claim barred), the indefinite-causal-structure family, and one corpus-internal calibration row (34's own machinery, which cannot feed the kill) — is typed by displayed fields under 69's extraction protocol and setting/input rule: "The paper classifies the families by displayed fields, not by reputation or expected success." The emptiness kill is scoped to skeleton grade, with the minimum skeleton displayed and one row, the process-matrix/switch row, proven to meet it at skeleton grade. The ceiling is strict: nothing here is a construction or an audition, and no row is a winner — 34's flag is exactly as set after this paper as before it; (d5) and (r5) stay open at their owners' seats; the preparation artifact is recorded as complete without opening, numbering, or reserving any successor program; no physics claim, no adoption, no datum. OrdStatus(order-sector classes) = scoped at typing/classification grade. "An order sector is a type, not a timeline." / "Typed is not constructed." / "Admissible-as-typed is not selected."
This paper is the seventh in a sequence developing an interpretive framework that redescribes physical reality as a causally consistent history of information updates rather than as a collection of objects independently existing within spacetime. Two prior results set up its question. The third paper (Uchida, 2026c) characterized public finality—recoverability of a committed fact by many mutually independent accessors, non-disturbingly and repeatably—as a broadcast redundancy: an effectively classical pointer value proliferated across many disjoint environmental fragments, approximately in the form of spectrum broadcast structure. The fourth paper (Uchida, 2026d) read the evaporating black hole as the regime in which a committed interior record crosses, at the Page time, from hidden to recoverable encoding, reconstructible from the radiation through entanglement-wedge reconstruction; and the parent framework (Uchida, 2026a, §10) had flagged holographic quantum error correction as a place its language of protected, redundantly encoded, locally recoverable records might find content. The present paper makes that content precise as a structure theory of recoverable records, kept separate from any claim of new physics or any suggestion that the framework derives the underlying results. The central claim is conditional and structural. The framework does not claim that every recoverable piece of information is a code, that the universe is literally a quantum error-correcting code, or that AdS/CFT describes our universe. It claims that when information qualifies as an erasure-recoverable record because an encoding preserves a logical degree of freedom—or a logical operator algebra—against a specified family of physical erasures, then that recoverability is formalized as an exact or approximate quantum error-correcting structure, in which each correctable erasure leaves an authorized complementary region supporting a recovery map. That formalization is itself an established mathematical classification; the paper's interpretive contribution is to read the series' recoverable record through it, and to set it against a second, distinct architecture. Broadcast redundancy makes the same effectively classical value legible from many disjoint fragments; erasure-correcting redundancy protects noncommuting logical quantum information by keeping each correctable erased region ignorant of—approximately private with respect to—the protected logical content, while suitable remaining regions recover it. The formal engine of this distinction is the duality between correctability and privacy (the information–disturbance tradeoff), not the no-cloning theorem; no-cloning, with the access structure of quantum secret sharing, survives as the reason independently usable recovery regions for arbitrary unknown logical quantum information must overlap. Three commitments keep the reading honest. First, it separates three senses of "record"—protected logical quantum information, a committed classical record, and a publicly final record—and insists that recovery or decoding restores access to logical quantum information but does not, by itself, commit a classical value; commitment additionally requires measurement, pointer imprinting, decoherence, and stabilization (Uchida, 2026f), and public finality a still later broadcast. Second, holographic quantum error correction is offered as the sharpest worked model of the distinction, with exactly solvable tensor-network toy codes (exact within the model) kept distinct from realistic semiclassical holographic reconstruction beyond those limits (generally treated as approximate), and both from the further question of state-dependence. Third, code-recoverability does not by itself imply public finality: protected quantum information can be recoverable from authorized regions while remaining non-public. The framework relabels code structure as the architecture of recoverability and derives none of it, making no independent empirical prediction. Its disciplined thesis is that codes protect recovery; broadcasts create publicity—different architectures that a single physical history may nonetheless realize in sequence.
After an experiment runs, exactly two decisions could still be made by a person: does this outcome exclude the theory's declared region, or did an auxiliary — source, tap, detector, calibration — fail, voiding the run instead; and who is licensed to say so. Any protocol that leaves either decision open after data exist has left a seat for a post-hoc refutation authority — the Global Falsifier, the imagined adjudicator who owns the modus tollens. Paper 44 of this sequence closed its photonic tapped-switch probe F_A at pre-registration-grade design ("the IV gate itself remains future laboratory work"); Paper 48 froze the shared adjudication layer as a handover template, typing everything design-specific as named pre-data slots behind a failable acceptance gate. What stands between that design and a laboratory's first shot is a single object: a completed, gate-passing handover instance. This paper — the forty-ninth in the sequence, centre panel of its Execution Program (Papers 48-50) — builds that object, the reference instance H_A, at the executable level the template promised. Delivered: the delta specification from the cited apparatus class (the Goswami-type photonic quantum switch) to 44's tapped configuration, component by component against the frozen budget rows; the schema completion with field-level primitives for every field, extensions add-only and partition-admissible; the frozen analysis pipeline in algorithmic normal form — IV gate, measured-calibration transfer, estimators, and the leaf router implementing 48's tap rows verbatim, under a language-independent rounding convention so floating-point artifacts cannot flip a frozen boundary direction — with the reference implementation shipped as a hash-anchored code appendix (deposited companion file) behind a declared production-scoring boundary; the concrete exclusion rule list with its veto-impact report and a frozen post-exclusion N rule disposing shortfall, surplus, and stopping at once; the S1-S4 blinding runbook; and the mock-data challenge, executed in three parts — a deterministic core exhaustive over every router branch and both tie directions, an adversarial set generated through a recorded, deposited exchange by a party in the auditor role that did not implement the scoring code, and a public-beacon-seeded bulk for totality and determinism. All three parts pass; the routing tables are generated by executing the shipped implementation and are typed non-evidence in the sentence that reports them. The acceptance gate is then run condition by condition on the deposited pair. One finding is published along the way: 48's coverage condition, read literally, would force every per-design instance to claim all seven leaves; the pre-committed completion evaluates coverage over the design's assigned leaf subset plus the shared non-execution leaf — an annotation to 48, never a retro-edit. The verdict is InstanceStatus_A = reference-complete, with an empty instance gap register: every gate condition passes, and what remains is ownership — acceptance transfers when a laboratory signs, auditor-reproducibly, over the instance plus its re-gated extensions; reference-complete never shortens to accepted, and an immutable core guarantees that no admissible extension can change what a null excludes, who may say so, or how boundary cases route. Displayed at the close is the channel's frozen exclusion content — the carve a future null already names, [1.7e-1, 2.1] at the conservative corner — beside the pre-priced Duhem escape map. Nothing is carved, scored, or measured; no band or feasibility verdict moves; the data paper 49-D stays reserved behind its unlock conditions. What retires is the interpretive seat: refutation content as a public, frozen, executable object — falsifiability without a Falsifier.
Paper 44 of this sequence closed its U3 construction program at readiness: two budgeted experimental designs — a photonic tapped-switch probe and a circuit-QED recorded-versus-unrecorded differential — and an execution map, a pre-registered decision tree assigning every future measured outcome its pre-committed consequence, with scoring left "by whoever holds the data." Paper 47 then closed the theory-side debt ledger, leaving the executed-measurement route as the corpus's only live route to a framework-forced world-side shift. What 44 did not freeze is the layer between data-in-hand and leaf-adjudicated: shot schemas, exclusion policy, blinding order, stopping rule, claim authority. That unfrozen layer is where a post-data analysis authority could still sit — the researcher degrees of freedom and forking paths of the methodological literature, answered in precision physics by blind analysis and pre-registration. This paper — the forty-eighth in the sequence, opening panel of its Execution Program (Papers 48-50) — freezes exactly that layer, before any laboratory exists, as the template of a handover package executable and adjudicable by strangers: every analysis decision the package will ever need is either frozen here or typed as a named pre-data slot behind a failable acceptance gate, so that once a design-specific instance is accepted, no one — including the author, who operates no laboratory and has made that fact load-bearing — retains a seat from which a future outcome could be re-interpreted. Four document-level layers are delivered. Record layer: field-level shot schemas for both designs, with every field and every raw source typed into an audit or scoring partition; an add-only extension rule; ledger, provenance, and dataset-role conventions; and an exclusion grammar under which every excluded run is a first-class recorded event and every exclusion predicate is outcome-unread by construction. Statistics layer: the estimators, the null-carving formula, a leaf-decision normal form (frozen statistics, thresholds, tie rules per leaf), and the stopping rule, inherited verbatim from Papers 43/44 with the arithmetic re-checked; and a four-stage blinding architecture ending in a scoring stage at which everything is readable and nothing is choosable. Authority layer: a claim-authority table over the map's seven leaves and four post-data speaking roles — the data holder scores; the series designer holds no scoring cell; the decisive leaf is licensed only to the data holder's own successor paper — plus the handover normal form H with a seven-condition acceptance gate, a data-existence certificate with a public deposit obligation, a mock-data challenge typed as non-evidence, and pre-committed unlock conditions for the reserved data papers 49-D/50-D. Verdict layer: a six-check closure audit (leaf coverage; arithmetic consistency; partition admissibility; blinding soundness; exclusion soundness; authority totality), each check mechanical against the delivered artifacts, with the pre-committed either-way verdict ProtocolStatus in {closed, gapped} scoring template closure — an executable package exists per design only as an accepted H-instance. The audit is run and every check passes: closed at the template layer, quarantine register empty. Papers 49 (tap-channel exclusion protocol) and 50 (recorded-unrecorded differential protocol) are unlocked to instantiate H per design; only the unlock-conditioned data papers 49-D/50-D will ever score a datum. The retired authority is the Global Statistician — post-data analytic discretion personified: its authority reading is retired by construction (no analysis decision has a post-data seat left to occupy, the per-design property conferred by each accepted H-instance), its innocence reading by the imported methodological record, and its substantive face — one uniquely correct inference framework — is declined, not refuted. Nothing world-side moves: no datum is scored, no band or floor shifts, the feasibility verdicts stay 44's, U3Status is restated not re-scored; if no laboratory ever executes, the program rests where 44 left it — undecided, with an implementable and now administrable discriminant — a complete record, not a failure.
This paper is the forty-second in a sequence developing an interpretive framework that redescribes physical reality as a causally consistent history of information updates rather than as a collection of independently existing objects. It is the recovery panel of the twelfth grouping — the U3 construction program — and it adopts nothing new. Paper 41 adopted, at the claim level L3-A defined there, the common record-history kernel P_ker (one classical-quantum GKSL-type generator whose fixed Lindblad operators are export-event operators driven by a classical, adapted record-export intensity), computed a nonempty admissible coefficient space, and left four obligations to this paper: the recovery conditions RC1/RC2a/RC2b and the single calibration RC3, under a binding ledger (one calibration, spent on lambda_dec only; lambda_diff carried as a gate-bounded prediction band; thresholds never re-tuned). This paper executes them. RC1 is proved as a theorem: in any sector containing no certified export channels, the kernel's matter/order dynamics is exactly unitary — a sector statement, not a coefficient limit, and not a claim that unexported mass ceases to gravitate. RC2a is proved as a bounded-residual theorem: the Newtonian baseline drift is fixed by definition to the imported Newtonian classical-quantum mean dynamics, the export correction is shown to be cell-mean-zero by the decomposition bookkeeping (with the adopted sourcing law's first moment checked — a real failure mode, not realized), and the residual mean-source shift is dipole-suppressed below every declared class tolerance delta_g(class); the honest status row records that no current class co-locates audited export density with a Newtonian-verification measurement, so the world-side probe is today bounded, not probed. RC2b is split in advance: an executed compatibility check — the Jacobson-route hypotheses of Paper 36 reduced row by row to declared criteria with declared tolerances, of which the normalization row is satisfied (G an input, never an output), the entropy-monotonicity row is satisfied under a displayed record-basis pinching condition, and the local-equilibrium and Clausius-defect rows are fixed as criteria whose concrete exhibition and evaluation are assigned to Paper 43's audit program (no row failed; the which-entropy circularity imported as a live defect) — and a typed covariant debt (relativistic microcausality, dressed sources, species), named and not discharged. RC3 is spent exactly once, on the single calibration benchmark declared before any dephasing datum was consumed: the measured measurement-induced dephasing of a dispersively read superconducting qubit (Schuster et al. 2005; channel model Gambetta et al. 2006). The audit is non-vacuous by construction: certified exports are the output-port photons insofar as their field modes are amplified into the retained quadrature record at the declared grain — the certified intensity is nu_eff = eta_cert kappa nbar with eta_cert = eta_ext eta_rec, the output coupling times the record-chain efficiency, both declared as intervals — and the per-event pointer-overlap deficit is computed from the device's spectroscopic parameters (2 chi / kappa = 1.0, so the which-state information is fully exported per photon). The declared linear fit proves mathematically inapplicable at this device's measurement strength, so the extraction is executed as a documented protocol-deviation conservative bound — a declared-tolerance ceiling read against the source's own validated line-shape model, from text-stated quantities only, with a full sensitivity table — and the source's data show no excess dephasing beyond that model: the calibration returns not a point but the pre-declared gate-truncated band Lambda_cal = [5e-5, 2.1], computed at the conservative corner of the declared inputs (including the record-chain efficiency) so the dominant systematics sit inside the band; the two bookkeepings of standard budget and kernel excess are ledger-distinct while the physical carriers may overlap. The band is nonempty, so the normalization survives: lambda_dec is bounded above, for the first time, by a declared-tolerance reading of laboratory line-shape data (a factor of five below Paper 41's illustrative interferometric ceiling at the most conservative record-chain reading, tightening to a factor of fifty under the architecture-level reading), lambda_dec = 0 is never carried (the trade-off floor truncates from below), and the normalized kernel — the lambda_dec band, the lambda_diff prediction band, the computed order coefficient, the unchanged thresholds — is handed to Paper 43. The retired authority is the Global Reducer: the demand that an adopted dynamics earn its standing only by a globally certified exact reduction of the established theories. Its methodological face falls to the construction — exact-but-sector-relative unitarity, on-average Newtonian sourcing over declared classes, imported-route compatibility at declared order, and declared-calibration normalization are how correspondence is legitimately held; its substantive face (one true global reduction) is declined, not refuted. Every result is conditional on P_ker; the calibration datum is spent, never scored; no gate is crossed and no U3 is claimed.