# 30, The Living Exclusion Table: what the first tabletop test of gravitationally induced entanglement kills, leaves standing, and can never decide (ADJUDICATOR program)

**Program decision document, v1.0 (2026-09-25).** Program of record: `29_dream_question_4.md` (sha256 `81fb1485eb6c1fab…`; doc 29 §4 the recommendation incl. the corrected-rider paragraph, doc 29 §4.3 (2); doc 29 §5 the mandates M-A1..M-A8, PILOT-ADJ with P-A-ctrl and P-A1..P-A7 and their declared parity, KA-ctrl and KA-0..KA-6, the engines, the prime bands and the scout lanes L1–L4, read in full and applied). **Grade: a bounded DUTY, never a discovery claim.** Files of record (sha256 prefixes as hashed by this document's leg at 2026-09-25 00:45:13 CEST): pilot M-A1 [PIL] `analysis/adj_pilot.py` (`27721ebd…`; registrations R1 `5985191a…` 18:21:42, R2 `b26d2298…` 18:33:29, R3 `9fa20c2c…` 18:39:51; tokens 9,000,011 / 9,000,041) → `adj_pilot_results.md` (`d19efbd0…`), `adj_pilot_rows.jsonl` (`31ca4dd2…`, 76 records), `adj_pilot_state.json` (`10b6ed66…`), `adj_pilot_run.log` (`fd3868b9…`); verifier gate M-A2 [GATE] `analysis/adj_pilot_verifier.md` (`911b1b03…`; registration `69371897…` 20:46:34, addendum `e7d1acd0…` 21:06:56; tokens 9,100,009 / 9,100,033), **PASS**; scout lanes L1–L4 [LANES] `papers/26_adjudicator/SCOUT_ADJ_LANES.md` (`6a8fd081…`; R0 `43ab6cf5…` 21:28:33, R1 `9fbc096e…` 21:34:45; no prime), no kills, KA-3 not fired; bounded reproduction M-A4 [REPRO] `analysis/adj_repro_r04.md` (`1784ba8b…`; registration `077985df…` 21:43:15; script `48a892c3…`; no prime), R04 reproduced of record; the in-house note M-A3 [NOTE] `30a_exclusion_table.md` **Final v1.0** (`3217a3a4…`, not HELD; its v0 `3edeab69…` preserved byte-identical as `analysis/adj_note_finalize_v0_snapshot.md`); the two adversarial legs M-A6, rigor [RIG] `analysis/adj_note_rigor.md` (`0657195a…`; tokens 9,100,067 / 9,100,079), **PASS-WITH-REPAIRS**, and experimentalist steelman [STL] `analysis/adj_note_steelman.md` (`3b5645e1…`; tokens 9,200,017 / 9,200,021), **PASS-WITH-REPAIRS**; the finalization ledger [FIN] `analysis/adj_note_finalize.md` (`e7ffe76e…`); the integrity co-runner M-A8 `analysis/scope_cert.md` (`f01e32f7…`; tokens 9,000,049 / 9,000,059) and its verifier `analysis/scope_cert_verifier.md` (`6d6e1aa7…`; tokens 9,100,051 / 9,100,061), **CONFIRMED-WITH-DEFECTS**, reported here and never ruled. Companion documents: [W] `06_witness_theorem.md` (Paper B's Theorem W, W1–W6, Definitions 1–3, Corollaries W1/W2, Theorem W′), [E1] `05a_E1_visibility_note.md`, [E1M] `21a_E1M_magic_note.md`, [WM] `21_witness_magic.md` (Paper M's Theorem M), docs 01–03 (the Aziz–Howl reproduction and bound), `analysis/van_hove_adjudication.md` (the template, 18/18), `analysis/lemma_M_mediation.md` (Lemma M), doc 27 §3.9 (the Fork-1 corollary). This document's own leg: registrations `analysis/adj_doc30_registration_R0.txt` (sha256 `50ce97ac…`, frozen 00:45:59, before the first network query) and `…_R1.txt` (`6256d94d…`, frozen 00:53:29, before any line of this document and before any of the leg's computations); its records are `analysis/adj_doc30_*` (§8).

**The question (doc 29 §4, dream form, verbatim).**
<!-- V: 29_dream_question_4.md -->
> **When the first tabletop experiment observes, or fails to observe, gravitationally induced entanglement, which published account of gravity does it kill, which does it leave standing, and what will no experiment of its class ever decide?**
<!-- /V -->

**Operational form (doc 29 §4 in substance).** Over the registered model set M26, 23 rows, every published classical, hybrid, stochastic, device-independent or platform account that claims to entangle, not to entangle, or to be indistinguishable on BMV/GIE-adjacent protocols through 2026-09-24, frozen in the pilot docstring before any row was read, and with the licensed instruments only (Theorem W with W1–W6 and Definitions 1–2 as amended by the van Hove adjudication; Theorem W′ and E1's null-space clause; Theorem M with M0–M4, Lemma M's fingerprint table; Lemmas C and T2; doc 03's bound; the doc 04 / 05 certificates; the E1 / E1-M thresholds; the Fork-1 magnitudes; doc 19's S / K carriers), type every row in three columns, (P) what a certified positive does to it, (N) what a null does to it, (S) its contribution to the saturation map, and compute the two cells no committed certificate decides: N-1 (the minimal-noise class of arXiv:2603.26075 at the committed geometry) and M-1 (the white-noise visibility F\* at which the noisy gravitational gate enters the convex hull of the two-qubit Clifford unitary channels).

**Status of the record.** Doc 29 judging (2026-09-24; no trigger fired) → PILOT-ADJ (M-A1; KA-ctrl fired at the first control run and was repaired, PASS-AFTER-DATED-REPAIR; census 23 rows; N-1 INSTANCE; M-1 branch 1; the pilot's headline "KA-0 fires" later refuted) → verifier gate (M-A2; **PASS**; R04 re-typed NEEDS-REPRODUCTION; KA-0 not fired) → scout lanes L1–L4 (no kills; KA-3 not fired, OCCUPIED-BELOW-THRESHOLD; 12 accounts flagged without verdict) and the bounded reproduction (M-A4; R04; KA-1 fired by its letter, verdict-neutral) → assembly of the note (M-A3; 30a v0) → the two adversarial legs (M-A6; PASS-WITH-REPAIRS each; no verdict cell moved) → finalization (30a Final v1.0, not HELD) → **this document**. The co-runner SCOPE-CERT (M-A8) ran in parallel (17:05–20:21) and hands its riders to the owner's post-2026-10-30 block. **Gates (§7.4): G30-1 closing audit at the untouched reserve ≥ 9,300,001, owed only if a quotation event is ever wanted; G30-2 fresh re-sweep incl. the post-2026-10-30 Perimeter leg, OWED (calendar); G30-3 external quotation, BLOCKED behind both; G30-4 pointer rows, RULED here: NONE ENTERED, Papers B and M untouched.** Licensing is never posting; defer-until-big governs all external timing.

---

## 1. Verdict up front

### 1.1 Trigger ruling FIRST (KA-6) and the scoop ruling (KA-3; T4)

**The four Adjudicator triggers, re-dated at the start of this leg (2026-09-25 00:46:57–00:48:44 CEST, before the 02:00 CEST arXiv announcement) and at its end (01:11:47–01:13:35).** Own script `analysis/adj_doc30_triggers.py`; net strings a verbatim copy of the finalization leg's, asserted identical by AST extraction before the first query (8/8 assignments, key order equal); windows arXiv `submittedDate:[202609150000 TO 202609252359]`, INSPIRE `de ≥ 2026-09-15`, citers `de ≥ 2026-09-01`; states `analysis/adj_doc30_trig_{start,end}.json`.

| trigger (firing rule of record) | start (00:46:57–00:48:44) | end |
|---|---|---|
| **T1** MOVW ref. [7], fires iff 2511.07348 reaches v ≥ 2 or an MOVW author posts the promised full paper | **NOT FIRED.** 2511.07348v1, published = updated = 2025-11-10T17:51:36Z; arXiv au:Marletto_C / Oppenheim_J / Vedral_V / Wilson_E 0 in the window; INSPIRE Vedral 1 (2609.22976, weak values, not gravity; typed since doc 29 §0), the others 0; citers of recid 3081450 (cc 20) since 09-01: 2609.09931, 2609.04104 (rows R23, R22; neither by an MOVW author); the two title nets on [7]'s title return exactly {2511.07348, 2510.19969}, 2510.19969 still v1 | **NOT FIRED**, identical to the start (2511.07348v1 and 2510.19969v1, unchanged; the same feeds, citers and title nets) |
| **T2** a BMV/GIE collaboration's first data or dated data timeline | **NOT FIRED.** arXiv GIE abstract net 1: 2609.22501 (quadratic-curvature GIE theory, row R21); INSPIRE GIE titles 2: 2609.26818 (Joseph, a squeezing proposal, a flagged account, §2) and 2609.22501; group feeds Aspelmeyer 2609.17787 (a levitated-micromirror platform), Mazumdar 2609.24998 (not GIE) and 2609.22501, all others 0; no data, no timeline | **NOT FIRED**, identical to the start |
| **T3** the Andriolo–Cius–Comar–Amaral pipeline paper (`a Amaral, B`; D. W. P. Amaral is never counted) | **NOT FIRED.** every net 0, window and all-time conjunctions | **NOT FIRED**, identical to the start |
| **T4** a direct scoop of Theorem W or Theorem M | **NOT FIRED.** every W / M title and abstract net 0 against the controls (INSPIRE `abstracts.value:"black hole"` 142; arXiv `cat:quant-ph` 609 and `all:"black hole"` 184, newest 2609.28469 of 09-23); recent citers of 2004.01189 (→ 2609.10535), 2503.18734 (→ 2609.27390), 1707.06050 (9) and 1707.06036 (8), every id already typed by the record | **NOT FIRED**, identical to the start; controls unmoved (609 / 184 / 142) |
| KA-3 ledger-shape nets (the gate's addition) | 0 on both indices | 0 |

**Like-for-like:** 0 differences at the start against the finalization leg's end state (2026-09-25 00:31:43–00:33:33), the rigor leg's end (23:41:09–23:42:57), the steelman's end (23:30:16–23:34:03) and the assembly's end (22:43:55–22:46:11); R0's ten forecasts F0.1–F0.10 all confirmed. At the end (01:11:47–01:13:35, still before the 02:00 CEST announcement): 0 differences from this leg's start and from all four earlier end states; the controls did not move (609 / 184 / 142).

**Ruling: no trigger has fired at any leg of the program, the pilot (three checks), the gate, the lanes, the reproduction, the assembly, both M-A6 legs, the finalization and this leg (§1.3, M-A7). KA-6 NOT FIRED; there is no row R24.** MOVW's ref. [7], as the lanes read it from 2511.07348's own reference list:
<!-- V: papers/26_adjudicator/SCOUT_ADJ_LANES.md -->
[7] Marletto, C. Oppenheim, J. Vedral, V. & Wilson, E. Classical gravity cannot mediate entanglement. Manuscript in preparation, 2025.
<!-- /V -->
It announces a mediation / responsibility criterion and carries the comment's own title, so the version check of 2511.07348 stays load-bearing: a v2 that replaces the comment is a possible form of posting. The day it posts, its row enters first and its criterion is typed against Lemma M before anything else (KA-6; G30-5).

**Scoop ruling.**
- **KA-3 (framing scoop), NOT FIRED, by the lanes' registered letter K1–K4** (K1 posted; K2 ≥ 5 GIE model classes; K3 a uniform ledger of ≥ 2 named premises marked per account; K4 ≥ 3 of its accounts are M26 accounts first posted on or after 2025-07-01). The one near-threshold object is **Di Biagio 2511.02683's Table II** (present in v4, retained in v5 of 2026-09-16): seven theories (six distinct classes) marked against {statespace, mediation, classicality}, K1–K3 hold, **K4 fails** (only Aziz–Howl is an M26 account of that date). **Occupancy: OCCUPIED-BELOW-THRESHOLD, cite and distinguish, no race.** The distinction, as the note states it (30a §10): Di Biagio's classicality is the GPT simplex while Definition 2 has three senses (K, D, O); his ledger has no null column and no saturation map and does not type the 2026 rows; his conclusion coincides in form with Theorem W's disjunction. Every later leg's ledger-shape nets returned 0, this leg's included.
- **T4 (a direct scoop of Theorem W or Theorem M), NOT FIRED** (the table above).
- **Credited, never claimed.** The table's verdicts belong to the rows' own texts and to the house's licensed theorems; the ledger adjudicates assumptions, never authors or priority. Di Biagio's Table II is the prior assumption ledger (the record's "adversarial mini-ledger", doc 06 v1.1); Corollary W2 is the house's own earlier scoreboard; N-1 is Fabiano–Fujita–Matsumura–Carney's theorem (itself an extension of the Kafri–Taylor inequality, the committed H1b clause) evaluated at the committed geometry; the device-independent price is the Horodecki criterion evaluated on the committed state; M-1 is a convex-hull certificate on the committed gate, no occupancy sweep was registered for it and no priority sentence is made for any cell.

### 1.2 The typed outcome and the answer at record grade

**The registered outcomes (doc 29 §4), verbatim:**
<!-- V: 29_dream_question_4.md -->
> **Typed outcomes of equal standing:** SCOREBOARD-BY-DEFINITION (every row readable off its own text, N-1 decided by committed text, the table assembles as a note: the order of record's own "a note, not a front"); SCOREBOARD-WITH-ROWS (one or two rows need a bounded reproduction on the van Hove template); ROW-FLIP (a reproduction contradicts a paper's own stated computation, reported first); COUNTEREXAMPLE (a model inside W1–W5 with a Definition-2-classical mediator delivering a Definition-1 positive, Theorem W in question; stop everything; the inversion is published first); SATURATED (the map confirms that no BMV-class discriminator splits Theorem W's disjunction above resolution, a typed fact, never a slogan).
<!-- /V -->

**Outcome type of record (each registered outcome by its own letter; their conjunction; ROW-FLIP printed first by its own "reported first"; no criterion outside the letters adds, suppresses or ranks one, R1 A4): ROW-FLIP (by its letter; verdict-neutral) + SCOREBOARD-WITH-ROWS (one row, R04, reproduced of record) + SATURATED.** Not SCOREBOARD-BY-DEFINITION; not COUNTEREXAMPLE. No single-word headline is selected beyond this conjunction: doc 29 registers the five outcomes as of equal standing and registers no ranking, so a ranking would be a definition introduced after the readout.

| outcome | its letter | holds? | the record |
|---|---|---|---|
| **ROW-FLIP** | a reproduction contradicts a paper's own stated computation, reported first | **YES, by its letter** | The M-A4 reproduction of R04 (Struyve, arXiv:2510.20991 v2) found the printed eq. (31) ×2 on all four Newton–Schrödinger mean-field phase terms relative to the paper's own eq. (29) on its own normalized state (21); the printed eq. (46) and the W_NS curve of Fig. 1 inherit it (reproduced W_NS 0.07775, printed 0.28683 at the committed geometry). **What the letter does not mean here:** no verdict cell of R04 or of any other row moved; no premise of Theorem W or Theorem M is touched (KA-1's re-audit clause did not apply); the NSB chain the gate assigned, eqs. (35) and (47), reproduced exactly. The contradiction is in the paper's printed arithmetic for its NS member; the paper's qualitative claim of distinct witness signatures stands. |
| **SCOREBOARD-WITH-ROWS** | one or two rows need a bounded reproduction on the van Hove template | **YES** | One row, R04, NEEDS-REPRODUCTION by the gate's ruling (its NSB (N-b) verdict rested on a computation neither its text nor a committed certificate settled); reproduced of record: NSB (N-b) REFUTES by E1's clause, thin, ×1.2474. KA-4 not fired (1 ≤ 2). |
| **SATURATED** | no BMV-class discriminator splits Theorem W's disjunction above resolution | **YES** | Six entries SAT-1..SAT-6 with sizes (§3.5); SAT-1's size now carries its parametrization, which does not affect the status. |
| SCOREBOARD-BY-DEFINITION | every row readable off its own text, N-1 decided by committed text | **NO** | R04 needed a bounded reproduction (the gate's D1). KA-0 did not fire; the phrase "a note, not a front" is **not entered** as this program's outcome (the pilot's contrary headline is refuted, §1.6). N-1 was decided by committed text, that conjunct held. |
| COUNTEREXAMPLE | a model inside W1–W5 with a Definition-2-classical mediator delivering a Definition-1 positive | **NO** | P-A4 TRUE at the pilot, the gate, the reproduction and both M-A6 legs; Theorem W stands on every row. The steelman's position-readout completion of R04's NSB member (two-setting witness 1 + C − C² = 1.1319 from product wave functions) violates W6 (ii), a premise of the theorem, lies outside the row's computation and carries no verdict; KA-2's letter says "inside W1–W5" while the theorem's premise set is W1–W6, flagged for the owner (§7.5, candidate (D)). |

**Beside the letters, carrying no type:** the reproduction leg read the outcome as "SCOREBOARD-WITH-ROWS plus a verdict-neutral KA-1 rider", because no cell's value class moved; that criterion is not in doc 29 §4's letter, and under the frozen-rule clause it cannot suppress an outcome whose letter holds (the note's own ruling, 30a §13).
**The kills that fired, neither stopped the program:** KA-ctrl (the pilot's first control run, 18:39:08; the quote-comparison layer; dated repair R3; re-run 4/4, PASS-AFTER-DATED-REPAIR) and KA-1 (the reproduction, by its letter; verdict-neutral). KA-0, KA-2, KA-3, KA-4 and KA-6 did not fire; KA-5 held at every leg (§1.5).

**The answer, at record grade (a transcription of 30a Final v1.0 §1; every condition is part of the answer).**

*If the first tabletop run certifies entanglement*, a sound spin–spin witness above the certified budget with the electromagnetic channel priced and per-run statistics (Definition 1's three clauses), under W1–W6, it licenses exactly one sentence, Theorem W's:
<!-- V: 06_witness_theorem.md -->
> A certified positive (all three Definition-1 clauses) licenses: "no classical degree of freedom carried this entanglement, either a non-classical mediator did, or the interaction is a nonseparable direct quantum gate with no carrier." Experimentalists should quote that sentence.
<!-- /V -->
Against the 23 registered accounts it **refutes 9 on their own formulations**, the Aziz–Howl c-number class, as R03 with its seven comments and R06 (a comment adjudicating the same class) type it; Struyve's additively separable semi-classical potentials (R04; its Bohmian NSB member only for a spin readout whose pointer is not the diamonds' own position); the Koopmanian commuting mediator (R09), every commutative C\*-algebra mediator (R11), Oppenheim's classical-quantum gravity in its locally correlated realisations (R12), the non-entangling members of the minimal-noise class of arXiv:2603.26075 (R13), the effective Markovian classical-quantum description of Tomizuka–Takeda (R15) and a classical gravitational-wave background as a driver (R23, scope rider), **reclassifies 3 without refuting them**, because what they call classical is not classical in Definition 2's sense (the van Hove hybrid R10, Di Biagio's Diósi–Penrose and Aziz–Howl readings R05, Płodzień et al.'s pair term R07; each certified to carry exactly its non-classical ingredient), and is **compatible with the other 11** (R01, R02, R08, R14, R16–R22).

*If it fails to see entanglement at sufficient sensitivity and the Z-conditioned fringes show no second difference (the extracted angle Δ̂ consistent with 0)*, the null certifies unconditionally that gravity imparted no second-difference phase in the probed regime; read under linear phase additivity it refutes only the conjunction {quantized gravity ∧ linearized coupling ∧ QM for matter at these scales}; it contradicts, wherever resolvable, every row that predicts such a phase (R05's listed theories, R07's pair term, R08's derived positive by its own contrapositive, R10, R13's non-local member, R18's pair term, R19, R20, R21); it leaves every non-entangling account standing, because a null is their own prediction. A witness null with Δ̂ ≠ 0 is not this branch: it is the classical-channel fingerprint, a diagnostic that refutes nothing. A null never feeds a classicality claim, modified-coupling quantum models predict a null too.

*If that null also comes with full visibility*, E1's pair/single ratio R flat at 1 on the fixed-phase contours for both masses **and** a clean single-mass scan for both masses, with reaches r_R on the ratio and u_s on the single-mass contrast such that −ln(1 − r_R) − ln(1 − u_s) ≤ 211/1944 (at r_R = REACH, u_s ≤ 0.0939), it also refutes Oppenheim's CQ gravity (R12, through the single-mass scan, because its forced decoherence is partner-independent), the **whole non-entangling class of arXiv:2603.26075 at once (cell N-1: ×5.358 when both reaches equal REACH, on the hold-only phase)**, Struyve's NSB member **thinly (×1.2474; 0.88σ per contour point, about 2.5σ at the scan level) and only under that paper's Born-rule statistics with a non-positional spin readout**, and R15's Markovian limit. A flat ratio alone refutes none of them: a member whose forced noise is partner-independent leaves R at exactly 1 on both masses (its V_s is 0.897144 on each).

*What no experiment of its class can decide* is the saturation map (§3.5): whether the entangler was a mediator or a direct gate (SAT-1: ≤ 3.3 × 10⁻¹³ rad in doc 27's parametrization, 8.38 × 10⁻¹⁴ rad at the committed point, a transient that is exactly 0 from 2.33 ps into the hold), a retardation test (SAT-2: 0.667 ps at 200 µm), quantization beyond non-commutativity (SAT-3: ⟨n_g⟩ ≈ 1.3 × 10⁻⁵⁷ radiated gravitons for C²-smooth ramps), van Hove versus quantized linear gravity (SAT-4: 0 in every mass-side statistic), the Clifford-exception coupling (SAT-5: τ_π = 25.0136 s, hold-only, a family theorem and never a lab point) and local-hidden-variable matter statistics (SAT-6: split only by a Bell test, unfiltered two-setting CHSH above 0.98799697, above 0.956106 with local filtering, the minimum over all Bell tests not committed, against the witness's 0.864790). The honest form of any "closes the loophole" sentence is this map.

*One cell carries a physics reading (M-1, §3.4):* under the campaign's white-noise benchmark the noisy gravitational gate is outside the convex hull of Clifford unitary channels at every visibility above **F\* = 1/(1 + 4|sin Δ|) = 0.447343871026845**, far below the witness's own threshold F_ent = 0.864790, at every visibility where the witness can fire, the modelled gate is certifiably not a Clifford mixture: the witness, not the gate's magic, is the bottleneck. A gate-typing number and a Paper-M row, never a law.

**Ceiling riders (the order of record's, fused from day one; verbatim, doc 29 §4):**
<!-- V: 29_dream_question_4.md -->
**Ceiling riders, fused from day one.** DUTY grade: no sentence of this program claims a discovery; every verdict row is a corollary of the house's licensed theorems and the model's own text, labelled so before the run. A positive certifies "non-classical" in Definition 2's sense and the Theorem-W disjunction (a non-classical mediator, or a nonseparable direct gate with no carrier), never "quantum", never "gravity is quantum". A null never feeds "gravity is classical" (doc 06 §5; 21a §7). "Loophole-free" is never used bare, the saturation map is its honest form. The ledger adjudicates assumptions, never authors or priority; "reclassified, not refuted" wherever a model's "classical" lies outside Definition 2. Published exclusion bounds from sensor data are cited as row data, never re-derived (§4.2 (v)). "Gravity is a magic factory" never, M-1 is a gate-typing number. "Undecidable" never (N1). Licensing is never posting; defer-until-big governs every export; whether D1 is "big" is the owner's call and is never argued in this program's text.
<!-- /V -->

### 1.3 Per-mandate results (each transcribed from its own leg's record; grades: COROLLARY = decided by definitions and the row's own text, COMPUTED, CERTIFIED, TRANSCRIBED)

- **M-A1, PILOT-ADJ (the entry calculation) [PIL].** Two instances on one script and state: the first ran 2026-09-24 18:12:54–18:41:35 and died on a network error after its texts stage; the resume re-verified every registration hash, every text hash and the order of record before any new stage (19:29:56) and ran to 20:27:23, re-freezing and re-registering nothing. Order of record: stage T triggers 18:13:03–18:15:42 (none fired; one new non-firing hit, 2609.26818) → R1 frozen 18:21:42 (M26 as 23 rows with 9 sub-entries, the four typing categories, the N-1 / M-1 objects, the predictions and kill criteria transcribed; 29 verbatim blocks checked) → R2 18:33:29 (the instruments, the value-class procedure C1–C5, the N and S column rules, the control's expected verdicts, the N-1 translation rule, the M-1 convention and its near-corollary NC-M1, the battery sources, the pilot's own forecasts; 46 verbatim blocks) → control run 1, 18:39:08: **KA-ctrl FIRED** (type and P class matched 4/4; criterion (iii) failed at R01 and R03 on the quote-comparison layer, a line-end hyphen and typographic quotes) → R3 18:39:51 (the dated repair of the comparison layer only) → control run 2, 18:39:54: 4/4 MATCH, **PASS-AFTER-DATED-REPAIR** → texts 18:40:02–18:41:35 → [resume] triggers 19:30:01 (none) → battery 19:35:24, **15/15** committed anchors regressed before any live cell (Δ = −211/672; F_ent 0.864790; F_magic 0.975555 and 0.9755547; N_min = 1105 with the N = 1104 deficit 17/276971520; M₂ = 0.104445; F_STAB = 0.981666; the Fork-1 phase; v_DI 0.98799697; the d/c timing; τ_π; ⟨n_g⟩ by transcription) → census 20:03:52: **23/23 DAF-BY-DEFINITION**, 0 NEEDS-REPRODUCTION, 0 NOT-INSTANTIABLE, 0 OUT-OF-SCOPE; 90 deciding sentences verified verbatim (70 on the 19 non-control rows, 20 on the control rows); 9 sub-entries typed (2608.17688 OUT-OF-SCOPE, no verdict); P classes 9 / 3 / 11 → N-1 20:04:49: **INSTANCE**, decided by committed text at the first act (the row's floor is "noise rate ≥ entangling rate", E1's committed H1b clause in the row's generality) and confirmed by the registered evaluation (worst-case deficit 0.0528232369 against REACH 0.0098588524, ×5.358) → M-1 20:18:32: **F\* = 1/(1 + 4|sin Δ|) = 0.447343871026845, certified width < 10⁻⁴⁰; branch 1** (own ⟨H, S, CNOT⟩ closure of 11520 Pauli-transfer matrices equal as a set to the committed engine's census, run as an isolated copy `50a53ea9…` verified before and after; a separating functional with entries in {0, ±1/3, −1}; an explicit decomposition over 220 Clifford unitaries with weights affine in (cos Δ, sin Δ)) → end-of-leg triggers 20:22:23–20:24:54 (none). Pins: 154 files before / resume / after; README.md changed externally at 20:24:03 (not written by the leg). The pilot's headline, "KA-0 fires: a note, not a front", was **refuted by the gate** (§1.6).
- **M-A2, the verifier gate [GATE]: PASS (not PASS-KA0-NOTE).** 20:37:50–21:22; triggers none at the start and end (52 nets identical); registration frozen 20:46:34, before any row text was opened by the gate and before any own computation. Own routes throughout: the control 4/4 under its own matcher and re-typing; the P class agrees on 23/23 rows, **the type on 22/23, R04 re-typed NEEDS-REPRODUCTION** (D1, the only verdict-moving defect: the NSB member's (N-b) verdict rested on census arithmetic that neither the row's text nor any committed certificate states, and its "near-corollary" label was introduced after the readout); hence **KA-0 does not fire by its letter** and the outcome is SCOREBOARD-WITH-ROWS with one row. N-1 by a disjoint route (its own symbolic derivation of the row's eq. (34) from two dephased qubits under H = J Z⊗Z; INSTANCE; the same 0.0528232369 and ×5.358; the worst-case exponent Δφ/4, first order in the phase). M-1 by a disjoint route (an own Sp(4, F₂) census, 720 × 16 = 11520 channels; an own facet with entries in {−1, 0, 1}, 29 nonzero, 6858 Cliffords tight; an own 185-element vertex decomposition; F\* certified with width 3.3 × 10⁻¹¹⁷; branch 1), plus data: over all 36720 four-qubit stabilizer states a float LP locates F\*_CSP = F\*_Cliff, LP-located, not certified. Defects D1–D7 graded (D2 text currency, four stale on-disk texts, one with a changed diagnosis; D3 provenance, six abstracts fetched before R2's freeze; D4 R13's premise; D5 R12's quote; D6 scope riders on R05, R15, R23; D7 its own log overstatement, corrected). Pins 209/209 byte-identical.
- **L1–L4, the scout lanes [LANES]** (21:28:14–22:11; R0 before the first query, R1 before any lane readout; no prime). **No KILLS anywhere** (L1 0 of 32 entries; L2/L3 0 of 104 graded hit records). L4: triggers none at the start and end, 0 differences from the gate's end. L1: identity 34/34; currency, exactly the gate's four stale texts (2512.13675 v2 → v4; 2604.19696 v2 → v3, its diagnosis changed to diffusion between the objects, i.e. TRANSFER; 2511.02683 v4 → v5, retitled; 2508.03052 v3 → v4, a typo fix), replacement sentences supplied; 85/90 pilot quotes present in the latest versions; the zero-computed-content test PASS on every P-bearing entry and consistent with the gate's typing on all 23 rows; the R04 test found no second instance; MOVW's ref. [7] identity read from 2511.07348's PDF. L2: **KA-3 NOT FIRED, OCCUPIED-BELOW-THRESHOLD** (Di Biagio's Table II fails K4). L3: 0 new citers of the six watch ids since Scout B's baseline; 32 distinct citers, dated and graded. Grades over 104 hit records: KILLS 0, FORCES-RIDER 10 records (9 accounts, all flagged rows), CITE-AND-DISTINGUISH 19, NO-OVERLAP 58, ROW 17; **12 GIE-adjacent accounts outside M26 flagged without verdict** (§2). A disclosed deviation from the registered reading method: 13 entries read in full, 19 at claim-sentence depth. No forecast refuted (6 + 21). Pins 282/282.
- **M-A3, assembly of the in-house note [NOTE].** The assembly leg (22:19:24–22:49; R0 22:19:57, R1 22:24:37, frozen before any line was written; triggers none; checks clean at its third run, CK-A 28/28, CK-B 13/13 verbatim blocks, CK-C 72/72 row quotations re-found on their pages, CK-D 0 banned strings, CK-E 23/23, CK-F 5/5; pins 376/376) wrote v0 (`3edeab69…`); after the two M-A6 legs the finalization (2026-09-24 23:59:41 → 2026-09-25 00:37; R0 00:00:36, R1 00:12:00; certifications 15/15; one asserted patcher with 107 edits; checks all clean on the final text, CK-A 41/41, CK-B 14/14, CK-C 76/76, CK-D 0, CK-E 23/23, CK-F 8/8, CK-G 26/26, CK-H 6/6; pins 465/466, the only change the note itself) produced **30a Final v1.0 (not HELD)**: every repair of both legs folded or, where it touches a committed document, drafted as a record-side candidate; no verdict cell, typed outcome, kill ruling, licensed sentence or verbatim quotation moved. The note is in-house and is never posted by this program.
- **M-A4, bounded reproduction, one row (R04) [REPRO].** 21:27:07–21:58; triggers none (56 nets, start and end identical and equal to the gate's end); registration 21:43:15 with every printed equation transcribed into a frozen EQ block; development runs only after the freeze. **Eqs. (35) and (47) REPRODUCED exactly** ((47) by two routes, the row's own average and Tr(ρ_ens W)); (28), (43), (44), (45) reproduced; **(31) and (46) CONTRADICTED**, the ×2 of the Newton–Schrödinger phase, verdict-neutral, no W/M premise touched (KA-1 by its letter; doc 29's P-A3 FALSE by its letter). On the reproduced NSB ensemble: ρ_NSB = ρ₁ ⊗ ρ₂, exactly the product of the single-mass marginals of the Newtonian quantum state; V_c = |cos(Δ/2)| for both masses and both partner outcomes; V_s = 1; Δ̂ = 0 exactly (no conditional imprint, the census's identification with 21a §7 item 1 is replaced). **The NSB (N-b) cell REFUTES by E1's clause, THIN:** forced deficit 1 − cos(Δ/2) = 0.0122982745690 against REACH = 44521/4515840, ×1.2474; the closed-form certificate 1 − cos(Δ/2) ≥ Δ²/8 − Δ⁴/384 ≥ Δ²/10 whenever Δ² ≤ 48/5 (at the committed Δ the difference is 955128023467/391541969387520), the ratio tending to exactly 5/4 as Δ → 0, thin at every geometry by construction. P REFUTED-BY-POSITIVE (Definition 2 (D)); N-a LEAVES-STANDING; S NONE. Forecasts F1–F16 all confirmed. Pins 232/232. Sequencing, disclosed: the leg was assigned after the gate and ran beside the lanes, before the assembly; it wrote no assembly text.
- **M-A5, optional computed rows: NOT EXECUTED.** No leg opened any of them (TM-26 (a) and (c); DI-LOC's readout-stage signalling budget); the note cites Scout B's SGI-window probe only as scout-grade data in R20's (S) cell, never as a verdict. Nothing of the table depends on them; they stay optional, never headline.
- **M-A6, the two adversarial legs, each PASS-WITH-REPAIRS; no verdict cell moved.** *Rigor* [RIG] (22:49–23:48; R0, R1, and R1-A a correction of R1 frozen before any M-1 computation; tokens 9,100,067 / 9,100,079): every number re-derived by own routes with exact enclosures (31/31 plus the R04 reproduction 7/7); **M-1 certified a third time by a disjoint route** (a four-class normal form of the census, 576 + 5184 + 5184 + 576 = 11520; a 128-element symmetry group that includes the transpose anti-automorphism; a facet with 31 nonzero entries in {±1/2, ±1}, 3634 Cliffords tight; a decomposition over 50 Clifford unitaries, minimum weight 0.0244453; width 5.7 × 10⁻⁷¹); 13 rows re-typed from full text by a pre-registered selection rule, all 13 P classes equal; 72/72 quotations by a second extraction route; defects D1 (SAT-1's scope label), D2 (the one-line summary), D3 (R08's linearity premise) and notes N1–N13; pins 396/396. *Experimentalist steelman* [STL] (23:03–23:37; tokens 9,200,017 / 9,200,021): every verdict quantity regressed exactly (14/14); **two defects blocking export**, D1 (a null with full visibility needs a clean single-mass scan as well as a flat pair/single ratio: a member of N-1's own class with partner-independent forced noise gives R ≡ 1 on both masses with V_s = 0.897144) and D5 (Theorem W′'s null needs Δ̂ consistent with 0: a one-way classical channel gives a witness null with Δ̂ = Δ); D2 (R04's NSB cells need a non-positional spin readout; the position-readout completion reads 1.1319 while passing single-mass fringe statistics exactly, W6 (ii)'s lab certificate is necessary, not sufficient), D3 (SAT-1 at the committed point 8.38 × 10⁻¹⁴ rad, exactly 0 from 2.33 ps), D4 (the device-independent price is inequality-specific: 0.956106 with local filtering), D6 (V_s as E1 defines it), D7 (every margin is on the hold-only phase), D8–D14; M-1's branch-1 reading robust under the optimal entanglement test and the campaign's branch-diagonal noise; pins 399/399. The finalization folded every item (the HELD test found no load-bearing item unrepairable from the record).
- **M-A7, standing sweeps and triggers.** The four triggers were re-dated at the start and the end of every leg; none fired anywhere:

  | leg | start | end | differences |
  |---|---|---|---|
  | doc 29 judging (Scout B; the judge's spot check of T1) | 2026-09-24 15:18–16:04 | J-D 16:31:54 |, (index-validated zeros) |
  | pilot (M-A1) | 18:13:03–18:15:42 (stage T); 19:30:01 (resume) | 20:22:23–20:24:54 | 47 nets identical to stage T at the resume and at the end |
  | gate (M-A2) | 20:39:20–20:41:22 | 21:15:57–21:18:00 | 52 nets identical |
  | reproduction (M-A4) | 21:28:27–21:30:47 | 21:50:50–21:53:10 | 56 nets identical, = the gate's end |
  | lanes (L4) | 21:29:19–21:31:14 | 21:59:06–22:00:56 | 0 vs the gate's end; + the two T1 title nets |
  | assembly (M-A3) | 22:20:46–22:23:05 | 22:43:55–22:46:11 | 0 vs the lanes' end |
  | steelman (M-A6) | 23:05:10–23:09:04 | 23:30:16–23:34:03 | 0 vs the assembly's end |
  | rigor (M-A6) | 23:08:31–23:10:29 | 23:41:09–23:42:57 | 0 vs the assembly's end |
  | finalization | 2026-09-25 00:01:45–00:03:39 | 00:31:43–00:33:33 | 0 vs the rigor, steelman and assembly ends |
  | **this document** | **00:46:57–00:48:44** | **01:11:47–01:13:35** | **0 vs the finalization, rigor, steelman and assembly ends at the start; 0 vs this leg's start and the four earlier ends at the end** |

  M-A7's named sweeps ran inside the lanes (the device-independent turn's and the noise-floor program's citers, dated; the classical-carrier genre; the ledger-shape occupancy of the Carney, Dakić, Bassi and Kent groups, Foo et al. and the Martín-Martínez–Perche line) and are re-owed at G30-2. The Perimeter post-2026-10-30 leg is doc 27's (G26-2) by doc 29 §5; this program's own share of it is G30-2 (§7.4).
- **M-A8, SCOPE-CERT, the dated integrity co-runner (reported, never ruled here; never headline).** Co-runner 2026-09-24 17:05–18:14 (`analysis/scope_cert.md`; registration 17:18:18 plus two addenda; tokens 9,000,049 / 9,000,059; canon pins byte-identical): **DONE**. Adversarial verifier 18:34–20:21 (`analysis/scope_cert_verifier.md`; two instances, the first died on a network error; tokens 9,100,051 / 9,100,061; 78 pins byte-identical): **CONFIRMED-WITH-DEFECTS**, the computed content reproduces by disjoint routes; the defects are in the drafted riders' wording and one disclosure, and no number is wrong. Its answer to doc 29's w = 4 question: **yes**, the recurring gravity anchor at lab-dipole width w = 4 carries the classical anchor reading 1/(νm̃₂) with an empty quantum rational set at every row (400 rows in the verifier's own route); off the registered width the quantum rational self-reading set is empty at all 3,600 rows. Its riders are **drafted, not entered**, for the owner's post-2026-10-30 block (doc 27 §7.7: the G26-3 re-ruling and the G26-4 decision on the doc 13 §4 row). None of it bears on the exclusion table, and this document rules none of it.

### 1.4 Registered predictions, fate (declared parity; refutations reported first; near-corollaries never findings)

**The order of record's predictions (doc 29 §5):**

| prediction | forecast | outcome | ruled by |
|---|---|---|---|
| P-A-ctrl (MANDATORY): the census re-types the four adjudicated rows (Aziz–Howl; the van Hove hybrid; 2605.20330; 2512.14834 v2) exactly | mandatory | **FAIL at the first run** (the quote-comparison layer) → KA-ctrl → dated repair R3 → **PASS-AFTER-DATED-REPAIR**; reproduced 4/4 by the gate's own matcher and re-typing | pilot; gate |
| P-A1: ≥ 75 % of the rows DAF-BY-DEFINITION | ~70 % | **TRUE** (pilot 23/23; gate 22/23 ≥ 18) | gate |
| P-A2: at most two rows NEEDS-REPRODUCTION | ~70 % | **TRUE** (one: R04) | gate |
| **P-A3: no reproduction contradicts a paper's own stated computation** | ~90 % | **FALSE by its letter**, Struyve's eq. (31) / (46) / Fig. 1 (×2 in the NS phase); verdict-neutral; KA-1 by its letter | reproduction |
| P-A4: Theorem W stands (no row inside W1–W5 with a Definition-2-classical mediator and a Definition-1 positive) | ~99 % | **TRUE** at every leg | all legs |
| P-A5: N-1 is an instance of E1's committed clause | ~55 % | **TRUE**, INSTANCE, decided by committed text; ×5.358 at equal reaches on the hold-only phase; the descriptor "an O(Δφ²)-order single-mass deficit" **refined**: the forced deficit is first order in the entangling phase | pilot; gate |
| P-A6: M-1 branch (F\* < 0.864790 / band / ≥ 0.975555) | 55 / 40 / 5 % | **branch 1** (F\* = 0.4473); branch 3 was excluded in advance by the pilot's registered near-corollary NC-M1 | pilot; gate; rigor |
| P-A7: the saturation map's magnitudes regress; v_DI = 0.98799697; SATURATED | ~99 % | **TRUE**; its parenthetical "at the committed geometry, the light-crossing phase is 3.3 × 10⁻¹³ rad" is doc 27's parametrization (record-side candidate (E)) | pilot; both M-A6 legs |

**The judge's odds (doc 29 §4.3), against the outcome:** PILOT-ADJ returns a typed census with its control reproduced (~95 %), TRUE, after the dated repair; SCOREBOARD-BY-DEFINITION (~45 %, the leading forecast), **not reached**; SCOREBOARD-WITH-ROWS (~35 %), reached; KA-4 (~12 %), not; ROW-FLIP (~5 %), reached by its letter; COUNTEREXAMPLE (< 1 %), not; the note assembled and through its two adversarial legs within one cycle (~85 %), TRUE (Final v1.0 on 2026-09-25, the day after the judging); SCOPE-CERT lands its riders at record grade before the post-meeting block (~90 %), the co-runner and its verifier are done and the riders are drafted; the block itself falls after 2026-10-30. The judge's hand estimate that "put F\* within reach of 0.864790 from either side" (doc 29 §4.3 (1)) is **refuted**: F\* = 0.4473. The forecast that ~75 % of the rows would be decided by reading (doc 29 §4.2 (i)) is exceeded: 22 of 23.

**Each leg's own forecasts (every refutation is a result; full tables in each record):**
- *Pilot* (R1, R2-J): **refuted**, DAF 17–20 of 23 (it found 23; the gate then 22); NOT-INSTANTIABLE 1–4 (0); the control 4/4 on the first run (2/4); the M-1 point forecast F\* ∈ [0.70, 0.92], median 0.83 (0.4473). Confirmed, NEEDS-REPRODUCTION 0–2, P-A4 / P-A7, N-1 INSTANCE, M-1 branch 1, F\* < F_magic, a facet normal with denominators ≤ 16 (≤ 3), the battery, the token fingerprints. Its "KA-0 fires" (~20 %) held in the pilot and was then refuted by the gate.
- *Gate*: **refuted**, type agreement on all 23 rows (22); an own facet with entries in {0, ±1/3, −1} (its facet has entries in {−1, 0, 1}); every pilot quote found by its first extractor (87/90, the other three extraction artefacts confirmed by a second); PASS-KA0-NOTE (PASS); F\*_CSP equal to the octahedron bound (F\*_CSP = F\*_Cliff at float grade).
- *Lanes*: none refuted (6 R0 + 21 R1 forecasts).
- *Reproduction*: its own F1–F16 all confirmed; the one refutation it produced is doc 29's P-A3.
- *Assembly*: **refuted**, CK-A clean on its first run (run1 crashed and run2 failed on check-code defects, not on the note).
- *Rigor leg*: **refuted**, its R1 description of the pilot's census route (a Pauli-transfer closure, not a unitary one; corrected by R1-A before any M-1 computation) and FC-2 (the SAT-1 defect sat in the body labels, not in Appendix A).
- *Steelman*: **refuted**, a 1/c² residual for a smooth closed split / hold / recombine sequence (it is 1/c⁴); λ\* equal to the L1-ball bound 0.764027 (λ\* = 0.514230, LP-located); the sixth decimal of λ_ent (0.924876042); the completion's frame maximum exactly 1 + C − C² (1.131908, off the symmetric frame).
- *Finalization*: **refuted**, every fold present on the first check run (one ×5.358 lacked its phase label).
- *This document's leg*: R0's F0.1–F0.10 all TRUE (§1.1); R1's FR.1–FR.6 TRUE (regression 21/21; certifications 22 primes and 3 composites; no draw at or above 9,300,000; the leg manifests; the pilot's five files; §5); FR.7–FR.12: FR.7 TRUE (the first document check run was all clean); FR.8 TRUE; FR.9 TRUE (no trigger fired at the end); FR.10 NOT EXERCISED (the end check ran at 01:11:47, before the 02:00 CEST announcement, and the control stayed 609); FR.11 TRUE (497/497 byte-identical); FR.12 TRUE (21/21).

### 1.5 The kill-criteria ledger (doc 29 §5, each by its letter)

| criterion | its letter (abridged; verbatim in doc 29 §5) | ruling | ruled by |
|---|---|---|---|
| KA-ctrl | P-A-ctrl fails ⇒ the census method is broken; stop before any other row; repair dated; re-run | **FIRED** at the pilot's first control run (18:39:08); repaired (R3, 18:39:51); re-run 4/4 (18:39:54); the census proceeded | pilot; gate (own route 4/4) |
| KA-0 | every row DAF-BY-DEFINITION and N-1 decided by committed text ⇒ "a note, not a front" | **NOT FIRED** (R04 NEEDS-REPRODUCTION); the pilot's "fires" refuted; the phrase not entered | gate |
| KA-1 | a reproduction contradicts a paper's own stated computation ⇒ reported first; a touched W / M premise is re-audited | **FIRED BY ITS LETTER** (eq. (31) / (46)); verdict-neutral; no W / M premise touched, so the re-audit clause did not apply | reproduction |
| KA-2 | a model inside W1–W5 with a Definition-2-classical mediator delivering a Definition-1 positive ⇒ stop everything | **NOT FIRED** at any leg; the letter's "W1–W5" against the theorem's W1–W6 flagged for the owner (candidate (D)) | all legs; steelman (the gap) |
| KA-3 | an assumption-ledgered classification of the 2026 model set posts elsewhere ⇒ cite-and-distinguish | **NOT FIRED** (K4 fails for Di Biagio's Table II); OCCUPIED-BELOW-THRESHOLD; every later leg's nets 0 | lanes; every later leg |
| KA-4 | more than two rows need computation ⇒ stop; the owner decides | **NOT FIRED** (one row) | gate |
| KA-5 | wording bans | **held**: every leg's report scanned by code (0 banned strings outside the ban lists and the verbatim quotations); this document scanned by CK-D (§5) | every leg |
| KA-6 | a trigger posts mid-run ⇒ its row enters first | **NOT FIRED** at any leg (§1.1, §1.3 M-A7) | every leg |

### 1.6 Parity record (findings against expectations, refutation-first)

(i) **The pilot's headline did not survive its gate.** The pilot typed all 23 rows DAF-BY-DEFINITION and ruled that KA-0 fires; the gate re-typed R04 as NEEDS-REPRODUCTION because one null-column verdict rested on a small calculation that no text and no committed certificate states, labelled "near-corollary" only after the readout. The frozen-rule clause worked as designed: a label introduced after a readout carried no verdict. The owed calculation then confirmed the pilot's number and verdict exactly, only the row type moved. (ii) **Doc 29's P-A3 (~90 %) is FALSE by its letter.** The one reproduction the program owed was registered over the row's whole calculation section, not only the mandated equations, and that wider scope found a factor-2 slip in a sibling member's printed phase (the NS member, eq. (31)), verdict-neutral, touching no premise, reported first. (iii) **The first control run failed** on a quote-comparison artefact (a line-end hyphen, typographic quotes); the typing layer had matched 4/4; recorded as PASS-AFTER-DATED-REPAIR, never as a first-run pass. (iv) **M-1 landed far from every forecast:** the pilot's point forecast [0.70, 0.92] and the judge's "within reach of 0.864790" are both refuted by F\* = 0.4473; the physics reading is branch 1, the ordering the doc 29 odds favoured (~55 %). (v) **P-A5's descriptor was wrong in its order:** the forced deficit is first order in the entangling phase (worst-case exponent Δφ/4), not O(Δφ²). (vi) **A judge-side decided-at-freeze miss:** doc 29 priced P-A6's third branch at ~5 % although a near-corollary excludes it (a channel in the Clifford hull maps |++⟩ to a stabilizer mixture, so F\* ≤ F_magic always); the pilot registered it as NC-M1 before computing, never as a finding. (vii) **A size printed with the wrong label since doc 27:** SAT-1's 3.3 × 10⁻¹³ rad is doc 27 §3.9's value at Δφ ≈ 1 rad over 2 s and d = 200 µm, a supremum over time in a step-onset model; at the committed operating point the same form peaks at 8.38 × 10⁻¹⁴ rad and the discrepancy of the second difference is exactly 0 from 2.33 ps into the hold (both M-A6 legs, exact; SATURATED unaffected; record-side candidates (A) and (E)). (viii) **The v0 note stated two lab conditions too weakly** (a null with full visibility as a flat ratio alone; W′'s null without Δ̂): the steelman blocked export on both, and Final v1.0 carries the repaired conditions; every verdict cell was right under the note's own premises. (ix) **The literature moved under the rows:** four on-disk texts were stale, one with a changed diagnosis (2604.19696 v3: TRANSFER by diffusion); the currency check now runs before any quotation. (x) **Against process:** the pilot died once mid-run and resumed on its frozen registrations; the gate's D3 (six abstracts fetched before R2's freeze; no R2 rule depends on them); the lanes read 19 entries at claim-sentence depth; the reproduction ran outside the binding order's listed sequence (disclosed; the owner's to rule); the rigor leg drew two tokens doc 29 had not pre-assigned (9,100,067 / 9,100,079, inside the verifier band; disclosed and certified); this leg's three code-level corrections (§5). (xi) **In the judge's favour:** COUNTEREXAMPLE did not arise and Theorem W stands on every row; KA-4 did not fire (one reproduction, not three); the note passed both adversarial legs within one cycle; every verdict number regressed exactly at every leg (the pilot's battery 15/15, the gate's, the rigor leg's 31/31, the steelman's 14/14, the finalization's 15/15 certifications and this leg's 21/21).

---

## 2. Literature position (the rows are the literature; occupancy; the watch)

**The model set.** M26 is 23 rows frozen in the pilot's R1 before any row was read, Schlegel–Kumar–Paterek–Dakić 2605.20330 (R01, the first row); Schlegel–Dakić–Del Santo 2512.14834 v2 (R02); Aziz–Howl (Nature 646, 813) with its seven comments and rebuttals 2511.07348, 2511.20717, 2511.00852, 2511.19242, 2512.13675, 2604.19696, 2604.16276 (R03); Struyve 2510.20991 (R04); Di Biagio 2511.02683 (R05); Tibau Vidal–Iyer 2607.03429 (R06); Płodzień et al. 2605.06577 (R07); Plávala 2508.03052 (R08); Marletto–Vedral 2507.11713 with 2608.17688 (R09); the van Hove hybrid 2601.21555 (R10); Ludescher–Loveridge–Galley–Müller 2507.13201 (R11); Oppenheim's postquantum classical gravity (PRX 13, 041040) with Oppenheim–Sajjad 2605.05375 (R12); Fabiano–Fujita–Matsumura–Carney 2603.26075 v2 (R13); Li–Miki–Chen 2605.26215 (R14); Tomizuka–Takeda 2604.06891 (R15); the device-independent Kent–Pitalúa-García 2109.02616 (R16) and Das–Alimuddin–Storz–Chu–Bose 2608.08778 (R17; Athreya et al. 2609.02009 as experimental context, never typed); the platform / theory rows Wakakuwa 2607.19141 (R18), frame dragging 2606.31678 (R19), Braccini–Serafini–Bose 2602.19306 (R20), quadratic curvature 2609.22501 (R21), Salih 2609.04104 (R22) and Nandi 2609.09931 (R23), 32 texts in all (23 primaries, 9 sub-entries). Every quotation in the note is from the latest arXiv version (32/32 latest on the arXiv API at 2026-09-24 23:26:15, the rigor leg); the four texts the pilot read at an older version (2512.13675, 2604.19696, 2511.02683, 2508.03052) carry the latest version's sentences, and their verdicts are unchanged.

**Occupancy of the ledger shape: OCCUPIED-BELOW-THRESHOLD (KA-3 not fired; §1.1).** Di Biagio's Table II is the one prior per-account assumption ledger of GIE models; it is cited and distinguished on every export, and the record already engages it as the "adversarial mini-ledger" (doc 06 v1.1). The RMP reviews (2311.09218, 2410.07262) fail K4 by date; Foo–Mann–Zych (2302.03259, npj Quantum Inf. 12 (2026)) is an assumption critique, not a classification; Goto et al. 2606.30853 and Anastopoulos 2609.22890 carry no ledger.

**Accounts outside M26, listed with no verdict (the frozen-rule clause: "a row or category added after reading is flagged and carries no verdict"; each enters only by a dated re-registration).** Trillo–Navascués 2411.02287 (the Diósi–Penrose model predicts GIE; the record's own typing is doc 06 v1.1's reconciliation, all of it the direct gate); Angeli–Carlesso 2501.16253; Lin–Mondal 2510.23584; Tibau Vidal–Marletto–Vedral–Chiribella 2506.21122 (a "locally classical" mediator with non-locally-tomographic coupling, Definition 2 (K) requires standard composition); Marletto–Vedral 2510.19969 (MOVW's ref. [8], never their [7]); Spalvieri–Garmier–Giacomini 2512.19806; Joseph 2609.26818 (a squeezing proposal; the one account new since the doc 29 judging); Howl–Cooper–Hackermüller 2304.00734 (PRA 114, 023306 (2026)); Muciño 2606.00255; Miki–Li–Chen 2605.26240; Saldanha–Marletto–Vedral 2602.12266 (a non-entanglement witness, which Theorem W does not type); Zarei et al. 2510.20587, twelve in all (30a §10).

**The must-cite block (30a §11, carried on every export):** Theorem W′'s anchors (Matsui 2607.20867; 2510.20991; Trillo–Navascués 2411.02287's closing remark); the classical-carrier genre (2512.14834 v2 next to 2605.20330); the inherited magic-line block of 21a §9 (with M-1, a Paper-M row); the lanes' adds 2210.14921 (Perche–Ragula–Martín-Martínez, SAT-3's scope) and 2302.03259; the W-row citations 2208.09489 (Martín-Martínez–Perche) and 2512.17024 (Boulle–Franzmann); the row-level canonical citations (2601.21555 for "reclassified, not refuted"; Di Biagio's Table II; 2507.13201; 2603.26075 v2 behind N-1; 2109.02616 and 2608.08778 behind SAT-6).

**Method note (affects grading).** INSPIRE is the citer source of record (`refersto:recid:`), `abstracts.value:` the abstract net, arXiv queried at ≥ 4 s spacing with a window control and a `cat:quant-ph` control; the lanes read 13 entries in full and 19 at claim-sentence depth (abstract, introduction, conclusions and every claim sentence caught by a registered keyword net), the residual risk is body text outside that net; the rigor leg re-typed 13 of the 23 rows from full text by a pre-registered selection rule and moved no value class.

**Watch list (G30-5):** MOVW ref. [7] (the version of 2511.07348 and the two title nets); the device-independent turn's citers (2608.08778, 2109.02616, 2609.02009); the noise-floor program's citers (2603.26075, 2605.26215, 2604.06891); the classical-carrier genre (2605.20330, 2512.14834 v2, 2607.13135); the twelve flagged accounts' successors; the Perimeter "Relativity Reframed" contributions after 2026-10-30 (G30-2).

---

## 3. Results

### 3.1 Run record, pins, primes (green before any verdict row)

| leg | window (CEST) | registrations, frozen before the computations they govern | canon pins | tokens | ruling |
|---|---|---|---|---|---|
| doc 29 judging | 2026-09-24 ~15:18–16:32 | `dq4_judge_probe/registration.txt` 16:30:24 | 43 byte-identical | none | the program recommended at DUTY grade |
| SCOPE-CERT co-runner (M-A8) | 17:05–18:14 | 17:18:18 (+ 2 addenda) | 58 entries (57 distinct paths), byte-identical | 9,000,049 / 9,000,059 | DONE |
| pilot (M-A1) | 18:12:54–18:41:35 and 19:29:56–20:27:23 | R1 18:21:42; R2 18:33:29; R3 18:39:51 | 154 (README changed externally, 20:24:03) | 9,000,011 / 9,000,041 | 23/23 DAF (later 22 + 1); N-1 INSTANCE; M-1 branch 1 |
| SCOPE-CERT verifier (M-A8) | 18:34–20:21 | 18:38:09 (+ 5 addenda) | 78 byte-identical | 9,100,051 / 9,100,061 | CONFIRMED-WITH-DEFECTS |
| gate (M-A2) | 20:37:50–21:22 | 20:46:34; addendum 21:06:56 | 209/209 | 9,100,009 / 9,100,033 | PASS; R04 NEEDS-REPRODUCTION |
| reproduction (M-A4) | 21:27:07–21:58 | 21:43:15 | 232/232 | none | R04 reproduced; KA-1 by its letter |
| lanes (L1–L4) | 21:28:14–22:11 | R0 21:28:33; R1 21:34:45 | 282/282 | none | no kills; KA-3 not fired |
| assembly (M-A3) | 22:19:24–22:49 | R0 22:19:57; R1 22:24:37 | 376/376 | none | 30a v0 |
| rigor (M-A6) | 22:49–23:48 | R0 22:57:01; R1 23:13:43; R1-A 23:18:24 | 396/396 | 9,100,067 / 9,100,079 | PASS-WITH-REPAIRS |
| steelman (M-A6) | 23:03–23:37 | R0 23:04:14; R1 23:17:08; R2 23:25:53 | 399/399 | 9,200,017 / 9,200,021 | PASS-WITH-REPAIRS |
| finalization | 2026-09-24 23:59:41 → 2026-09-25 00:37 | R0 00:00:36; R1 00:12:00 | 465/466 (only the note, its object) | none | 30a Final v1.0, not HELD |
| this document | 2026-09-25 00:40 → 01:13:41 (pins after; the manifest last) | R0 00:45:59; R1 00:53:29 | 497 + 21 paper files: 497/497 and 21/21 byte-identical | none | this document |

Nothing in the program was decided by a prime; every token was a regression pin (mod-p census fingerprints, mod-p replays of exact identities, residues of exact rationals). The closing-audit reserve ≥ 9,300,001 is untouched (§6).

### 3.2 The table at a glance (23 registered rows; 30a §4, transcribed)

Type: **DAF** = DAF-BY-DEFINITION; **REPRO** = NEEDS-REPRODUCTION, reproduced of record. N gives (N-a) Theorem W′ / (N-b) a null with full single-mass visibility. "where resolvable" = wherever the row's predicted phase is resolvable at the probed regime.

| row | account | type | (P) a certified positive | (N) a null: (N-a) / (N-b) | (S) saturation map |
|---|---|---|---|---|---|
| R01 | Schlegel–Kumar–Paterek–Dakić 2605.20330 *(control)* | DAF | COMPATIBLE: its classical reproduction does not reach the committed protocol (its own regime clause) | not reached / LEAVES-STANDING | NONE |
| R02 | Schlegel–Dakić–Del Santo 2512.14834 v2 *(control)* | DAF | COMPATIBLE: the spin witness is its "incompatibility beyond phase-space measurements", paired with Wigner-negative states | LEAVES-STANDING (no null claim) | NONE |
| R03 | Aziz–Howl, Nature 646, 813 (2510.19714) + 7 comments *(control)* | DAF | REFUTED-BY-POSITIVE: c-number field, Definition 2 (D); Corollary W2; prong (I.3) | LEAVES-STANDING / LEAVES-STANDING | NONE |
| R04 | Struyve 2510.20991 v2 (separable semi-classical potentials) | REPRO | REFUTED-BY-POSITIVE: (D); Corollary W2 names "Struyve" | LEAVES-STANDING / NS: LEAVES-STANDING; **NSB: REFUTES, thin ×1.2474, Born-rule and readout conditional** | NONE |
| R05 | Di Biagio 2511.02683 v5 | DAF | RECLASSIFIED-NOT-REFUTED: its entanglers are the direct pair gate (Diósi–Penrose) and quantum matter (Aziz–Howl) | refutes each listed GIE-predicting theory where resolvable; thesis stands / no forced deficit posited | NONE |
| R06 | Tibau Vidal–Iyer 2607.03429 | DAF | REFUTED-BY-POSITIVE: the AH class; its diagnosis is TRANSFER | LEAVES-STANDING / LEAVES-STANDING | NONE |
| R07 | Płodzień et al. 2605.06577 | DAF | RECLASSIFIED-NOT-REFUTED: the entangler is the linear pair operator (the direct gate) | REFUTES the pair term where resolvable / LEAVES-STANDING | NONE |
| R08 | Plávala 2508.03052 v4 | DAF | COMPATIBLE: an indirect inference, not a Definition-1 certificate | REFUTES its derived positive (by its own contrapositive) /, | NONE |
| R09 | Marletto–Vedral 2507.11713 (Koopmanian) + 2608.17688 | DAF | REFUTED-BY-POSITIVE: commuting mediator; prongs (I.2) / (I.5) | LEAVES-STANDING / LEAVES-STANDING | NONE |
| R10 | Ulbricht–Bermúdez Manjarres–Reginatto 2601.21555 (van Hove) *(control)* | DAF | RECLASSIFIED-NOT-REFUTED: non-commuting van Hove observables | REFUTES (it predicts the quantum secular phase) / no forced deficit stated | **mediator-sector algebra: 0 in every mass-side statistic** |
| R11 | Ludescher–Loveridge–Galley–Müller 2507.13201 | DAF | REFUTED-BY-POSITIVE: commutative C\*-algebra mediators; prong (I.2) | LEAVES-STANDING / LEAVES-STANDING | NONE |
| R12 | Oppenheim PRX 13, 041040 (1811.03116) + Oppenheim–Sajjad 2605.05375 | DAF | REFUTED-BY-POSITIVE: (O); scoped Corollary C5; prong (I.2) | LEAVES-STANDING / **REFUTES** (committed E1 clause, through the single-mass scan; corroborated by N-1) | NONE |
| R13 | Fabiano–Fujita–Matsumura–Carney 2603.26075 v2 | DAF | REFUTED-BY-POSITIVE: its named non-entangling members | non-entangling members LEAVE-STANDING / **N-1: REFUTES, ×5.358** (hold-only phase; both reaches at REACH) | NONE (its entropic/graviton split has no committed size) |
| R14 | Li–Miki–Chen 2605.26215 | DAF | COMPATIBLE: a platform threshold (coupling must beat thermal noise) | a noise-dominated null is not at sufficient sensitivity: LEAVES-STANDING | NONE |
| R15 | Tomizuka–Takeda 2604.06891 | DAF | REFUTED-BY-POSITIVE: the effective Markovian CQ description only | DIAGNOSTIC-ONLY / its Markovian limit is R12's class | NONE |
| R16 | Kent–Pitalúa-García 2109.02616 (device-independent) | DAF | COMPATIBLE: discharges W6 (ii) by spacelike separation | LEAVES-STANDING (no null claim) | **DI price: v_DI = 0.98799697 (unfiltered CHSH; 0.956106 filtered) vs 0.864790** |
| R17 | Das–Alimuddin–Storz–Chu–Bose 2608.08778 (device-independent) | DAF | COMPATIBLE: Theorem W's disjunction per source–probe pair | LEAVES-STANDING (no null claim) | NONE new (the Fork-1 class inside each pair) |
| R18 | Wakakuwa 2607.19141 (clock GIE) | DAF | COMPATIBLE: a CZ(Δ)-type gate on internal energies | REFUTES the pair term where resolvable; Scheme 1 bears on LPI only | NONE new (τ_π) |
| R19 | Wakakuwa–Petruzziello–Lantano–Huelga–Plenio 2606.31678 (frame dragging) | DAF | COMPATIBLE: linearized quantum gravity | REFUTES where resolvable | **retardation = the Fork-1 class (SAT-1: ≤ 3.3 × 10⁻¹³ rad in doc 27's parametrization; 8.38 × 10⁻¹⁴ rad at the committed point)** |
| R20 | Braccini–Serafini–Bose 2602.19306 (SGI) | DAF | COMPATIBLE: operator-valued Newtonian coupling | REFUTES where resolvable | **Clifford-exception coupling, τ_π = 25.0136 s** |
| R21 | van Manen–Blankenstein–Mazumdar 2609.22501 (quadratic curvature) | DAF | COMPATIBLE: Stelle gravity | REFUTES where resolvable | NONE (the Yukawa terms shift Δ by −5.3 % to −7.1 % at λ = 52 µm but split neither reading) |
| R22 | Salih 2609.04104 v6 (a horizon) | DAF | COMPATIBLE: the retarded potential is an effective quantum field by its own text | LEAVES-STANDING (no null claim) | NONE at the committed geometry |
| R23 | Nandi 2609.09931 (a GW mediator) | DAF | REFUTED-BY-POSITIVE: a classical GW background, (D) (scope rider) | LEAVES-STANDING the classical-GW class | NONE at the committed geometry |

**Tally.** P classes: **9 REFUTED-BY-POSITIVE, 3 RECLASSIFIED-NOT-REFUTED, 11 COMPATIBLE** (the pilot and the gate agree 23/23). Types: **22 DAF-BY-DEFINITION + 1 NEEDS-REPRODUCTION (R04)**. No row is NOT-INSTANTIABLE, OUT-OF-SCOPE, OUTSIDE-PREMISES or a counterexample; one sub-entry (2608.17688) is OUT-OF-SCOPE and carries no verdict. The zero-computed-content test and the Lemma-M fingerprint check passed on every P-bearing row. The committed geometry of every cell: m = 10⁻¹⁴ kg, d = 450 µm, Δx = 250 µm, τ = 2.5 s, Gm²/ħ = 63.3 µm·rad/s, hence **Δ = −211/672 = −0.313988 rad** (hold segment only), C = |sin(Δ/2)| = 0.156350, witness value 1 + C = 1.156350, F_ent = 1/(1 + C) = 0.864790. The R01 rider's geometry condition: at δx/d = 5/9 the quadratic term of the potential carries 56/81 of the committed Δ and the beyond-quadratic terms 25/81 (30.9 %), so the committed protocol lies outside that row's quadratic regime as well as outside its Wigner-positive and quadrature conditions.

### 3.3 Cell N-1, the minimal-noise class of arXiv:2603.26075 at the committed geometry: INSTANCE

The row proves that every Galilean-invariant, time-local, Newton-on-average model that does not entangle must inject a minimal noise. **First act, decided by committed text:** its floor (eqs. (34)–(36), App. B) requires 4Γ₁Γ₂ ≥ α_G²δx⁴ + β²δx⁴/4, hence Γ₁ + Γ₂ ≥ α_Gδx², the noise rate is at least the entangling rate, which is E1's committed H1b clause ("forced to decohere at a rate tied to the coupling that would have entangled") in the row's generality, an extension of the Kafri–Taylor inequality by the row's own account. **Evaluation (a confirmation of record, exact):** α_Gδx²/ħ = 211/4860 s⁻¹; with visibility exponents E_a = 2Γ_aτ, E₁ + E₂ ≥ 211/972 and E₁E₂ ≥ (211/1944)², so max_a E_a ≥ 211/1944; with both statistics read at REACH the adversary's best split of that mass's noise gives E = 211/3888 on each channel; **minimal forced deficit 1 − e^(−211/3888) = 0.0528232369 against REACH = 44521/4515840 = 0.0098588524, ×5.358 (hold-only phase).** The gate re-derived the row's eq. (34) itself from two dephased qubits (the bilinear coupling 2Gm²/d³ = 2α_G, the one R14 prints) and the rigor and steelman legs confirmed the exponent convention E_a = 2Γ_aτ from the row's own definition of Γ_a.
**Two reaches (the steelman's D1; a lab-facing condition introduced after readout, carrying no verdict):** E1 states its reach for the ratio R only (c < 0.1 at 2σ); the partner-independent share of a forced noise goes to the single-mass scan, whose reach u_s the campaign must state. The class is cornered iff −ln(1 − r_R) − ln(1 − u_s) ≤ 211/1944; at r_R = REACH this is u_s ≤ 1 − e^(−211/1944)/(1 − REACH), printed as u_s ≤ 0.0939 (certified value 0.0939233). The member a flat ratio alone lets through carries all of its forced noise partner-independent: R ≡ 1 on both masses and V_s = e^(−211/1944) = 0.897144 on each.
**Data, no verdict:** with the exact committed phase in place of the row's quadratic expansion the deficit is 0.075495 (×7.66), the quadratic-expansion phase 211/972 is the exact |Δ| times 56/81, so the expansion is conservative; symmetric CQ-form noise gives a V_s deficit of at least 0.102856; the forced deficit is first order in the phase; on E1's total-phase contours (0.5–0.9 rad at the design point) the conservative ratio is ×3.31–×1.78, reaching 1 only at a total phase of 1.52 rad.
**Scope:** time locality is the row's premise (non-Markovian classical-quantum dynamics, R15, lies outside); both masses must be read (V_c from the pair shots, V_s from left-only and right-only single-interferometer runs in the same run blocks); the row's dissipator depends on positions only, a choice it calls relaxable; the class's entangling members are not N-1's object. **N-1 is a statement about what a future null would do, never an exclusion claimed today, and a deficit of that size tracking Δφ is a diagnostic, never a certificate.**

### 3.4 Cell M-1, the gravitational gate's own Clifford-hull threshold: F\* = 1/(1 + 4|sin Δ|), branch 1

**The object:** F\* = sup{F : F·𝒰_Δ + (1 − F)·𝒟 ∈ conv(two-qubit Clifford unitary channels)} at Δ = −211/672, with 𝒰_Δ the channel of CZ(Δ) in the calibrated frame and 𝒟 the completely depolarizing channel; the admissible F form the interval [0, F\*]. The cell was live because doc 21 typed the unitary gate and 21a's 0.975555 is a threshold on states. **Result: F\* = 1/(1 + 4|sin Δ|) = 0.447343871026845, certified from both sides by three disjoint routes:**

| | pilot | gate | rigor leg (M-A6) |
|---|---|---|---|
| census | own ⟨H, S, CNOT⟩ closure (11520), equal as a set to the committed engine's (isolated copy) | own Sp(4, F₂) enumeration, 720 × 16 = 11520 | two own routes equal as sets, incl. a four-class normal form 576 + 5184 + 5184 + 576 with no complex arithmetic |
| upper side | functional with entries in {0, ±1/3, −1}; ⟨W, R_U⟩ = 1 − 4 sin Δ | own facet, entries in {−1, 0, 1}, 29 nonzero, 6858 tight | own facet invariant under a 128-element group, 31 nonzero entries in {±1/2, ±1}, 3634 tight |
| lower side | explicit decomposition over 220 Clifford unitaries (point classes 1, 192, 3, 24), weights affine in (cos Δ, sin Δ), positive at Δ | own 185-element vertex decomposition, an exact identity in (cos Δ, sin Δ) | own orbit decomposition over 50 Clifford unitaries, minimum weight 0.0244453 |
| certified width | < 10⁻⁴⁰ | 3.3 × 10⁻¹¹⁷ | 5.7 × 10⁻⁷¹ |

**Branch 1 (F\* = 0.4473 < F_ent = 0.864790 < F_magic = 0.975555).** For every white-noise visibility in (0.4473, 0.8648] the noisy gate is outside the convex hull of Clifford unitary channels while the committed witness is silent; for every visibility where the witness can fire the gate is certifiably outside that hull; there is no visibility band in which the witness fires while the noisy gate is already a Clifford mixture, the anti-alignment of doc 21 (a maximal witness at Δ = π on a gate with no magic) has no noise-level counterpart at the committed geometry under this benchmark. The ordering holds for every entanglement certificate, not only the committed witness: the PPT threshold of the benchmark state is 1/(1 + 2C) = 0.761789.
**Scope and wording.** (i) White noise only; under the campaign's own branch-diagonal dephasing (steelman data, re-certified by the finalization, not of record) the hull question reduces exactly to the 32 diagonal Cliffords, a certified one-mass bound gives λ ≤ 1/(1 + |sin Δ|) = 0.764027, an LP locates the symmetric boundary at λ\* = 1 − √(|sin Δ|/(1 + |sin Δ|)) = 0.5142 (LP-located), while the dephased state is entangled only above 0.855799 and the committed witness fires only above 0.924876, branch 1 survives the change of noise model. (ii) Against the larger completely-stabilizer-preserving set the gate's float LP located F\*_CSP = F\*_Cliff: LP-located, not certified. (iii) M-1 is a property of the modelled channel; a laboratory magic certificate still needs Definition 1 and M0–M4 (Theorem M), and a lab measures M-1 only by process tomography, which the campaign does not run; its lab-facing content is a consistency guarantee. (iv) A gate-typing number, a Paper-M row, never a law, never a statement about the mediator, never a hardness claim.

### 3.5 The saturation map, what no BMV-class experiment can split (SAT-2..SAT-6 at the committed geometry; SAT-1 in doc 27's parametrization with its committed-point value)

| entry | discriminator | what it would split | size | rows | status |
|---|---|---|---|---|---|
| SAT-1 | the light-crossing phase \|V_AB\|·d/c | Theorem W's reading (I), a mediator, from (II), a direct gate | ≤ 3.3 × 10⁻¹³ rad in doc 27 §3.9's parametrization (Δφ ≈ 1 rad over 2 s, d = 200 µm; a supremum over time in a step-onset model); **8.38 × 10⁻¹⁴ rad at the committed point**, a transient that is exactly 0 from 2.33 ps into the hold | R17 (per pair), R19, R22 | SATURATED |
| SAT-2 | retardation / locality timing | an onset delayed by ≥ d/c against onset at t = 0⁺ | d/c = 0.667 ps at 200 µm, against a 2.5 s hold | R19 | SATURATED (infeasible) |
| SAT-3 | field quantization beyond non-commutativity | a mediator's own radiated quanta | ⟨n_g⟩ ≈ 1.3 × 10⁻⁵⁷ per run (C²-smooth quintic ramps; transcribed from doc 06 W1 and 05a §9) | R23 | SATURATED |
| SAT-4 | the mediator-sector algebra | van Hove prequantum vs quantized linear gravity | 0 in every mass-side statistic (the polaron identity) | R10 | SATURATED (needs mediator-sector probes) |
| SAT-5 | the Clifford-exception coupling (Theorem M) | a maximal witness on a gate with no magic | τ_π = 1680π/211 s = 25.0136 s (hold-only; a family theorem, never a lab point) | R18, R20 | outside every registered platform |
| SAT-6 | local-hidden-variable matter statistics vs quantum statistics (W6 (ii)) | the witness's reliance on standard quantum mechanics for the masses | two-setting CHSH above v_DI = 1/√(1 + sin²(Δ/2)) = 0.98799697 unfiltered at perfect detection, above 0.956106 with local filtering; the minimum over all Bell tests not committed; against the witness's 0.864790 | R16, R17 | split only by a Bell test |

**SAT-1 at the committed operating point** (both M-A6 legs, exact; re-certified by the finalization and regressed by this leg): applied branch by branch, Θ_b(t) = V_b·max(0, t − s_b/c) with V_b = Gm²/(ħ s_b); every branch's offset V_b·s_b/c equals Gm²/(ħc), so the offsets cancel in the second difference; the retarded-minus-instantaneous discrepancy peaks at (25/63)·Gm²/(ħc) = 8.38 × 10⁻¹⁴ rad at t = (d − Δx)/c = 0.667 ps, the same number as (|Δ|/τ)(d − Δx)/c, and is exactly 0 from t = (d + Δx)/c = 2.33 ps on, long before readout. For a mass-conserving smooth split the retarded potential has no O(1/c) term at all (steelman data; a Liénard–Wiechert toy leaves a 1/c⁴ residual). **SATURATED holds: no row supplies a discriminator between Theorem W's two readings above these sizes; the disjunction is the permanent deliverable of the BMV class.** The only on-paper decider is Corollary W1's cited embedding argument, which the table never consumes.
**Phase basis.** Every margin and threshold above is on the hold-only Δ. E1's contours are fixed in total accumulated phase, 0.5–0.9 rad at the design point; there NSB's ratio runs ×1.2435–×1.2290, N-1's conservative ratio ×3.31–×1.78, F\* 0.343–0.242 against F_ent 0.802–0.697, and v_DI 0.971–0.917 (this leg's regression, N20). No value class changes over that range.

### 3.6 The one reproduction (R04, Struyve 2510.20991 v2) and the KA-1 letter

The reproduction followed the van Hove template: the paper's own calculation first, then the cell typed. **Reproduced exactly:** eqs. (35) (all 16 branch amplitudes) and (47) (the NSB witness, W_NSB = 0.100298105048 at the committed geometry, by the paper's own average and by Tr(ρ_ens W)), and (28), (43), (44), (45). **Contradicted:** eq. (31), the printed Newton–Schrödinger mean-field phases are exactly twice what the paper's eq. (29) gives on its own normalized state (21); the only reading that restores the print sums the partner's potential over both of its branches instead of averaging it, which the paper's normalization excludes, and eq. (46) and Fig. 1's W_NS curve inherit it (reproduced W_NS = 1 − cos²(γtδ/(d² − δ²)) = 0.07775, printed 0.28683; the printed curve peaks at 6.949 s, the reproduced one at 13.90 s, both at the figure's own CODATA constants). Every R04 cell holds under both forms of (31), so the contradiction is **verdict-neutral**; the paper's claim that its two semi-classical members give distinct witness signatures survives the correction.
**The NSB cell:** under the paper's Born-rule statistics (its p10) the ensemble is exactly ρ₁ ⊗ ρ₂, the product of the single-mass marginals of the Newtonian quantum state; V_c = |cos(Δ/2)| for both masses and both partner outcomes, V_s = 1, Δ̂ = 0, **(N-b) REFUTES by E1's clause, thin:** 1 − cos(Δ/2) = 0.0122982745690 against REACH, ×1.2474; at E1's statistics that is 0.88σ per contour point and about 2.5σ at the scan level. **Conditions:** the paper's Born-rule statistics (p10) and quantum-equilibrium initial configuration (p7), its own p5 says Bohmian semi-classical positions are generally not Born-distributed, and a spin-readout pointer that is not the diamonds' own position. With a Stern–Gerlach re-split along the interferometer axis and a position readout, the paper's own p5 mechanism makes each partner's σ_z outcome its hold-time branch; that completion (the steelman's, outside the row's computation, no verdict) gives V_c = 1, Δ̂ = Δ, R = 1 and a two-setting witness of 1 + C − C² = 1.1319 from product wave functions, a local-hidden-variable statistic whose joint outcomes are not Born, a W6 (ii) failure that single-mass fringe statistics cannot detect. SAT-6's Bell test, or ⟨Y_A Y_B⟩ in the calibrated frame divided by V_s,A·V_s,B (1 for the quantum state against cos²(Δ/2) = 0.9756 for the completion, a gap of C² = 0.0244), separates it. **Record note for the owner (drafted, not entered; doc 01 is committed):** doc 01 §2.5 lists Struyve's separability theorem and witness curves (eqs. 44–47) among its unreproduced items; (44), (45) and (47) are now reproduced, (46) is contradicted by the factor 2 inherited from (31), and the separability theorem holds exactly at the BMV instance.

### 3.7 What a campaign's outcome licenses (30a §9, condensed; a restatement of the rows that carries no verdict of its own)

| outcome | certifies | refutes (on the account's own formulation, under the premises named) | never |
|---|---|---|---|
| certified positive (Definition 1, all three clauses; W1–W6) | non-classical in Definition 2's sense: Theorem W's disjunction | R03 with its seven comments; R04 (the NSB member for a non-positional spin readout); R06; R09; R11; R12 with its sub-entry; R13's non-entangling members; R15's effective CQ description; R23's classical-GW class (scope rider). Reclassified, not refuted: R05, R07, R10 | a verdict between readings (I) and (II); anything beyond the disjunction |
| the same, with E1-M's angle interval excluding πZ, plus M0–M4 | in addition: the coupling implemented a nonlocal non-Clifford gate (Theorem M) | as above | magic of the mediator; hardness |
| witness null at sufficient sensitivity with Δ̂ consistent with 0 (Theorem W′, linear phase additivity) | no second-difference phase in the probed regime | {quantized gravity ∧ linearized coupling ∧ QM for matter}; wherever resolvable R05's listed theories, R07's pair term, R08's derived positive (its contrapositive), R10, R13's non-local member, R18's pair term, R19, R20, R21 | a classicality claim |
| null with full visibility (R flat within r_R **and** each mass's V_s within u_s, both masses; −ln(1 − r_R) − ln(1 − u_s) ≤ 211/1944) | no forced classical-channel noise down to those reaches | in addition: R12 (through the single-mass scan); N-1's whole non-entangling class (×5.358 at equal reaches, hold-only); R04's NSB member (thin ×1.2474; conditional); R15's Markovian limit | a classicality claim |
| witness null with Δ̂ ≠ 0 and a deficit tracking Δ², or E1's signature 2 | a diagnostic only | nothing | a certificate of anything; a classicality claim |

**Two conditions a lab can drop by accident:** a flat pair/single ratio alone is not a null with full visibility (it needs the single-mass scan too), and a witness null alone is not Theorem W′'s null (it needs Δ̂ consistent with 0). **Is a null informative at all?** Under symmetric branch-diagonal noise the committed witness reads V_s·C + V_s² in the calibrated frame, so a null carries information only above V_s = 0.924876 plus the witness's margin; below V_s = 0.855799 the delivered state is separable at the committed geometry (R14's noise-dominated region on this platform). The full lab decision rule, every number certified, is 30a §9.

---

## 4. The answer at supported strength (what is corollary, computed, certified, transcribed, conditional)

- **Corollary (decided by the licensed theorems' definitions and the account's own text, typed before the run; never a finding):** every cell of the 22 DAF rows, and R04's (P), (N-a), (S) and NS cells. This is the table's bulk, as doc 29 §4.3 (1) said it would be: the duty is thin as discovery, and that thinness is its honesty.
- **Computed of record by one bounded reproduction:** R04's NSB (N-b) cell, REFUTES by E1's clause, thin (×1.2474), conditional on the paper's Born-rule statistics and a non-positional spin readout; and the KA-1 letter (the NS member's printed factor 2), verdict-neutral.
- **Decided by committed text and confirmed by an exact evaluation:** N-1 = INSTANCE (×5.358 at equal reaches, hold-only).
- **Certified from both sides by three disjoint routes:** M-1, F\* = 1/(1 + 4|sin Δ|) = 0.447343871026845, branch 1.
- **Committed magnitudes (near-corollaries registered so they are never findings):** SATURATED and the sizes SAT-2..SAT-6 at the committed geometry; SAT-1 in doc 27's parametrization with its committed-point value; v_DI.
- **Data only, carrying no verdict:** the position-readout completion of R04's NSB member (1.1319) and its discriminators; F\*_CSP (float); the dephasing hull boundary λ\* (LP-located); the Liénard–Wiechert 1/c⁴ toy; the phase-basis figures; R21's Yukawa shift of Δ (−5.3 % to −7.1 % at λ = 52 µm, transcribed from the steelman and the finalization); the generality of the filtered-CHSH figure (general complex filters found nothing lower, float grade).
- **Conditional, with the condition part of every sentence:** the positive column on W1–W6 and all three clauses of Definition 1 (W6 (i) is not certifiable, as in every Bell test; W6 (ii)'s single-mass certificate is necessary, not sufficient; W4's signatures are infeasible at BMV, which is why the conclusion is a disjunction); the null column on sufficient sensitivity, on Δ̂ consistent with 0 and on linear phase additivity; the full-visibility verdicts on both statistics for both masses with stated reaches; the NSB verdicts on the row's Born-rule statistics and a non-positional spin readout.
- **Prediction-grade only:** nothing.
- **What the table is not.** Not an exclusion claimed today, every "refutes" is what a future certified outcome would do to an account on its own formulation, under the premises named; a row left standing is not confirmed. Not a judgement of any paper or author; not a ranking of theories; not a statement that gravity has, or lacks, any property, a positive certifies Theorem W's disjunction only and a null never feeds a classicality claim.

---

## 5. Verification record

**The program's adversarial and second-path legs** (each with its own registrations frozen before its computations, its own trigger checks and its own pins; §3.1):
- **The gate** re-derived every row from full text (INSPIRE primary; own poppler route), re-typed every row by R1 / R2 exactly, re-derived N-1 symbolically and certified M-1 by an own census, facet and decomposition; ruled PASS with one verdict-moving defect (D1, the R04 type) and six others.
- **The lanes** re-read all 32 texts (identity 34/34 on INSPIRE and arXiv; 13 own arXiv fetches byte-identical to the pilot's and the gate's; currency; 85/90 quotes in the latest versions, replacements supplied), ran the zero-computed-content test at full-text grade, graded 104 hit records against the rows by registered categories, and found no kill.
- **The reproduction** proved every identity symbolically with the phases decomposed on independent atoms, certified every failure with its own interval evaluator at three physical points, showed its checker has teeth with a mutation control, and regressed the gate's logged numbers 7/7.
- **The rigor leg** re-derived every number by its own routes (31/31; R04 7/7), certified M-1 a third time, re-typed 13 rows from full text (13/13 P classes equal), re-found 72/72 quotations by a second extractor and 38/38 Appendix-A values in their sources.
- **The steelman** regressed every verdict quantity (14/14), confirmed N-1's exponent convention by its own derivation of the row's eq. (34), and attacked every row's reading, the map's sizes and N-1's and M-1's conventions from the bench.
- **The finalization** certified every newly printed number by its own exact arithmetic before entry (15/15), changed the note only through an asserted patcher that is the record of what changed, and closed with its checks clean on the final text and its pins 465/466 (only the note).

**This document's leg (R1 sections C–F; every forecast registered before the computation it predicts):**
- **Triggers** at the start (00:46:57–00:48:44): none fired; 0 differences from four earlier end states (§1.1). At the end: none fired (01:11:47–01:13:35); 0 differences from the start and from the four earlier end states; the controls unmoved (the check ran before the 02:00 CEST announcement).
- **Regression (NUMS, 00:56:39–00:57:20): 21/21 PASS**, N1–N21 by own exact arithmetic (Fractions; alternating Taylor series with explicit remainders for sin, cos and exp; Machin π; isqrt square roots; atanh logarithms; a certified sign change for the filtered-CHSH root): Δ = −211/672 by the closed form and by the four-branch sum; C, F_ent, F_magic, REACH; the N-1 chain (211/4860, 211/972, 211/1944, 211/3888; 0.0528232369, ×5.358; 0.075495, ×7.66; 0.102856); the two-reach bound 0.0939233 and 0.897144; the NSB deficit 0.0122982745690, ×1.2474 and the exact certificate 955128023467/391541969387520 with Δ² ≤ 48/5 and the limit 5/4; F\* = 0.447343871026845 with F\* < F_ent < F_magic; v_DI 0.98799697 > F_ent; τ_π = 25.0136 s; 0.667 ps and 2.33 ps; SAT-1's 3.3356410 × 10⁻¹³ of record and 8.38 × 10⁻¹⁴ at the committed point, identical to (|Δ|/τ)(d − Δx)/c; the dephasing thresholds 0.924876, 0.855799, 0.764027, 0.761789 and λ\* 0.5142 with λ\* < λ_PPT < λ_ent; the completion's 1.131905, 0.975555 and 0.0244; the filtered root 0.956106; F_STAB 0.981666, M₂ 0.104445, N_min = 1105 with the N = 1104 deficit exactly 17/276971520; the phase-basis figures and the 1.52 rad crossing; 0.88σ and 2.5σ. Every printed digit and every printed inequality regressed; no record number needed correcting.
- **The prime ledger (PRIMES, run of record run3, 00:57:58): D1 PASS**, the 22 primes named in §6 certified by sympy `isprime` and own trial division; 9,100,001 = 269 × 33829, 9,200,001 = 3 × 277 × 11071 and 9,400,001 = 13 × 317 × 2281 factored; 9,100,009, 9,200,017 and 9,400,009 are the first primes of their bands; 9,100,067 and 9,100,079 are the next primes after 9,100,061. **D2: 23 occurrences of integers in [9,300,000, 9,500,000) across 241 record files, 21 are listings of the reserve, the successor band or the pilot's named-elsewhere set, and 2 are the byte size of a PDF (9,357,466); no draw.**
- **The record (RECORD, run of record run2, 00:58:50):** nine leg manifests verified entry by entry, the gate's 22/22, the reproduction's 16/16, the lanes' 39/39, the rigor leg's 35/35, the steelman's 26/26, the finalization's 28/28, SCOPE-CERT's 19/19 and its verifier's 28/28; the assembly's 18/20, the two differences exactly the expected ones (its entry for the note is the v0 text, whose sha256 `3edeab69…` equals the preserved snapshot's; its log entry predates the log's last line). The pilot's five record files equal the shas the gate recorded at 20:38:08 (5/5). Every leg's pin table is unchanged between its before and after hashes (gate 209, lanes 282, reproduction 232, assembly 376, rigor 396, steelman 399, SCOPE-CERT 58 entries / 57 distinct paths and 78), except the finalization's (only the note, its object) and the pilot's (only README.md, changed externally at 20:24:03).
- **The document checks (DOC; CK-A..CK-H, R1 F):** run1 (01:10:39, on the text with its end-of-leg placeholders) was ALL CLEAN, CK-A 62/62 registered printed forms present, with the regression 21/21; CK-B 6/6 verbatim blocks found in their sources; CK-C no row-PDF quotation (every row quotation stays in 30a); CK-D 0 banned strings outside the ban list and the verbatim blocks, and the 8 sentences pairing "null" with "classical" and 2 pairing "positive" with "quantum" read by hand (each states the ban, describes a model or a diagnostic, or is a "never" cell); CK-E 23 row lines equal to 30a §4's (row, type, P class), 9 / 3 / 11 and 22 + 1, equal to the gate's tally; CK-F the registered section order, the five gates with statuses, the untouched reserve, Papers B and M; CK-G 116 section references resolve; CK-H every leg verdict word and pin count present. The check of record is the last DOC run in `analysis/adj_doc30.log`, on this final text (its sha256 is logged with it).
- **Pins:** 497 canon and record files hashed at 00:45:13 and again at 01:13:41, after the end-of-leg trigger check (01:13:41): 497/497 byte-identical, nothing changed (this document is not in the list; a final re-hash after its last write is logged as pins final); the 21 paper files (paper_A, paper_B, paper_C, paper_F, paper_M, paper_S) hashed at 00:51:07 and at the end (01:13:41): 21/21 byte-identical.
- **Dated corrections of this leg's own code (none touches a verdict, a source or a registered rule):** (1) PRIMES run1's D2 pattern matched digit runs inside hex digests (114 artefact items); the pattern now excludes alphanumeric neighbours; (2) the correction line for (1) was logged at 00:57:43, before its patch took effect (the first patch attempt failed on an assertion), so run2 repeated run1; run3 (00:57:58) is the run of record, and the stamps were corrected in the log; (3) RECORD run1 compared the pilot's nested pin tables at the wrong level and counted SCOPE-CERT's duplicated pin-list entry once; run2 is of record. The document checks needed no code correction.

---

## 6. Ledger

- **Bands and primes (every token certified by sympy `isprime` AND own trial division at its leg and re-certified by this leg; route-disjoint; nothing decided by a prime):** PILOT-ADJ **9,000,011 / 9,000,041**; SCOPE-CERT co-runner **9,000,049 / 9,000,059**; the gate **9,100,009 / 9,100,033**; SCOPE-CERT verifier **9,100,051 / 9,100,061**; the rigor leg of M-A6 **9,100,067 / 9,100,079** (the next primes after 9,100,061, inside the rigor / verifier band; doc 29 §5 pre-assigned only the steelman's M-A6 tokens, disclosed, not a breach); the steelman **9,200,017 / 9,200,021** (spares 9,200,027 / 9,200,029 certified, not drawn). The lanes, the reproduction, the assembly, the finalization and this leg drew none. Composites of record never drawn: 9,100,001 = 269 × 33829; 9,200,001 = 3 × 277 × 11071; 9,400,001 = 13 × 317 × 2281. **Closing-audit reserve ≥ 9,300,001 (prime; 9,300,007 / 9,300,017 / 9,300,023 certified): UNTOUCHED**, the designated consumer of G30-1, and of nothing else. **Successor band ≥ 9,400,001 (first prime 9,400,009; 9,400,019 / 9,400,021 / 9,400,043 certified): reserved fallow for the next program.** Every earlier reserve (≥ 6,600,001; ≥ 7,300,001; ≥ 7,700,071; ≥ 8,100,073; doc 27's audit reserve ≥ 8,550,001; doc 28's ≥ 8,900,029) is untouched: every token of this program lies in the ADJUDICATOR bands at or above 9,000,011, and the pilot's registered token discipline places doc 28's reserve below 9,000,010.
- **Canon pins:** every leg hashed its pins before and after (§3.1, §5): no committed document (`NN_*.md`, README, SUBMISSION_STATUS) was written by any leg of this program; no file under `paper_*/` changed during the program (every one carries an August 2026 mtime, `paper_B/` 2026-08-03, `paper_M/` 2026-08-28, and this leg's supplementary pins of all 21 paper files are compared at its end); README.md changed once during the pilot's end-of-leg trigger pass (20:24:03) by an external process and carries no verdict; the one pre-existing file the program changed is `30a_exclusion_table.md`, by the finalization, its assigned object (v0 preserved). This document is a new file.
- **Engines:** one committed engine was run, by the pilot only, as an isolated scratch copy with its sha256 verified before import and after the run (`witness_magic_lattice.py` `50a53ea9…`, for M-1's census); the co-runner M-A8 and its verifier ran committed engines (`closeduni_lattice.py`, `anchor_lattice.py`) the same way; every other leg used its own code only. The nine committed engine prefixes matched doc 29 §5 before and after the pilot.
- **Files of record** (sha256 prefixes): header of this document; each leg's `*_MANIFEST.sha256` (verified by this leg, §5). New records of this program: `analysis/adj_pilot*`, `adj_pilot_verifier*`, `adj_repro*`, `adj_note*` (assembly, rigor, steelman, finalization), `scope_cert*`, `papers/26_adjudicator/` (the lanes' report, manifest, 34 PDFs and scripts), `30a_exclusion_table.md`, and this document with `analysis/adj_doc30_*`.
- **Dated deviations and amendments (each disclosed in its own record):** the pilot's death mid-run and its resume on frozen registrations, with its script edited between stages (every stage's script sha256 in its state); the pilot's R3, the one definition introduced after a readout, which carries only the control's criterion (iii); the gate's D3 (six abstracts fetched before R2's freeze; no R2 rule depends on them); the reproduction assigned after the gate and run beside the lanes, before the assembly (doc 29 §5 lists lanes → assembly freeze → reproductions; the owner rules whether the order matters); the lanes' corpus path `papers/26_adjudicator/` against doc 29 §5's `papers/25_adjudicator/`; the lanes' reading-method deviation (19 entries at claim-sentence depth); the rigor leg's tokens not pre-assigned by doc 29; the steelman's two-reach condition u_s, introduced after readout as a lab-facing condition carrying no verdict; the SCOPE-CERT co-runner's pin list naming one file twice; this leg's supplementary paper pins (added after its main pins, before any line of this document) and its three code-level corrections (§5).
- **Process:** every leg ran on python3.12 and wrote new files only (the finalization also changed its assigned object, the note); no leg reports a background process left running; every leg after the pilot wrote a manifest (the pilot's five record files are pinned by the shas the gate recorded, re-verified here; the assembly's log postdates its own manifest line, disclosed by the rigor leg's N11); the owner folds README and memory.

---

## 7. Consequences

### 7.1 Pointer-row ruling (G30-4): **NONE ENTERED, the default holds; Papers B and M untouched; candidate rows drafted**

The material-change rule, as the house precedent quotes it (doc 27 §7.1, from doc 15 §6.1 + v1.1):
<!-- V: 27_anchor_margin.md -->
adding any pointer row would be a fresh registration decision, not an entitlement of this run
<!-- /V -->
No licensed sentence of Theorem W, Theorem W′ or Theorem M changed status in this program: every verdict of the table is a corollary of them, and the two computed cells are a scope confirmation of a committed clause (N-1) and a gate-typing number (M-1). The ruling per target:

- **Paper B (doc 06: Theorem W, W′, Corollary W2, the certification map), NO ROW, NO TEXT CHANGE.** Candidates drafted, NOT entered: **(B1)** the W6 (ii) certification-map cell (the finalization's candidate (B)): "single-mass fringe statistics (Born), necessary, not sufficient: a Bohmian semi-classical completion read out by position passes them exactly while its joint statistics are not Born; a Bell test is the only general discharge"; **(B2)** a scoreboard pointer beside Corollary W2: "[ADJUDICATOR, 30a §4] typed against the 2026 model set, a certified positive also refutes on their own formulations the Koopmanian commuting mediator, every commutative C\*-algebra mediator, the covariant CQ theory, the non-entangling members of the minimal-noise class of 2603.26075, the effective Markovian CQ description of 2604.06891 and a classical gravitational-wave background driver (scope rider), and Struyve's NSB member for a non-positional spin readout; it reclassifies without refuting the van Hove hybrid, Di Biagio's Diósi–Penrose and Aziz–Howl readings and Płodzień et al.'s pair term"; **(B3)** the lab-facing conditions of doc 06 §5's null statements: "full visibility" read as a flat pair/single ratio **and** a clean single-mass scan on both masses (the steelman's D1), and Theorem W′'s case read with Δ̂ consistent with 0 (21a §7's first sub-branch; the steelman's D5). P-A5 TRUE means doc 06 §5's E1 clause needs no scope rider for the minimal-noise class: at the committed geometry that class is an instance of it.
- **Paper M (doc 21, 21a: Theorem M, E1-M), NO ROW, NO TEXT CHANGE.** Candidate drafted, NOT entered: **(M1)** "[ADJUDICATOR, 30a §8] under the campaign's white-noise benchmark the noisy gravitational gate F·𝒰_Δ + (1 − F)·𝒟 at Δ = −211/672 lies outside the convex hull of the two-qubit Clifford unitary channels for every F > F\* = 1/(1 + 4|sin Δ|) = 0.447343871026845 (certified by three disjoint routes), below the witness threshold 0.864790: 21a's witness-first gating rule acquires a channel-level reason; a gate-typing number, never a law."
- **Doc 27 §3.9 (the Fork-1 corollary's printed size), doc 29 §5 (KA-2's letter; P-A7's parenthetical), doc 01 §2.5 (Struyve's unreproduced items)**, NO ROW; candidates in §7.5.
- **The observer line (docs 13, 25, 27, 28)**, not this program's: SCOPE-CERT's drafted riders go to the owner's post-2026-10-30 block (doc 27 §7.7), where they are ruled; nothing here enters or pre-empts them.

**Papers B and M are untouched:** no file under `paper_B/` or `paper_M/` changed (their mtimes are of 2026-08-03 and 2026-08-28; this leg's supplementary pins of all 21 paper files, before and after: 21/21 byte-identical at 00:51:07 and 01:13:41). Any future entry of a drafted row is a fresh registration decision of the owner; a row that would be quoted externally waits for G30-2.

### 7.2 Paper assessment

**The table's home is the in-house note, `30a_exclusion_table.md` Final v1.0 (the 05a / 21a pattern), and it is not posted by this program.** It is not a Paper B or Paper M amendment: its verdicts are corollaries of those papers' theorems plus the rows' own texts, and what those papers could receive is the drafted rows of §7.1, each a fresh decision at the paper's own gate process. **D1**, doc 29 §7's composition of Theorem W and W′, Theorem M, E1 and E1-M, the living exclusion table and the saturation map, now has its last leaf at in-house grade, through both adversarial legs; whether D1 is "something big" is the owner's call under defer-until-big and is not argued here. No export is gated open by this document: G30-1 and G30-2 precede any quotation.

### 7.3 Never-claimables (this program's, on top of the inherited stacks, doc 06 §5, 21a §7, doc 22 §6.7, doc 26 §4.1, doc 27 §7.3, doc 29 §7's last paragraph and 30a §12, all in force)

- **A positive** certifies non-classicality in Definition 2's sense and Theorem W's disjunction only, never a verdict between reading (I) and reading (II) (SAT-1..SAT-4), never "quantized GR" (which would consume SAT-2's infeasible timing or Corollary W1's cited argument, and must be flagged as doing so).
- **A null** never feeds a classicality claim; a witness null with Δ̂ ≠ 0 never carries Theorem W′'s verdicts; a flat pair/single ratio alone never carries the full-visibility verdicts.
- **"Reclassified, not refuted"** is the only wording for R05, R07 and R10; "consistent with a positive" is never "confirmed by a positive"; a row left standing is never "confirmed".
- **N-1 and the NSB verdict** are statements about what a future null would do, never exclusions claimed today; the NSB margin is always printed as thin (×1.2474) and conditional (Born-rule statistics; a non-positional spin readout), never as an order-of-magnitude exclusion.
- **ROW-FLIP** is never printed without "by its letter" and "verdict-neutral"; the flip is a printed factor 2 in the paper's NS member, and never "Struyve refuted" or "the paper is wrong about its no-go".
- **SAT-1's 3.3 × 10⁻¹³ rad** is never the size at the committed geometry (it is doc 27's parametrization; 8.38 × 10⁻¹⁴ rad at the committed point, a transient that vanishes 2.33 ps into the hold).
- **M-1** is a gate-typing number about a modelled channel under white noise: never a law, never a statement about the mediator, never a hardness claim, never a laboratory magic certificate (that needs Theorem M's M0–M4).
- **The ledger adjudicates assumptions, never authors or priority;** no priority or discovery sentence for any cell; published exclusion bounds are row data, cited and never re-derived; a flagged account carries no verdict until a dated re-registration types it.
- **The observer-line scope riders (SCOPE-CERT)** never travel with the table, and the table never travels with them.
- **Whether D1 is "big"** is never argued; licensing is never posting.

### 7.4 Gates

| gate | content | status |
|---|---|---|
| **G30-1** (closing audit at the untouched reserve ≥ 9,300,001; doc 29 §5's binding order: "a closing audit only if a quotation event is ever wanted") | a third-band re-execution through own-route second paths: the N-1 chain and the NSB deficit by a fresh exact route; M-1's certificate against a fourth census route; the saturation sizes; the typing of a pre-registered sample of rows re-derived from full text; a hand probe registered before any engine; must catch any record defect of the chain (the assembly manifest's path conventions, the pilot's README pin, the co-runner's duplicated pin entry) | **NOT SCHEDULED, owed only for a quotation event; gates G30-3.** Tokens 9,300,007 / 9,300,017 / 9,300,023 certified and reserved for it |
| **G30-2** (fresh re-sweep, incl. the post-2026-10-30 Perimeter leg) | the four triggers and the KA-3 ledger-shape nets with the Carney, Dakić, Bassi, Kent, Foo and Martín-Martínez–Perche feeds; L3's citers (the device-independent turn, the noise-floor program, the classical-carrier genre) against the lanes' baselines; the currency of all 32 row texts (a new version of any row is re-read and its cell re-derived under the frozen rules); the twelve flagged accounts and their successors; MOVW ref. [7]; **after 2026-10-30:** the Perimeter "Relativity Reframed" meeting, PIRSA C26010 recordings, the Indico 2058 timetable and contribution export, INSPIRE `cnum:C26-10-26.5`, every contribution graded against the 23 rows, the ledger shape and Theorems W / M (targets: entanglement, classical gravity, mediator, witness, BMV, GIE, Bell, LOCC, decoherence, noise, magic, non-Clifford); INSPIRE primary; it may run jointly with doc 27's G26-2 post-meeting leg | **OWED (calendar: after 2026-10-30); gates G30-3** |
| **G30-3** (external quotation of 30a or of this document) | whole-with-riders only (30a §12, §7.3 above, the ceiling riders); the steelman's export gate (its D1 and D5) is met in 30a's text | **BLOCKED until G30-1 and G30-2**; then a dated amendment of this document rules it conjunct by conjunct; licensing is never posting; defer-until-big |
| **G30-4** (pointer rows) | the material-change rule | **RULED THIS DOCUMENT: NONE ENTERED; Papers B and M untouched;** candidate rows (B1)–(B3), (M1) drafted, not entered (§7.1) |
| **G30-5** (standing) | MOVW ref. [7] (KA-6: its row enters first and its mediation criterion is typed against Lemma M before anything else; the ledger-shape occupancy is then re-ruled); the four triggers at every leg of any successor; M-A7's named sweeps; the per-export re-sweep inherited from the earlier programs | **STANDING** |

### 7.5 Record-side candidates (the owner's; drafted, NOT entered; none moves a verdict)

(A) Doc 27 §3.9 (i) and its M-L8 (i) line: SAT-1's "3.3 × 10⁻¹³ rad at BMV" carries its parametrization (Δφ ≈ 1 rad over 2 s, d = 200 µm; a supremum over time) and the committed-point value 8.38 × 10⁻¹⁴ rad, zero from 2.33 ps (the finalization's text, [FIN] §6 (A)). (B) Doc 06 §1.2's W6 (ii) certification row: "necessary, not sufficient" ([FIN] §6 (B); = (B1) of §7.1). (C) The gate's ruling 1 of its §7 ([GATE] §7): V_s as E1 defines it, from single-interferometer runs, not from the pair shots ([FIN] §6 (C); a dated erratum line for the owner's record, the gate's report never edited by another leg). (D) Doc 29's KA-2 and COUNTEREXAMPLE letter: "inside W1–W5" could name W1–W6, the theorem's premise set, or add that Definition 1's separable bound is sound under W6 (ii) ([FIN] §6 (D)). (E) Doc 29 §5 P-A7's parenthetical: "≤ 3.3 × 10⁻¹³ rad in doc 27's parametrization, 8.38 × 10⁻¹⁴ rad at the committed point" ([FIN] §6 (E)). (F) The assembly manifest's mixed path conventions, a note only; a closed leg's manifest is never rewritten ([FIN] §6 (F)). (G) The owner's fold of README and memory: 30a Final v1.0 and this document v1.0 ([FIN] §6 (G), extended). (H) SUBMISSION_STATUS's race-check line on Di Biagio 2511.02683: v5 exists (2026-09-16, retitled), and three of the four v4 sentences the house quoted are gone (the lanes' owner item). (I) The T1 watch definition carries [7]'s title identity, "Classical gravity cannot mediate entanglement", the comment's own title (the lanes' owner item). (J) Doc 29 §5's corpus path `papers/25_adjudicator/` against the lanes' `papers/26_adjudicator/` (the lanes' owner item). (K) Doc 01 §2.5's list of unreproduced Struyve items: (44), (45), (47) reproduced; (46) contradicted by the factor 2 of (31); the separability theorem exact at the BMV instance ([REPRO] §6). (L) Whether the reproduction's place in the sequence (after the gate, beside the lanes, before the assembly freeze) matters under doc 29 §5's binding order (disclosed by the reproduction and the note).

### 7.6 Next calculation

When policy calls, in order: **(1)** G30-2's post-2026-10-30 leg, the only calendar-bound work of this program, jointly with doc 27's G26-2 post-meeting leg if the owner so decides; **(2)** the owner's post-meeting block, which receives SCOPE-CERT's riders (doc 27 §7.7's G26-3 re-ruling and G26-4 decision), the observer line's business, not this program's; **(3)** the owner's decisions on the record-side candidates (A)–(L) and the drafted rows (B1)–(B3), (M1); **(4)** G30-1 only if a quotation event is wanted, then G30-3 by a dated amendment of this document; **(5)** the successor, which doc 29 §6 prices and does not promote: the second dimension of holography (runner-up 1) only after its own order-of-record pre-check, or a fresh multi-frontier survey if the owner wants discovery, the successor band ≥ 9,400,001 is reserved fallow for it; **(6)** MOVW ref. [7], the one event in this table's lane that could re-price Lemma M and hence Paper B, typed first the day it posts. The table itself needs no further computation: it is complete at DUTY grade, and a new account enters only as a flagged row or by a dated re-registration.

### 7.7 Doc 29 §7's honesty section, answered point by point

- **"Is this formulation-level?", No, as ruled.** The program claimed no discovery; its decided conjuncts were typed as corollaries before the run; the four-instance mechanism (a question decided at freeze while posing as open) could not arise and did not. The risk the judge named was thinness, and the record realized it: 22 of 23 rows decided by definitions, one reproduction, two small computed cells.
- **"The owner wants frontier discovery, and this delivers none by construction."** It delivered none. It used a bounded cycle (the pilot's first stage at 18:12:54 on 2026-09-24 to this document on 2026-09-25) with one bounded reproduction and no posting implication.
- **"Its value may sit unused."** The table now exists at in-house grade; its external use is gated by G30-1, G30-2 and the owner's defer-until-big, the judgment doc 29 said could not be made while D1's third leg was missing can now be made.
- **"It re-derives other groups' calculations."** Bounded to one row, and it paid in the other direction: the whole-section scope found a verdict-neutral factor-2 slip in the paper's own NS member while confirming the NSB chain exactly.
- **"It can read as advocacy in a live dispute."** The rows are typed by assumptions; R03's verdict is read off Aziz–Howl's own v3 text at the committed geometry, and the house's own flip of record (Fork 2) is marked contested and not load-bearing; no priority sentence anywhere.
- **"The ledger shape can be scooped."** KA-3 did not fire; the prior ledger (Di Biagio's Table II) is below the registered threshold and is cited and distinguished.
- **"MOVW ref [7] can land mid-run."** It did not, at any of the program's trigger checks; the watch stands (G30-5).
- **"M-1 can come out on the unglamorous side (F\* ≥ 0.975555, ~5 %)."** It could not, the pilot's registered near-corollary NC-M1 excluded that branch before computing, and it came out on branch 1, the reading in which the witness, not the gate's magic, is the bottleneck.
- **"After four stops, is the duty the honest pick?"** On the record, the duty delivered what it was picked for: the founding document's third deliverable, the living exclusion table, with its stated failure mode made precise as the saturation map, at DUTY grade, with every refutation reported first, both adversarial legs passed and nothing false entered. Whether that makes D1 big is the owner's call.

---

## 8. Provenance of this document

*Read order (all before this leg's registration R0): doc 29 in full (doc 29 §4 incl. the corrected-rider paragraph, doc 29 §5); doc 27 v1.2 as the format precedent (its structure; doc 27 §1, doc 27 §5 to doc 27 §7); the pilot's script (the R1 / R2 / R3 docstrings in full, the stage code in structure), results, rows, state and log; the gate's report; the lanes' report, manifest, README, registrations, grades, ZCC, L2 / L3 and trigger states, log, pins and manifest; the reproduction's report, registration (with its EQ block), script, state, log, trigger / freeze / pins / final scripts, pin lists and manifest; 30a Final v1.0 (whole); the rigor and steelman reports (whole); the finalization ledger; the SCOPE-CERT co-runner's and verifier's report headers and verdict sections (for §1.3, M-A8 only); `~/.claude/procedures/` (the gated-frontier-research-front entry and its note-assembly and finalization variants). **This leg, in order (2026-09-25, CEST):** pins before, 497 files, 00:45:13 → R0 frozen 00:45:59 (sha256 `50ce97ac610af833…`) → the first act, 00:46:57–00:48:44 (none fired; 0 differences) → supplementary paper pins, 21 files, 00:51:07 → R1's 13 verbatim blocks checked 00:53:07 and R1 frozen 00:53:29 (sha256 `6256d94d6aad1195…`) → the regression NUMS 00:56:39–00:57:20 (21/21) → PRIMES runs 1–3, 00:57:25–00:57:58 (run3 of record) → RECORD runs 1–2, 00:58:24–00:58:50 (run2 of record) → this document written in checkpointed chunks from about 01:00 to 01:10 → the document checks (run1 01:10:39 on the text with placeholders, ALL CLEAN; the check of record is the last DOC run in the log, on the final text) → the end-of-leg trigger check 01:11:47–01:13:35 → pins after and final, paper pins after → the manifest, written last. Records (all new, `analysis/`): `adj_doc30_registration_{R0,R1}.{txt,sha256,stamp}`; `adj_doc30_{pins,paperpins,freeze,vcheck,triggers,checks}.py`; `adj_doc30_pinlist.txt`; `adj_doc30_pins_{before,after,final}.{txt,stamp}`; `adj_doc30_pins_papers_{before,after}.{txt,stamp}`; `adj_doc30_trig_{start,end}.json`; `adj_doc30_state.json` (every check run); `adj_doc30.log` (append-only, with the dated corrections); `adj_doc30_MANIFEST.sha256`. No committed document, note, pilot, gate, lane, reproduction, assembly, M-A6, finalization or SCOPE-CERT file was edited; no engine was imported or run; no prime was drawn; no background process was launched; python3.12 only. The owner folds README and memory.*

**Wording (KA-5).** This document is scanned by code for the standing bans outside the block below and outside its marked verbatim quotations (CK-D); every sentence that pairs "null" with "classical" or "positive" with "quantum" was read by hand.

<!-- BANLIST -->
Banned (quoted only here): "gravity is quantum", "gravity is classical", "QM derived", "requires QM", bare "loophole-free", "the house settles the debate", "gravity is a magic factory", "undecidable"; a null feeding a classicality claim.
<!-- /BANLIST -->

**Status: ADJUDICATOR DECIDED (2026-09-25) at DUTY grade, outcome of record ROW-FLIP (by its letter; verdict-neutral) + SCOREBOARD-WITH-ROWS (R04, reproduced of record) + SATURATED; no Adjudicator trigger fired at any leg; KA-3 not fired; the table (30a Final v1.0) in-house and unposted. G30-1 owed only for a quotation event (reserve ≥ 9,300,001 UNTOUCHED); G30-2 owed after 2026-10-30; G30-3 BLOCKED; G30-4 NONE ENTERED, Papers B and M untouched.**