Red-team: the demotion-and-derivation posture and its fixed cores

Status: Red-team audit — central iteration-2 conclusion survives at its core; synthesis-page compression over-claims relative to its own source notes Last updated: 2026-06-08 Iteration: 3

Target

This page adversarially stress-tests the iteration-2 central conclusion: that the most defensible unified research program is one of demotion-and-derivation, and that its load-bearing fixed cores are local QFT Type $\mathrm{III}_1$ for any $\Lambda$; the horizon area law; no-signaling / causal-order; the Gleason–Busch-forced Born form; and unitarity.

The conclusion lives in two places that this audit treats as the attack surface:

• the synthesis note's registry item REG-SYNTH-1 and its Verdict (unified-program-synthesis.md, lines 79–81 and 99–111), and
• the FINDINGS iteration-2 update, which compresses the same posture into a single sentence (../FINDINGS.md, lines 96–97).

The audit's posture: be adversarial but fair. The wiki's deflationary core is the project's best work, and the goal here is not to demolish it but to find where the synthesis page over-compresses what its own better-hedged source notes establish. Each crack below carries its severity and the disposition it received from a second-pass referee, because in four of six cases the referee materially weakened or refuted the crack — and recording that is itself part of keeping the wiki honest. INFERENCE

A meta-result up front, established by re-checking every load-bearing 2025 citation: all three pivotal 2025 sources verify exactly as the wiki characterizes them — Giddings arXiv:2505.22708 (and sequel 2510.24833), Kudler-Flam–Leutheusser–Satishchandran Phys. Rev. D 111, 025013, and DESI DR2 arXiv:2503.14738. ESTABLISHED So the cracks below are about framing and propagation, not about misquoted physics.

Cracks found

Crack 1 — "Type $\mathrm{III}_1$ for any $\Lambda$" as a fixed core of a gravitational program

Severity claimed: serious. Referee disposition: kept but downgraded to MINOR — "internal contradiction" rejected, "clarity fix" retained. CONTESTED → clarity fix

The attack: listing local QFT Type $\mathrm{III}_1$ for any $\Lambda$ as an [ESTABLISHED] fixed core (REG-SYNTH-1 line 81; FINDINGS line 97) appears to contradict the wiki's own algebraic note, which — citing the verified Giddings results — holds that gravitational dressing makes spacelike-separated dressed observables generically fail to commute already at leading order in $G_N$, so that the $\mathrm{III}\to\mathrm{II}$ crossed product is a truncation of a more general structure (algebraic-background-independence.md, Arg-6, Obs-2, OP-CP1). The charge: the synthesis silently upgrades to ESTABLISHED-fixed an object the note tags OPEN/CONTESTED.

Why the referee weakened this. The crack conflates two distinct algebras:

• the undressed local net ($G_N = 0$), which is the unique hyperfinite Type $\mathrm{III}_1$ factor (Connes 1973/74; Buchholz–D'Antoni–Fredenhagen 1987; Haagerup 1987) — [ESTABLISHED] on every page, for all signs of $\Lambda$, and unaffected by Giddings, who takes the $\mathrm{III}_1$ net as input to his dressing;
• the dressed crossed-product Type II algebra ($G_N \neq 0$), whose nonperturbative survival is what the note actually tags OPEN/CONTESTED (Obs-2, OP-CP1) and what Giddings shows to be a leading-order special case.

So there is no cross-page status contradiction: both pages keep undressed-$\mathrm{III}_1$ as ESTABLISHED and the dressed-algebra fate as OPEN. INFERENCE, high The defining lines already carry the qualifier — synthesis line 19 reads "before gravitational dressing," and REG-SYNTH-2 line 85 repeats "before dressing." The genuine residue is a one-line copyedit: the bare fixed-cores parenthetical (REG-SYNTH-1 line 81; FINDINGS line 97) drops the "undressed / $G_N=0$" qualifier present two lines up, so read in isolation it could be misread as a claim about the gravitational theory. INFERENCE

The honest three-way grading the binary "fixed core" flattens:

• undressed local net is Type $\mathrm{III}_1$ for any $\Lambda$ — [ESTABLISHED];
• dressing to an observer/clock replaces it by a crossed-product Type II (purely algebraically, via Connes–Takesaki — no Killing field needed for the construction) — [ESTABLISHED, given a background];
• whether any von Neumann-algebraic subregion description survives nonperturbatively in $G_N$ — [OPEN/CONTESTED] (Giddings).

Verdict on the crack: the Giddings physics it cites is correct and ESTABLISHED; its inferential leap (that this impeaches "$\mathrm{III}_1$ for any $\Lambda$") fails because it attacks the wrong algebra. Action: append "(undressed, $G_N=0$)" to the two list-shorthands. Not a contradiction.

Crack 2 — the "fixed cores" list is rhetorically homogeneous but its members are not

Severity claimed: serious. Referee disposition: SOUND, kept at MINOR (presentation/propagation defect, not a substantive self-contradiction). INFERENCE

The attack: the five cores are not the same kind of object, and three are demonstrably not "fixed" in the strong sense the flat framing implies — facts the wiki establishes elsewhere but does not propagate into the headline. This crack the referee found sound on all three factual sub-claims:

• (a) Area law. Conceded as an input, not an output (synthesis line 23), yet derived in AdS by Ryu–Takayanagi and output from matter loops by Sakharov — so it is contingent/derivable in at least two frameworks, load-bearing specifically for the Jacobson–Padmanabhan–Verlinde lineage, per A-21 in ../ASSUMPTIONS_LEDGER.md. ESTABLISHED that it is derivable elsewhere
• (b) No-signaling. REG-SYNTH-3 already concedes it is not quantum-forcing: PR / Popescu–Rohrlich boxes are no-signaling-but-non-quantum, so it cannot do the singling-out work. It belongs with the GPT axiom set (local tomography + purification + reversibility), not on a par with Gleason/Busch. ESTABLISHED — Popescu–Rohrlich 1994
• (c) Unitarity. A-2 is explicitly "conditional on standard QM being final" (echoed at ../FINDINGS.md line 42) and is deliberately violated by GRW/CSL/Diósi–Penrose, which the wiki treats as empirically live (A-3-REF). A core that live competitor theories violate is conditional, not unconditionally fixed. ESTABLISHED — GRW 1986; CSL

So the set mixes one background-conditional theorem ($\mathrm{III}_1$), one forced functional form (Born), one GPT-insufficient shared constraint (no-signaling), one derivable-elsewhere quantity (area law), and one competitor-violated assumption (unitarity). The list's rhetorical force — "these are FIXED, everything else is suspected-derivative" — outruns what each member can bear.

Where the referee trimmed the crack. Two overclaims: (i) "demonstrably NOT fixed" overshoots — the accurate charge is heterogeneous in epistemic type and conditional in three different senses, not that the items fail to function as held-constant premises; the program holds the area law fixed precisely because its microscopic origin is unsupplied. (ii) "Homogeneous" is the crack's own gloss, not language in REG-SYNTH-1. The right reading of "fixed" is methodological (a held-constant working baseline), not ontological. INFERENCE

Verdict on the crack: a valid propagation gap — the iter-2 referee already corrected REG-SYNTH-2 (overclaim) and REG-SYNTH-3 (mis-scoping), but those caveats were never carried into the FINDINGS line-97 headline. Action: annotate each core in the headline with its status-type (background-conditional / form-only / GPT-insufficient-but-load-bearing / input-but-derivable-elsewhere / competitor-violated-conditional). Sharpens, does not refute, the program.

Crack 3 — "demotion-and-derivation" is mislabeled a falsifiable program

Severity claimed: serious. Referee disposition: SOUND on the terminological point, kept at MINOR; the crack's stronger "not physics" thesis rejected as naive falsificationism. INFERENCE

The attack: as a meta-strategy — "hold the cores fixed, prosecute the rest as suspected-derivative, each to a yes/no verdict" — the program forbids no observation and commits to nothing checkable. No observation or theorem could refute the program (as opposed to one of its object-level conjectures like OP-dS-FIRSTLAW). The genuinely falsifiable content is the bracketed [SPECULATIVE] wager (synthesis line 39: algebra/entanglement before geometry is true of our $\Lambda>0$ universe), which the program quarantines precisely because it is the part that could be wrong. The wiki half-sees this ("defensibility comes entirely from its scoping," line 37) but does not draw the consequence: a strategy whose defensibility comes from scoping is unfalsifiable by scoping.

The legitimate kernel. Predicating Popperian falsifiability of a set of methodological imperatives is a category error: imperatives are not truth-apt (Popper 1959). The correct kind-word is appraisable, not falsifiable. The falsifiable content lives entirely in (a) the seven decisive object-level deliverables (OP-dS-FIRSTLAW; the $m^2$-vs-$m^3$ BMV scaling HYP-GIE-DISCRIM; objective-collapse bounds A-3-REF; $w(z)\neq-1$ ASSUME-ETERNAL-dS) and (b) the bracketed wager. The structural observation is real: the line-37 "defensibility from scoping" is the immunization move that removes the program-level commitment a basic statement could contradict. INFERENCE

Why the referee rejected the strong thesis. The crack's conclusion — "only the bracketed wager is physics" — rests on naive single-hypothesis falsificationism, the position Lakatos (1970/1978) corrected. The program maps exactly onto a Lakatosian research programme: the fixed cores are the hard core; the line-37 scoping is the negative heuristic; "prosecute each suspicion to a verdict" is the positive heuristic; the seven deliverables are the protective belt. Such programmes are appraised as progressive vs degenerating, not as falsifiable/unfalsifiable — and they are checkable at the programme level via the wiki's own graduation criteria (synthesis lines 65–76): progressive iff a deliverable yields a genuine novel derivation of geometry (a $\Lambda>0$ entanglement first law delivering linearized Einstein; or emergent Killing/horizon structure from a non-symmetric modular flow, breaking the Bisognano–Wichmann/CHM circularity); degenerating iff every rigorous instance stays a re-encoding while every would-be derivation stays unachieved (the current state). [INFERENCE; the demarcation criterion the crack invokes is itself CONTESTED — Duhem–Quine, Kuhn, Feyerabend — which strengthens the conclusion, since Lakatos also denies programme-level falsifiability]

This also surfaces a live internal inconsistency the crack under-credits: FINDINGS line 97 already calls the same construct a "posture" / "sharp, coherent strategy ... not a unification" and pointedly never calls it falsifiable, while REG-SYNTH-1 line 81 does.

Verdict on the crack: the label is wrong, the "not physics" thesis is wrong. Action: strike "falsifiable program" from REG-SYNTH-1; substitute "an appraisable (progressive/degenerating) research strategy whose object-level conjectures and bracketed wager are independently falsifiable."

Crack 4 — the six-lens convergence is over-credited as independent

Severity claimed: serious. Referee disposition: the meta-claim (that the wiki over-credits) REFUTED; the technical input-audit retained as housekeeping. INFERENCE for the audit; REFUTED for the charge

The attack: reading H0/H1/H3/H4/H6/H7 convergence on "algebra/causal-order precedes geometry" as evidence for the wager tacitly assumes the lenses are independent, when they share load-bearing inputs:

• H0/H4 (modular/crossed-product) both run on Reeh–Schlieder + Tomita–Takesaki + a modular flow that is geometric only for Bisognano–Wichmann/CHM symmetric vacua (Arg-5, Obs-3) — i.e. they converge because they use the same Killing/conformal-symmetric AdS/large-$N$ inputs;
• H1/H3/H6 (causal-order) all rest on the continuum Malament/HKM theorem, which presupposes a distinguishing Lorentzian manifold; the discrete version (the causal-set Hauptvermutung) is unproven in all three;
• H7 (QECC) is an AdS/large-$N$ reconstruction.

So the "six independent lenses" are not independent — they share (i) AdS/large-$N$ tractability, (ii) symmetry-to-make-modular-flow-geometric, and (iii) the continuum Malament theorem. Convergence of methods sharing their load-bearing assumptions is not independent confirmation; it is a streetlight artifact — the rigorous results cluster exactly where a symmetry or solvable corner makes them tractable.

Why the referee refuted the meta-claim. The wiki does not claim the convergence is evidence-for. FINDINGS line 48 calls the hypotheses "research directions, not results" that merely "converge on one wager," and immediately adds the wager "is currently confined to the anti-de-Sitter corner and has no distinguishing experimental test — which is exactly why it is not yet physics." ESTABLISHED — verbatim in the wiki ../HYPOTHESES.md line 278 already states "These are NOT independent" for precisely the crack's reasons, and REG-SYNTH-5 reads the unity as a shared substrate ("rigorous reorganization ... not yet derivation"), the thesis, not a buried admission. The crack thus refutes a position the wiki does not hold, while re-deriving the de-crediting the wiki already performs. INFERENCE

Technical precisions the referee added (these survive as a genuine sharpening): the dependency graph factors into 2–3 correlated clusters, not one — causal-order (H1/H3/H6, with H6 partly independent via Bombelli–Henson–Sorkin Poisson sprinkling), modular-algebraic (H0/H4), code (H7); the continuum Malament/HKM theorem (ESTABLISHED, presupposes a manifold) must be separated from the discrete Hauptvermutung (UNPROVEN, obstructed by Kleitman–Rothschild dominance of non-manifoldlike posets); and "Killing-symmetric vacua" should read "a state with geometric (Killing or conformal) modular flow," since the crossed product itself is purely algebraic. So "six independent votes" rightly degrades to "two-to-three correlated clusters sharing a solvable corner" — a Reichenbach common-cause / screening-off discount that is partial, not total.

Verdict on the crack: the wiki does not commit the over-crediting error, so the headline charge is refuted; the input-sharing audit is correct and worth consolidating. Action: add an explicit independence-audit, and always gloss "convergence" as descriptive coherence, never as evidential support.

Crack 5 (cosmetic) — Born form as a "fixed core" while Program B interrogates the framework's finality

Severity claimed: cosmetic. ESTABLISHED form / mildly circular framing

The Born functional form $p=\mathrm{Tr}(\rho E)$ is "forced" only conditional on non-contextuality of probabilities (which Bohm explicitly rejects while still reproducing Born statistics) and, in the qubit ($\dim=2$) case, additivity over all compatible effects within a POVM (Busch; Caves–Fuchs–Manne–Renes — strictly stronger than Gleason's PVM additivity). So "Born form is a fixed core" is really "Born form is forced given Hilbert space + non-contextuality + effect-additivity" — the core is downstream of scaffolding that is itself part of the quantum framework whose finality Program B is supposed to interrogate. Listing it as a flat fixed core is mildly circular. Lower severity because the wiki does carefully tag the form-vs-meaning split (born-rule-and-quantum-linearity.md); the crack is only in the synthesis page's compression. INFERENCE

Citation note (a genuine catch from iteration 2, preserved): the dim-2 result is Caves–Fuchs–Manne–Renes (Found. Phys. 34, 193, 2004), not "-Schack" / "-Renner-Schack" — REG-SYNTH-1 line 17 still reads "Caves-Fuchs-Renner-Schack" and should be corrected. ESTABLISHED — citation correction

Crack 6 (cosmetic) — $\mathrm{II}_1$-vs-$\mathrm{II}_\infty$ billed as the headline $\Lambda>0$ foothold

Severity claimed: cosmetic — a persistent emphasis error, not a factual one. ESTABLISHED

The wiki already correctly downgraded the strong "algebra-type encodes $\Lambda$-sign" version to a "structural correlation" (REG-SYNTH-2). But even the surviving weak version is thinner than the Verdict's repeated billing ("CLPW Type $\mathrm{II}_1$ as the strongest $\Lambda>0$ foothold," line 105): $\mathrm{II}_1$ vs $\mathrm{II}_\infty$ is set by trace normalizability, which is a property of the horizon/observer Hamiltonian spectrum (whether the area is bounded above), not of $\Lambda$ per se. Decisively, Schwarzschild–de Sitter (a $\Lambda>0$ spacetime) gives a product of $\mathrm{II}_\infty$ on each horizon (KLS, verified) — so $\Lambda>0$ does not uniquely fix the type. The foothold is real as observer-relative entropy bookkeeping but supplies no dynamics (the de Sitter note concedes this). ESTABLISHED The $\mathrm{II}_1$ case is a feature of the static patch's bounded area, not of $\Lambda>0$ as such. See de-sitter-and-emergent-spacetime.md.

What survives the attack

The audit confirms that the deflationary core is intact and is the project's best work INFERENCE, high:

• The 17 GR/QM "contradictions" are domain-mismatches or unsolved-but-consistent problems, not logical inconsistencies. Correctly argued; the referee verdicts in ../GAPS_AND_CONTRADICTIONS.md are sound. ESTABLISHED as an internal verdict
• The single deepest tension — the algebraic/holographic program encodes geometry, it does not generate it — survives every attack and is strengthened by the verified Giddings results. The verdict "rigorous reorganization of semiclassical gravity on a fixed background, NOT yet a derivation of geometry" is correct and well-sourced (Witten 2308.03663 self-scoping; Bisognano–Wichmann/CHM circularity; KLS requires a Killing field). INFERENCE, high
• The Born form/meaning split survives: form forced by Gleason ($\dim\ge3$) + Busch/CFMR ($\dim=2$, at the cost of stronger additivity); quantitative-probability meaning OPEN in no-collapse settings; every Everettian derivation re-imports a Born-equivalent premise at the unequal-amplitude step (majority critical view, not a proved no-go). ESTABLISHED form / OPEN meaning
• Linearity-as-derivative-of-no-signaling survives as [INFERENCE, strong]: Gisin/Polchinski exclude generic deterministic nonlinearity via signalling; stochastic objective-collapse (GRW/CSL) evades it and is empirically live. The wiki's correction that Bielińska–Eckstein–Horodecki 2024 refines rather than reaffirms is accurate. INFERENCE, strong / CONTESTED that exact linearity is fundamental
• The "PARTIAL coherence — two programs joined by thin bridges" verdict survives, and the refusal of the "it's all information" fusion is correct and disciplined ("information" names non-interchangeable objects across clusters). INFERENCE
• The de Sitter negative meta-result survives: no dS RT surface, no uncontested unitary dual, no CLPW-rooted dS-first-law $\Rightarrow$ Einstein chain. The verified DESI DR2 numbers ($3.1\sigma$ for DESI+CMB; $2.8$–$4.2\sigma$ depending on the SNe sample) correctly support tagging eternal-dS as a modeling choice, not a fact — while themselves being parametrization/sample-dependent, not established new physics. CONTESTED — DESI DR2 arXiv:2503.14738
• BMV-as-class-falsifier-not-quantization-proof survives, with the General Witness Theorem locality assumptions correctly flagged as the soft spot. INFERENCE for the exclusions; loophole status CONTESTED

What must be weakened or re-tagged

Each fix propagates a qualification the wiki already establishes in a deeper note up into the compressed synthesis/headline:

1. Re-tag the fixed-cores list as a heterogeneous, per-item set, not a flat block. Header: "a small set of robust-but-non-homogeneous anchors, each conditional on a stated background or premise."

   • (i) $\mathrm{III}_1$ — [ESTABLISHED] for QFT on a fixed background (undressed, $G_N=0$), any $\Lambda$ sign; [OPEN/CONTESTED] as a core of the gravitational theory (Giddings).
   • (ii) Area law — [INFERENCE] not a universal input; derived in AdS (RT), output in Sakharov; load-bearing for the Jacobson–Padmanabhan–Verlinde lineage specifically (A-21).
   • (iii) No-signaling — [ESTABLISHED] shared constraint but not quantum-forcing (PR-boxes); demote to "one of several GPT axioms."
   • (iv) Born form — [ESTABLISHED] forced conditional on non-contextuality + effect-additivity.
   • (v) Unitarity — [INFERENCE, conditional] on standard QM being final; violated by live GRW/CSL.

2. Drop "falsifiable" as a descriptor of the program. Replace with "an appraisable (progressive/degenerating) research strategy; its falsifiable content lives entirely in the seven decisive deliverables and in the bracketed [SPECULATIVE] wager." INFERENCE

3. Weaken "independent convergence" to "convergence of methods sharing load-bearing inputs." Add the streetlight caveat: the lenses are tractable in, and tested only in, the regimes (AdS, Killing horizons, free fields, finite-dim toys) where they share inputs, so their agreement is weaker evidence than independent convergence — and "convergence" must always be glossed as descriptive coherence, never as confirmation. INFERENCE

4. Stop billing $\mathrm{II}_1$/$\mathrm{II}_\infty$ as $\Lambda$-diagnostic. Keep CLPW as a genuine $\Lambda>0$ entropy-bookkeeping result, but state it as "a fixed-background, observer-relative finite-entropy result, with the $\mathrm{II}_1$ case a feature of the static patch's bounded area, not of $\Lambda>0$ as such" (KLS: Schwarzschild–dS is $\mathrm{II}_\infty\times\mathrm{II}_\infty$). ESTABLISHED

5. Fix the citation: "Caves-Fuchs-Renner-Schack" → "Caves–Fuchs–Manne–Renes" at REG-SYNTH-1 line 17. ESTABLISHED

Proposed registry items (with referee verdicts)

GAP-FIXEDCORE-INCONSISTENCY — referee: do NOT keep as "inconsistency"; recast as MINOR copyedit

Proposed (crack 1): a live cross-page status mismatch on whether Type $\mathrm{III}_1$ is a fixed core once gravity is on. Referee disposition: the premise is false — the two pages assign the same graded status (undressed-$\mathrm{III}_1$ ESTABLISHED on both; dressed-algebra fate OPEN on both; Giddings presupposes $\mathrm{III}_1$ as input). The only warranted action is the one-line copyedit appending "(undressed, $G_N=0$)" to the bare list-shorthands. Tag of the proposed item: [REJECTED] as a contradiction; [INFERENCE] for the clarity fix. The Giddings facts remain [ESTABLISHED].

GAP-PROGRAM-UNFALSIFIABLE — referee: KEEP, MINOR

Statement: "demotion-and-derivation" is described as a "falsifiable program" (REG-SYNTH-1) but, as a meta-strategy, is a research heuristic/programme that forbids no observation; its falsifiable content lives in the seven object-level deliverables and the bracketed wager. The structural feature: "defensibility from scoping" (line 37) is the negative-heuristic immunization that makes the strategy non-falsifiable as a single hypothesis. Resolution: predicate "falsifiable" on the deliverables and the wager; describe the programme as appraisable (progressive/degenerating) per Lakatos. Caveat the referee adds: the programme is not epistemically unaccountable — it retains diachronic appraisability; it merely makes no single Popper-falsifiable commitment. Tag: [INFERENCE] (standard application of Popper 1959 + Lakatos 1970/1978; the demarcation criterion is itself [CONTESTED], which strengthens rather than weakens the point). Rationale: applies the wiki's own anti-crank standard (../AGENTS.md §4 — "a direction with no distinguishing test is not yet physics") reflexively to the meta-strategy, not only to object-level hypotheses.

ASSUME-INDEPENDENT-CONVERGENCE — referee: KEEP, MINOR (as a refinement of the existing convergence note)

Statement: reading H0/H1/H3/H4/H6/H7 convergence as evidence for the wager assumes the lenses are independent; they are not (modular cluster shares geometric-modular-flow + AdS/large-$N$; causal-order cluster shares the continuum Malament/HKM theorem + the unproven Hauptvermutung; H7 is AdS/large-$N$). A Reichenbach common-cause / screening-off streetlight discount applies. Referee precisions (which the item must absorb): the six lenses collapse to ~2–3 correlated clusters, not one artifact (H6 partly independent via Bombelli–Henson–Sorkin sprinkling); separate the continuum Malament/HKM (ESTABLISHED) from the discrete Hauptvermutung (UNPROVEN); "Killing-symmetric vacua" → "state with geometric (Killing or conformal) modular flow." So the discount is partial, not total. Tag: [INFERENCE]. Note: this makes explicit the shared-input dependency that REG-SYNTH-5 already concedes ("one substrate") but that the Verdict's "mutually reinforcing across all three tracks" phrasing risks reading as independent corroboration.

Verdict

The iteration-2 conclusion is substantially MORE honest than typical foundational synthesis and survives at its core. INFERENCE, high confidence

The deflationary "no logical inconsistencies, only domain-mismatches" result holds; the "encodes-not-generates geometry" verdict is correct and is strengthened, not weakened, by the verified Giddings results; the Born form/meaning split and linearity-as-derivative-of-no-signaling survive; the PARTIAL-coherence verdict and the refusal of the "it's all information" fusion are disciplined and right. All load-bearing 2025 citations (Giddings 2505.22708; KLS PRD 111, 025013; DESI DR2 2503.14738) verify exactly as characterized. ESTABLISHED for the citation checks

Critically, the wiki has already pre-empted the naive forms of every attack in this brief: it concedes the area law is an input not a universal core (A-21); concedes no-signaling does not single out QM (REG-SYNTH-3); concedes unitarity is conditional (A-2); and concedes the streetlight effect (H2, HYPOTHESES line 278). The genuine cracks are therefore at a higher level — synthesis-page compression — and are real but not fatal:

1. The bare $\mathrm{III}_1$ list-shorthand under-qualifies ("undressed, $G_N=0$" dropped) — a clarity hazard, not the cross-page contradiction crack 1 alleged; the referee downgraded it from "internal inconsistency" to a one-line copyedit. INFERENCE
2. The fixed-cores list is rhetorically flat but its five members carry radically different statuses, three of which the wiki itself shows elsewhere are not unconditionally fixed — a valid propagation gap from the better-hedged source notes into the FINDINGS headline. INFERENCE
3. "Demotion-and-derivation" is mislabeled "falsifiable" when it is an appraisable Lakatosian heuristic whose only falsifiable content is the part it brackets — a genuine terminological/category fix; the crack's stronger "not physics" thesis is rejected as naive falsificationism. INFERENCE
4. The six-lens convergence over-credit charge misfires as a charge — the wiki explicitly denies the convergence is confirmation and explicitly flags non-independence — but the underlying independence audit is correct and worth consolidating. INFERENCE for the audit; REFUTED for the charge

Net: the conclusion is sound; the synthesis page over-compresses and mildly over-claims relative to its own better-hedged source notes. The remedy is entirely propagation and re-tagging, not retraction — push the notes' qualifications up into REG-SYNTH-1 and the FINDINGS line-97 headline, append "(undressed, $G_N=0$)" to the $\mathrm{III}_1$ shorthand, demote "falsifiable program" to "appraisable research strategy," add the shared-input caveat to the convergence claim, stop billing $\mathrm{II}_1$ as $\Lambda$-diagnostic, and fix the Caves–Fuchs–Manne–Renes citation. Confidence in this audit: high. INFERENCE, high

See also

• Unified program synthesis — the target (REG-SYNTH-1; Verdict)
• Algebraic background independence — the source note crack 1 leans on (Arg-6, Obs-2, OP-CP1, A-CP-KILLING)
• Born rule and quantum linearity — form/meaning split; CFMR citation
• de Sitter and emergent spacetime — CLPW $\mathrm{II}_1$ as $\Lambda>0$ entropy bookkeeping
• Tabletop quantum gravity — BMV class-falsifier
• Entropic and emergent gravity — Sakharov/RT area-law derivability
• ../FINDINGS.md — iteration-2 update (lines 42, 48, 96–97)
• ../HYPOTHESES.md — H0/H1/H3/H4/H6/H7; line 278 non-independence note
• ../ASSUMPTIONS_LEDGER.md — A-2, A-3-REF, A-21, A-CP-KILLING
• ../GAPS_AND_CONTRADICTIONS.md — the deflationary 17-contradictions result
• ../EPISTEMICS.md — tag conventions; the cardinal rule
• ../AGENTS.md — §4 anti-crank standard
• ../OPEN_PROBLEMS.md — OP-CP1 and the dS-first-law deliverable
• ../domains/quantum-field-theory.md · ../domains/general-relativity.md · ../domains/information-theory.md