Hypotheses: Candidate Research Directions Toward Background-Independent Physics

Status: Living page — active, refereed, low-to-medium confidence throughout Last updated: 2026-06-10

DISCLAIMER — READ FIRST. Nothing on this page is a claimed result. These are research directions: structured conjectures, each built by synthesizing established mathematics and landmark results, then stress-tested by an internal referee. No new physics, no new theorem, and no new experimental number is asserted as established here. Where a sub-result is established, it is cited to its real source and tagged accordingly; the unifying readings that turn those sub-results into a "direction" are explicitly SPECULATIVE or, at most, INFERENCE. Several directions converge on the same wager — that causal/algebraic/entanglement structure is more fundamental than metric geometry — and several share the same fatal-if-unmet obstacles (Weinberg–Witten, the Type III $_1$ obstruction, AdS-confinement of the rigorous results). Each entry carries a REFEREE VERDICT recording errors, overclaims, the refined statement, and a severity grade. One sub-thesis is recorded as a negative/limiting result. See EPISTEMICS.md for the marker definitions and GAPS_AND_CONTRADICTIONS.md for the contradictions these directions target.

These directions are ordered by the lens that produced them: the algebraic/modular lens (H0, H4), the causal-order lens (H1, H3, H6), the holographic/information lens (H2, H7), the amplitudes/positive-geometry lens (H5), and the operational/post-quantum lens (H8).

Index

ID	Lens	One-line	Tag	Confidence	Severity
H0	Modular / algebraic	Linearized Einstein eq = stationarity of relative entropy	SPECULATIVE (AdS sub-results INFERENCE)	medium	minor
H1	Causal order	Entanglement fixes conformal metric; relational time fixes the scale	SPECULATIVE	low	minor
H2	Holographic (negative)	Entanglement-geometry is not yet a universal theory — graduation criteria	INFERENCE (negative meta-claim)	medium	minor
H3	Causal order / amplitudes	Causal structure primary; celestial symmetry as the boundary observable	SPECULATIVE	low	minor
H4	Modular / algebraic	Type III $_1$ + modular flow as primitive; crossed product gives time & entropy	SPECULATIVE (inputs ESTABLISHED)	medium	minor
H5	Amplitudes	Positive geometry as primitive; locality & unitarity derived	SPECULATIVE (N=4 core ESTABLISHED)	low	minor
H6	Causal order	Causal-set substrate: order + counting = geometry	SPECULATIVE	low	minor
H7	Holographic / information	QECC substrate; Type III QFT & bulk geometry as code limit	SPECULATIVE	low	minor
H8	Operational	QM as the rigid limit of an operational info theory; post-quantum room mapped	SPECULATIVE (scaffolding ESTABLISHED)	very low	minor

All eight passed the referee (keep=true) at minor severity. None is relegated to the rejected section, but H2 is itself a negative/limiting result and is flagged as such.

H0 — Modular / Type-III algebraic emergence: linearized Einstein from relative-entropy stationarity

Statement. Take the underlying ontology to be a Haag–Kastler-type net of von Neumann algebras (operationally: which operations are jointly performable) with states, and with no prior length metric beyond an assumed causal order, no manifold, and no tensor-factorized Hilbert space. The conjecture: an emergent semiclassical metric exists precisely for states whose modular (Tomita–Takesaki) flow is, in a coarse-grained large- $N$ limit, geometrically localizable; and the linearized Einstein equation is the statement that relative entropy is stationary at first order around such states (the entanglement "first law"). Geometry becomes a derived bookkeeping of modular Hamiltonians; the metric is not quantized because it was never a fundamental field. SPECULATIVE for the universal claim; the AdS sub-results are INFERENCE.

Motivation. Attacks the background-dependence vs background-independence and holographic area-law vs volume-extensive Hilbert space contradictions at their shared technical root: local QFT algebras are Type III $_1$ factors with no trace and no tensor factorization, so there is literally no local density matrix to "sew space from." The honest move is to make the Type III modular structure primary rather than treat it as an obstruction. See GAPS_AND_CONTRADICTIONS.md and UNIFYING_PRINCIPLES.md (entanglement-as-glue).

Mathematical sketch. For $(\mathcal{A},\Omega)$ cyclic-separating, Tomita–Takesaki gives $S = J\Delta^{1/2}$ , modular Hamiltonian $K=-\log\Delta$ , modular flow $\sigma_t(a)=\Delta^{it}a\,\Delta^{-it}$ . The Araki relative entropy

S(\omega\,\|\,\psi) = -\langle\Omega|\log\Delta_{\psi|\Omega}|\Omega\rangle

(built from the relative modular operator $\Delta_{\psi|\Omega}$ ) is finite and monotone even where individual entropies diverge. For a ball-shaped region with vacuum modular Hamiltonian $K_A = \int_A f(x)\,T_{00} + \text{const}$ , the entanglement first law $\delta S_A = \delta\langle K_A\rangle$ for all small regions is equivalent (Faulkner–Guica–Hartman–Myers–Van Raamsdonk) to

\delta G_{\mu\nu} = 8\pi G\,\delta T_{\mu\nu},

the linearized Einstein equation. The crossed product (Witten; Chandrasekaran–Longo–Penington–Witten) adjoins the modular-flow generator together with an observer/clock, promoting Type III $_1$ to Type II, which admits a (observer-relative) trace and a renormalized generalized entropy $S_{\text{gen}}=\langle K\rangle + \text{Area}/4G + \text{const}$ . Monotonicity of relative entropy yields the QNEC $2\pi\langle T_{vv}\rangle \ge S''_A$ (Ceyhan–Faulkner), playing the role of an averaged null energy condition.

What it would explain. Why perturbative metric quantization fails (the metric is a collective modular response, not a fundamental field); why holographic entropy is area-extensive (the finite quantity is relative entropy / renormalized $S_{\text{gen}}$ , not the bare divergent von Neumann entropy); the thermodynamic flavor of the linearized Einstein equation (Jacobson); the operational survival of no-signaling and a coarse causal order even when naive Reeh–Schlieder locality fails.

Predictions / tests. (1) Whether the crossed-product Type II construction extends beyond AdS/large- $N$ to a state-independent generalized-entropy functional in de Sitter and in dynamically formed asymptotically-flat black holes — failure marks the program AdS-specific. (2) Whether QNEC can be derived purely from modular monotonicity with no assumed background metric. (3) Whether the nonlinear Einstein equation follows from second-order relative-entropy (Fisher-information) data — persistent failure is evidence the lens is incomplete.

Strongest objections. (i) Weinberg–Witten is sharpest: the algebraic theory must not contain a Lorentz-covariant conserved stress tensor in the same spacetime where a massless spin-2 emerges; the only escape is the emergent-extra-dimension/holographic route. (ii) Everything rigorous is AdS/large- $N$ ; de Sitter (our $\Lambda>0$ universe) has no established dual. (iii) Only linearized equations are obtained. (iv) Modular Hamiltonians are geometric only for special regions/states (Rindler, CFT balls); generically the flow is non-local, so "geometry = modular localizability" risks being stipulative. (v) It explains the form of dynamics given thermodynamic inputs but does not identify the microscopic degrees of freedom.

No-go theorems to confront. Weinberg–Witten (survived only via emergent extra dimension); Reeh–Schlieder (used constructively, but forbids strict local factorization); Haag's theorem / Stone–von Neumann failure (the framework is natively algebraic, so this is structural, not a bug); singularity theorems (the emergent metric must degrade gracefully where modular localizability fails); Coleman–Mandula / Haag–Łopuszański–Sohnius (emergent spacetime symmetry must factorize from internal symmetry in the IR).

Relation to existing programs. A synthesis, claiming no novelty for: Tomita–Takesaki modular theory and Haag–Kastler AQFT; the FGHMVR linearized-Einstein-from-first-law derivation; Jacobson's thermodynamic and entanglement-equilibrium derivations; the Casini–Huerta relative-entropy/QNEC line; and the Witten / CLPW crossed-product Type II reformulation. See UNIFICATION_LANDSCAPE.md and domains/quantum-field-theory.md.

Epistemic tag: SPECULATIVE (universal claim) / INFERENCE (AdS sub-results). Confidence: medium.

REFEREE VERDICT (H0) — keep, severity: minor. Sound as a live research program; sub-results correctly attributed. Key corrections: (1) Geometricity of modular flow is established only for Rindler wedges (Bisognano–Wichmann) and CFT-vacuum balls (Casini–Huerta–Myers); for generic states it is non-geometric, so the emergence criterion is partly stipulative and potentially circular. (2) The ansatz $K_A=\int_A f(x)T_{00}+\text{const}$ already imports geometry — $T_{\mu\nu}$ is defined by variation w.r.t. $g_{\mu\nu}$ and $f(x)$ is the CHM conformal weight; FGHMVR relates two geometric descriptions, not geometry-from-a-metric-free substrate. (3) "No prior metric" is too strong: an abstract causal poset already fixes the conformal class of the metric (Malament / Hawking–King–McCarthy). Read it as "no prior length metric beyond the assumed causal order." (4) Notation: relative entropy uses the relative modular operator $\Delta_{\psi|\Omega}$ (Araki). (5) The crossed-product trace/entropy is observer-relative (requires adjoining a clock). Overclaims flagged: universal scope (every concrete result is asymptotically-AdS, large- $N$ , strong-coupling); "linearized Einstein is relative-entropy stationarity" holds only at first order in holographic CFTs; " $\Lambda$ -energy conditions $\Leftrightarrow$ monotonicity" is established only as monotonicity $\Rightarrow$ QNEC, not the biconditional. Recommended tag: SPECULATIVE (universal) + INFERENCE (AdS), equivalent to the author's "principled-speculation."

H1 — Causal order before geometry: entanglement fixes the conformal metric, a relational clock fixes the scale

Statement. Separate the two pieces of a Lorentzian metric that the entanglement program conflates. Causal/conformal structure (light cones, $g_{\mu\nu}$ up to a local scale) is more fundamental and robust than the conformal factor (local volumes/proper times). Conjecture: causal order is primitive (a discrete or algebraic partial order on events); entanglement/mutual-information structure reconstructs the conformal metric on the coarse-grained order; and the missing scalar conformal factor — equivalently local clock rate — is supplied by a relational/thermodynamic time (Page–Wootters correlation with a clock subsystem, or modular flow as an intrinsic clock). This makes time emergent and relational from the start. [SPECULATIVE.]

Motivation. Targets three contradictions jointly: the problem of time (Wheeler–DeWitt $H|\Psi\rangle=0$ , no external $t$ ); background dependence vs independence (a partial order is background-free); nonlocality vs relativistic locality (causal order may survive where spatial separability fails). The conformal/scale split is principled: a Lorentzian metric is (conformal class) $+$ (volume element), and a causal order already fixes the conformal metric in the continuum (Malament / Hawking–King–McCarthy). So entanglement need supply only one scalar.

Mathematical sketch. Primitive: a locally finite partial order $(C,\prec)$ (causal set), or the inclusion order of local algebras. Continuum recovery: a bijective causality-preserving map between distinguishing spacetimes is a conformal isometry, so causal order $\Rightarrow g_{\mu\nu}=\Omega^2(x)\,\bar g_{\mu\nu}$ with $\Omega$ undetermined. Discreteness fixes the scale via $V\sim N$ (number of elements) — "order + number = geometry." Relational time: Page–Wootters, $H_{\text{tot}}|\Psi\rangle\!\rangle=0$ with $H_{\text{tot}}=H_{\text{clock}}\otimes\mathbb{1}+\mathbb{1}\otimes H_{\text{sys}}$ ; the conditional state $|\psi(t)\rangle=\langle t|_{\text{clock}}|\Psi\rangle\!\rangle$ obeys $i\,d_t|\psi\rangle=H_{\text{sys}}|\psi\rangle$ with no external time. Modular-clock version: the Connes–Rovelli thermal time hypothesis takes the modular flow $\sigma_t$ of a KMS state as physical time. Entanglement enters via $I(A{:}B)$ as a regulator-independent proxy and a candidate distance $d(A,B)\sim-\log(\text{correlations})$ .

What it would explain. Why there is no external time (time is correlation, reducing to Schrödinger evolution via Page–Wootters); how background independence coexists with QFT's fixed causal structure (microcausality as IR shadow of the partial order); a heuristic for small positive $\Lambda$ (Sorkin's causet $\Lambda\sim 1/\sqrt{V}$ , a genuine if crude pre-observation order-of-magnitude estimate); Lorentz-invariant discreteness via Poisson sprinkling.

Predictions / tests. Causal-set discreteness predicts no generic frame-dependent dispersion — a confirmed energy-dependent photon time-of-flight from GRBs attributable to vacuum dispersion would falsify the Poisson-sprinkling version (Fermi-LAT bounds already strongly constrain linear-in- $E$ Lorentz violation). The everpresent- $\Lambda$ scenario predicts a fluctuating dark energy $\sim 1/\sqrt{V_{\text{past}}}$ ; evolving- $w(z)$ analyses are a live discriminator. Causet propagation predicts tiny momentum diffusion ("swerves"), bounded by null results. Self-consistency: the relational clock must reproduce unitary QFT in the appropriate limit.

Strongest objections. Causal-set dynamics is unsolved (no demonstrated 4D Lorentzian continuum limit); "entanglement fixes the conformal metric" is asserted by analogy to AdS, not derived, and $I(A{:}B)$ is UV-divergent in continuum QFT; thermal time is state-dependent, hard to reconcile with universal laboratory time; Page–Wootters faces the Kuchař criticism (two-time correlators, Pauli's no-time-operator theorem); no matter sector.

No-go theorems to confront. Pauli's theorem (evaded by making time relational, not an operator); singularity theorems (discreteness must be shown to resolve incompleteness); Bell / no-signaling (must reproduce Tsirelson correlations without conspiratorial superdeterminism); Borde–Guth–Vilenkin (a genuine beginning or well-defined replacement); empirical Lorentz-invariance bounds (near-theorem-strength).

Relation to existing programs. Malament/HKM theorems; causal set theory (Bombelli–Lee–Meyer–Sorkin), including "order + number = geometry" and Sorkin's everpresent- $\Lambda$ ; Page–Wootters relational time and Rovelli's partial observables; Connes–Rovelli thermal time. No novelty claimed; the candidate content is the division of labor (entanglement $\to$ conformal metric; relational/modular time $\to$ conformal factor). See UNIFICATION_LANDSCAPE.md, domains/cosmology.md.

Epistemic tag: SPECULATIVE. Confidence: low.

REFEREE VERDICT (H1) — keep, severity: minor. Internally consistent; conflicts with no result. Central defect: the load-bearing identification "conformal factor = local clock rate" is asserted, not derived, and equivocates between (i) a local geometric volume/proper-time scale $\Omega(x)$ on a 4-manifold and (ii) a single global PW time parameter or a single modular flow. Routes A (causets, where $V\sim N$ already fixes $\Omega$ combinatorially) and B (relational/modular time) are ALTERNATIVES, not complementary — route A makes route B redundant unless an explicit map from a global/modular time to a local $\Omega(x)$ is supplied. The construction tacitly assumes the unproven causal-set Hauptvermutung, and treats the continuum Malament theorem as if it bootstraps geometry from a bare order. The Connes–Rovelli flow is dimensionless and state-dependent, so using it to supply the scale risks circularity. Overclaim: "addresses the problem of time" should read "addresses one (kinematic) facet" — Dirac observables, the multiple-choice and global-time problems, and Kuchař's objections remain. Mutual information does not in general define a true metric. Recommended tag: SPECULATIVE.

H2 — A negative/limiting result: entanglement-geometry is not yet a supported route to a universal theory

This entry is itself a negative result. Recording where a lens fails to yield a supported direction is a legitimate output (see EPISTEMICS.md).

Statement. Judged as a candidate for a universal theory (not as a tool within AdS/CFT), "spacetime and gravity emergent from entanglement" currently fails to clear two identifiable bars and should be pursued as a constrained program, not asserted as a framework. Bar 1: all rigorous RT/HRT/QES and entanglement-wedge results, and the first-law derivation of gravity, hold only in asymptotically-AdS / large- $N$ / strong-coupling Einstein gravity, recover only the linearized Einstein equation, and use the wrong sign of $\Lambda$ for our accelerating universe, with no established de Sitter or asymptotically-flat analog. Bar 2: local QFT algebras are Type III $_1$ — no normal trace, no factorization $\mathcal{H}=\mathcal{H}_O\otimes\mathcal{H}_{O'}$ — so the slogan's "entanglement entropy of region $A$ " and "qubits in $A$ " are not literally well-defined without a cutoff; the naive subsystem picture is mathematically ill-defined (not realizable), though not logically inconsistent, since modular theory provides the rigorous reformulation. [The negative meta-conclusion is an INFERENCE from established items below; CONTESTED by proponents.]

Motivation. Several supplied contradictions point the same way: ER=EPR is tagged SPECULATIVE precisely because generality beyond AdS is unestablished; the holographic-vs-volume contradiction is OPEN with the Type III caveat; the entanglement-glue principle is SPECULATIVE with RT-in-AdS only at INFERENCE. Asserting a universal framework from these would violate epistemic discipline. The honest deliverable is a precise set of graduation criteria.

Mathematical sketch (what IS vs is NOT established). (a) RT/HRT $S(A)=\text{Area}(\gamma_A)/4G_N$ and linearized-Einstein-from-first-law: derived only in the large- $c$ , large- $\lambda$ asymptotically-AdS limit; linear order only. (b) Local algebras are Type III $_1$ (Buchholz–D'Antoni–Fredenhagen; Fredenhagen): no trace, $\mathcal{H}\neq\mathcal{H}_O\otimes\mathcal{H}_{O'}$ ; the crossed-product fix yields Type II after adjoining modular data, again mainly in AdS/large- $N$ and de Sitter static-patch settings. (c) ER=EPR has no first-principles derivation for generic (non-thermofield-double) entanglement — no theorem that a generic entangled state is a smooth geometric connection; support is limited to the eternal black hole (TFD) and engineered traversable wormholes. The gap is structural, not a missing computation.

What it would explain. Why decades of AdS results yielded no de Sitter / real-cosmology counterpart (tractability may be intrinsic to the negatively-curved corner); why the deepest puzzles ( $\Lambda$ sign/value, the arrow of time, dynamically-formed black holes) live exactly where the lens is least grounded; why the "qubits sewing space" intuition keeps hitting the Type III wall.

Falsifiable graduation criteria. (1) A de Sitter ( $\Lambda>0$ ) RT/QES analog with a controlled dual and derived entropy. (2) A state-independent generalized-entropy functional and entanglement-wedge reconstruction in a dynamically formed, asymptotically-flat black hole (not a toy model). (3) Derivation of the full nonlinear Einstein equation from an entanglement principle without a background metric. (4) A concrete operational test of ER=EPR for generic entanglement — none currently exists, which is itself diagnostic.

Strongest objections (to this negative framing). "Programs are judged by fertility, not current universality" — and by that standard it thrives (Page curve in controlled models, crossed-product reorganization of QFT algebras, unification of black-hole thermodynamics with entanglement). AdS may be a legitimate solvable laboratory whose lessons transfer. Linearized Einstein from entanglement is already nontrivial; partial nonlinear extensions exist. Risk: a negative framing may discourage a promising direction; the calibrated stance is "support the program, withhold the universal claim."

No-go theorems to confront. Weinberg–Witten (any emergent graviton must route through an emergent extra dimension; same-spacetime versions are dead); Reeh–Schlieder $+$ Type III $_1$ (make the local-qubit picture ill-posed); singularity theorems and Borde–Guth–Vilenkin (singularities unreplaced outside toy models); the persistent absence of a unitary, normalizable de Sitter dual (not a theorem, but the mathematical fact the universal claim must overcome).

Relation to existing programs. A faithful restatement of established status: Maldacena's AdS/CFT (a conjecture restricted to anti-de Sitter); Ryu–Takayanagi/HRT; Van Raamsdonk's disentangling argument; Buchholz–Fredenhagen Type III results; the recognized open status of de Sitter holography. The contribution is purely meta-level. See UNIFICATION_LANDSCAPE.md, OPEN_PROBLEMS.md, GAPS_AND_CONTRADICTIONS.md.

Epistemic tag: INFERENCE (the negative meta-conclusion), resting on ESTABLISHED structural facts plus a SPECULATIVE element (ER=EPR). Confidence: medium.

REFEREE VERDICT (H2) — keep, severity: minor. Substantively correct and well-sourced. Two fixes. (1) Bar 2 said "mathematically inconsistent"; downgrade to "the naive tensor-factor/qubit picture is ill-defined (not realizable) for Type III $_1$ algebras, expressed instead via modular theory and relative entropy." Nonexistence of factorization is a domain-of-validity issue, not a logical inconsistency — no contradiction is derivable, and the technical RT/first-law results never relied on Type-I factorization (the looseness lives in popularizations). (2) "All rigorous derivations are AdS-confined" — true for the RT/HRT/QES/entanglement-wedge machinery, but Jacobson-type horizon-thermodynamic derivations (1995 Clausius "equation of state"; 2015 entanglement equilibrium) are non-AdS, though heuristic and equally short of a universal theory; restrict the word "all" accordingly. Citations verified: RT (2006), HRT (2007), QES (Engelhardt–Wall 2014), RT derivation (Lewkowycz–Maldacena 2013), FGHMVR (2013/14, linear order only), ER=EPR (Maldacena–Susskind 2013), wrong-sign- $\Lambda$ obstruction. Recommended tag: INFERENCE (negative meta-claim) on ESTABLISHED scaffolding + one SPECULATIVE element; better classed as a well-supported inference than as speculation.

H3 — Causal structure primary; conformal + volume reconstruct the metric; celestial symmetry as the boundary observable

Statement. Take the partial causal order, not the metric, as the irreducible primitive. A Lorentzian metric factorizes as $g_{\mu\nu}=\Omega^2\tilde g_{\mu\nu}$ ( $\tilde g$ = light cones / causal order; $\Omega$ = scale/volume). Proposal: a fundamental theory specifies only the causal order plus a counting measure; the smooth metric, locality, and the Poincaré "particle" notion are emergent IR shadows. The organizing symmetry is then the asymptotic (BMS / celestial / Carrollian) symmetry on the boundary of causal diamonds, with the S-matrix (or its celestial-CFT avatar) as a candidate primary observable in the asymptotically-flat sector. SPECULATIVE program built on one ESTABLISHED theorem.

Motivation. Targets the background-dependence contradiction: QFT needs a fixed metric for Poincaré irreps, microcausality, and a vacuum; GR forbids prior geometry. Causal order is the one piece GR keeps and that needs no prior length scale. Microcausality is reframed as an emergent IR property. The causal-diamond boundary is exactly where Bousso's covariant bound and celestial data live.

Mathematical sketch. Metric split $g_{\mu\nu}=\Omega^n\tilde g_{\mu\nu}$ ( $\sqrt{-g}=\Omega^n\sqrt{-\tilde g}$ in $n$ dimensions). ESTABLISHED, HKM 1976 / Malament 1977 For distinguishing (or strongly causal) spacetimes, the causal order determines topology, differentiable structure, and conformal metric; a volume element then fixes $g_{\mu\nu}$ . Discrete sharpening (causal set, hedged, not endorsed): a locally finite poset with $N\sim V/\ell^4$ . Boundary observable: at null infinity the residual symmetry is BMS = supertranslations $\rtimes$ global Lorentz $SL(2,\mathbb{C})$ , possibly extended to superrotations/Virasoro CONTESTED; celestial amplitudes transform as 2D-conformal correlators with weights $h=\tfrac{1}{2}(1+s+i\lambda)$ . ESTABLISHED at proven orders, Strominger et al. Leading/subleading soft theorems equal the supertranslation/superrotation Ward identities, e.g. $\lim_{\omega\to0}\omega A_{n+1}=(\textstyle\sum_i\cdots)A_n$ .

What it would explain. Microcausality and a fixed light-cone in IR QFT as the emergent conformal causal order; the problem of time (primary observable is the boundary S-matrix, not bulk Schrödinger evolution); natural area-counting on diamond boundaries; Wigner's particle classification as the flat-space limit where the conformal causal structure degenerates to exact Poincaré.

Predictions / tests. Celestial-CFT consistency (associativity/crossing, celestial OPE) are sharp constraints flat-space gravity amplitudes must satisfy. A discreteness-scale imprint in the volume sector (causet $\Lambda\sim1/\sqrt{V}$ , order-of-magnitude only). Lorentz invariance must remain exact: a robust energy-dependent photon dispersion would disfavor the Lorentz-invariant-discreteness form.

Strongest objections. Weinberg–Witten (the boundary theory must lack a local covariant $T_{\mu\nu}$ — exactly where least proven); the metric = conformal + volume reconstruction is rigorous only for smooth distinguishing/globally-hyperbolic spacetimes, not off the smooth locus; celestial holography is perturbative and the 2D "CFT" is non-unitary (continuous principal-series weights, distributional correlators); specifying consistent quantum dynamics on a poset is unsolved.

No-go theorems to confront. Weinberg–Witten; Coleman–Mandula / Haag–Łopuszański–Sohnius (BMS evades by being asymptotic, not a symmetry of a nontrivial massive S-matrix); Reeh–Schlieder (vacuum entanglement across the diamond boundary); Penrose–Hawking singularity theorems; Haag's theorem / Stone–von Neumann failure (the "one celestial CFT" picture must confront inequivalent representations).

Relation to existing programs. The metric/causal split is classic (Weyl; Ehlers–Pirani–Schild; Malament 1977; HKM). Causal-set theory supplies "order + number = geometry." Celestial holography, BMS/asymptotic-symmetry/soft-theorem equivalences (Strominger and collaborators), and Carrollian field theory supply the boundary side. The only novelty is the framing. See UNIFICATION_LANDSCAPE.md, domains/particle-physics.md, domains/general-relativity.md.

Epistemic tag: SPECULATIVE. Confidence: low.

REFEREE VERDICT (H3) — keep, severity: minor. Internally consistent; conflicts with no result or no-go. Corrections. (1) Reconstruction theorem: attribute precisely — HKM (1976) (topology + differentiable + conformal structure under a causal-isomorphism hypothesis) and Malament (1977) (weakened to a pure order bijection); the distinguishing hypothesis is essential and must be stated. This is the one genuinely ESTABLISHED piece. (2) "Order + number = geometry" is the causal-set Hauptvermutung — UNPROVEN, complicated by Kleitman–Rothschild dominance of non-manifoldlike posets; label SPECULATIVE, not on Malament's footing. (3) Tensorial care: $\det\tilde g$ fixed is coordinate-dependent; the geometric object is the volume density $\Omega^n$ , not $\Omega^2$ . (4) Distinguish BMS (global Lorentz) from extended/generalized BMS (superrotations/Virasoro — CONTESTED). (5) Celestial CFT is non-unitary (Mellin-transform continuous/complex weights, no standard OPE convergence); a complete celestial CFT for 4D gravity is not constructed. (6) "Crossing + Ward identities fix correlators with no bulk time" overstates: asymptotic symmetries fix only the soft/universal sector, not the hard dynamics. The S-matrix is privileged only under asymptotically-flat boundary conditions and is IR-subtle (Faddeev–Kulish/KLN); it is not defined in de Sitter/cosmology. Recommended tag: SPECULATIVE.

H4 — Operator-algebraic background independence: Type III $_1$ + modular flow as primitive geometry; crossed product gives time, entropy, area law

Statement. Make the net of observable algebras and their Tomita–Takesaki modular structure the primary object, prior to any metric. A "region" is a von Neumann algebra $\mathcal{A}$ (generically Type III $_1$ in QFT), not a set of points; the modular automorphism group $\sigma_t^\omega$ of a state $\omega$ supplies an intrinsic, state-dependent flow that can serve as emergent time. Coupling to gravity (an observer/constraint) implements the crossed product $\mathcal{A}\rtimes_\sigma\mathbb{R}$ , converting Type III $_1$ into Type II, restoring a trace and a renormalized entropy whose leading term is the generalized/area entropy. SPECULATIVE for the emergent-geometry thesis; inputs ESTABLISHED.

Motivation. Meets three contradictions with established mathematics: the problem of time (Tomita–Takesaki gives a canonical flow with no external clock — Connes–Rovelli thermal time); the holographic/area-law vs volume tension and the Type III obstruction (the crossed-product yields Type II with a trace and a finite, area-leading entropy); Haag's theorem / Stone–von Neumann failure are the central structural facts, not pathologies. Locality survives only as the net structure; subsystem/qubit pictures are explicitly denied.

Mathematical sketch. ESTABLISHED Tomita–Takesaki: $Sa|\Omega\rangle=a^\dagger|\Omega\rangle$ , $S=J\Delta^{1/2}$ , $\Delta^{it}\mathcal{A}\Delta^{-it}=\mathcal{A}$ , $J\mathcal{A}J=\mathcal{A}'$ . Connes: local QFT algebras are Type III $_1$ (no trace, no minimal projections, no density matrix). KMS: $\omega(a\,\sigma_t(b))=\omega(b\,\sigma_{t+i\beta}(a))$ — the modular flow is KMS with respect to its own generator (convention $\beta=1$ ); the physical Unruh temperature of the Rindler/boost state is $\beta=2\pi$ (Bisognano–Wichmann). INFERENCE, recent Crossed product: adjoin a bounded-below observer Hamiltonian (clock) and impose the modular constraint; the result is Type II $_\infty$ (black hole) or II $_1$ (de Sitter static patch), admitting a semifinite trace and a renormalized $S(\rho)=-\text{Tr}(\rho\ln\rho)=\langle\text{Area}\rangle/4G+S_{\text{out}}+\cdots$ . For a Rindler wedge, $K=2\pi\int_{\text{wedge}}\xi^\mu T_{\mu\nu}\,d\Sigma^\nu$ (boost generator), and $\delta\langle K\rangle=\delta S_{\text{ent}}$ gives the linearized Einstein equations.

What it would explain. The problem of time (modular/thermal time intrinsic to (algebra, state)); why naive QFT over-counts degrees of freedom (the true objects are Type III, only the Type II renormalized entropy is physical and area-leading); Unruh/Hawking thermality as a theorem about $\Delta^{it}$ ; black-hole $S=A/4G+S_{\text{out}}$ as a Type II trace obeying a second law.

Predictions / tests. Primarily structural: the crossed-product entropy must reproduce known generalized-entropy results across de Sitter, black-hole, and Rindler horizons — failure to extend bounds its scope. Predicts that any consistent quantum-gravity entropy is the trace of a Type II algebra (entropy differences UV-finite, absolute entropy not — matching the physical fact that only differences are observable). Modular/thermal time should agree with proper time along stationary observers in KMS states.

Strongest objections. Crossed-product results are recent and shown mainly in semiclassical / large- $N$ / AdS / de Sitter static-patch settings; full background-independent dynamical asymptotically-flat cosmology is OPEN. Thermal time is state-dependent — recovering a single shared classical time risks circularity (must pick the equilibrium state). Modular Hamiltonians are geometric only for special regions. This lens makes geometry emergent from algebra+entanglement, partly self-undermining within a "geometry-primary" framing — it is honestly an "algebra/information-primary" direction.

No-go theorems to confront. Reeh–Schlieder (the enabling cyclic-separating hypothesis, but it kills strict locality); Haag's theorem / Stone–von Neumann failure (the whole point); Weinberg–Witten (the emergent linearized Einstein eq must not be misread as a composite graviton with a covariant boundary $T_{\mu\nu}$ ); Bekenstein/Bousso bounds (the Type II trace-entropy must respect area bounds); Coleman–Mandula (modular flow is not an S-matrix spacetime symmetry).

Relation to existing programs. Built on ESTABLISHED Tomita–Takesaki and Connes classification, Haag–Kastler AQFT, Bisognano–Wichmann; plus Connes–Rovelli thermal time (SPECULATIVE but principled) and the recent (2021–2024) Witten / Chandrasekaran–Longo–Penington–Witten crossed-product Type II generalized entropy (INFERENCE, young), and the FGHMVR / Jacobson first-law $\to$ linearized Einstein. No novelty for the ingredients. See domains/quantum-field-theory.md, domains/information-theory.md, UNIFICATION_LANDSCAPE.md. Closely allied with H0.

Epistemic tag: SPECULATIVE (emergent-geometry thesis) / ESTABLISHED (modular-theory inputs) / INFERENCE (crossed-product entropy). Confidence: medium.

REFEREE VERDICT (H4) — keep, severity: minor. The author's original sketch was marked isSound=false for mislabeling the inferential bridge as nearly established; the refined statement above repairs this and is kept. Load-bearing corrections, now incorporated: (1) " $\beta=1$ KMS" is a convention of the abstract modular flow, not the physical Unruh temperature ( $\beta=2\pi$ w.r.t. boost energy) — the original conflated them. (2) Modular flow is geometric only for special (algebra, state) pairs (Bisognano–Wichmann wedge; Hislop–Longo CFT double-cone); generically non-geometric, so geometry is recognized a posteriori in symmetric cases, not derived in general. (3) The crossed product adjoins a bounded-below observer Hamiltonian, not "the modular Hamiltonian" (which is unbounded both ways and not affiliated to the Type III $_1$ algebra); the trace/entropy are observer-relative, perturbative in $G$ , and presuppose a fixed AdS or dS background — the program encodes/reconstructs geometry, it does not yet generate spacetime. (4) First-law $\Rightarrow$ Einstein is linearized and holographic. Headline overclaim curbed: "geometry and a preferred time DERIVED from algebra + state" is not delivered background-independently. Recommended tag: SPECULATIVE (synthesis) / ESTABLISHED (inputs) / INFERENCE (recent crossed-product result).

H5 — Positive geometry as primitive: spacetime, locality, unitarity as derived boundary properties

Statement. Posit that the fundamental object generating observables is a positive geometry — a region with boundaries equipped with a canonical differential form having logarithmic singularities exactly on those boundaries — and that physical amplitudes are the canonical form. Spacetime locality and unitarity are then emergent: factorization/positivity properties of the geometry's boundaries, not input axioms. The amplituhedron for planar $\mathcal{N}=4$ super-Yang–Mills is the proof-of-concept; the speculative extension is that gravity and the realistic Standard-Model S-matrix likewise arise from (associahedron-/cosmological-polytope-like) positive geometries. [SPECULATIVE, with an ESTABLISHED-within-planar- $\mathcal{N}=4$ core.]

Motivation. Engages the "scattering amplitudes / positive geometry" clause and the principle that the continuum and locality may be derived, not assumed. In the amplituhedron, BCFW recursion and even unitarity emerge from combinatorics, not from a Lagrangian with built-in microcausality. If the S-matrix data come from a geometry with no reference to a bulk metric, a fixed background is not a prerequisite for defining observables.

Mathematical sketch. [ESTABLISHED, within $\mathcal{N}=4$ SYM] A positive geometry $(X,X_{\ge0})$ has a unique canonical form $\Omega$ with $d\log$ singularities on all boundary components: for $[a,b]$ , $\Omega=dx\big(\tfrac{1}{x-a}-\tfrac{1}{x-b}\big)$ ; for the ABHY associahedron $A_{n-3}$ , $\Omega$ yields the tree-level bi-adjoint $\phi^3$ amplitude $m(\alpha|\alpha)$ for the corresponding ordering. The amplituhedron $A_{n,k,L}$ (image of $G_{\ge0}(k,n)$ under a $Z$ -map into $G(k,k+4)$ at tree level, with an extended loop-level geometry) has canonical form equal to the planar $\mathcal{N}=4$ SYM loop integrand. Locality: physical poles = codimension-1 boundaries; factorization $A\to A_L\times A_R$ = a boundary being a product of smaller positive geometries. Unitarity is encoded separately, via the $d\log$ /logarithmic-singularity structure (no double poles; leading singularities $\pm1$ ) and, at loop level, forward-limit/positivity properties. Cosmological polytope (Arkani-Hamed–Benincasa–Postnikov): wavefunction coefficients for conformally-coupled scalars with polynomial couplings in FRW/de-Sitter-limit backgrounds.

What it would explain. Why amplitudes are local (poles = boundaries) and unitarity-respecting ( $d\log$ /forward-limit structure) without those being axioms; the hidden simplicity (no spurious poles, all-loop integrands from one object) invisible in Feynman diagrams; a possible origin for cosmological correlators bypassing bulk time evolution; positivity-constrained EFT coefficients linking to the bootstrap.

Predictions / tests. Sharp integrand predictions (sign/positivity, singularity structure) checkable against independent unitarity computations — agreement is ESTABLISHED for many $\mathcal{N}=4$ cases; a discrepancy would falsify. Cosmological-polytope predictions for total-energy/partial-energy singularities of inflationary correlators are testable against in-in QFT. If extendable to gravity, it must reproduce graviton amplitudes via KLT/double-copy (ESTABLISHED); failure to embed the double copy would bound the program.

Strongest objections. Established only for planar, maximally supersymmetric $\mathcal{N}=4$ SYM — not gravity (whose amplitudes lack the simple $d\log$ /pole-positivity structure), not non-planar, not massive non-SUSY matter, not the Higgs/confinement sectors. It computes the integrand; loop integration (IR/UV, regularization, transcendental functions) is outside the construction. Even in the proof-of-concept, momentum-twistor kinematics presuppose 4D conformal spacetime and planarity — so "spacetime emergence" is partial; the background is smuggled into the kinematics.

No-go theorems to confront. Coleman–Mandula / HLS ( $\mathcal{N}=4$ saturates maximal SUSY; a real-world theory must respect the limits); Weinberg–Witten (a positive-geometry gravity theory must evade the massless-spin-2 stress-tensor obstruction — the double-copy origin is the likely route); Froissart-type unitarity/analyticity bounds (must hold after integration); Haag's theorem (sidestepped by working with on-shell integrand data — but the nonperturbative object the geometry computes must then be defined).

Relation to existing programs. The amplituhedron / positive-geometry program (Arkani-Hamed–Trnka 2013; associahedron/ABHY; cosmological polytope; curve-integral/surfacehedron work) is ESTABLISHED as computational technology and a reformulation of planar $\mathcal{N}=4$ SYM. Twistor theory (Penrose), BCFW recursion, generalized unitarity, and KLT/double copy are the backdrop. No novelty claimed beyond framing against the contradictions. See domains/particle-physics.md, domains/mathematics.md, UNIFICATION_LANDSCAPE.md.

Epistemic tag: SPECULATIVE (gravity/real-SM extension OPEN; integrand-level $\mathcal{N}=4$ core ESTABLISHED). Confidence: low.

REFEREE VERDICT (H5) — keep, severity: minor. Internally consistent; conflicts with no no-go; the ESTABLISHED-within- $\mathcal{N}=4$ core is genuine. Scope/attribution fixes (incorporated above). (1) The associahedron canonical form computes $m(\alpha|\alpha)$ for a single ordering; the full bi-adjoint $m(\alpha|\beta)$ matrix needs associahedron intersections — do not imply one polytope gives all. (2) The amplituhedron geometrizes the loop integrand, not the integrated amplitude. (3) "Linear map into $G(k,k+4)$ " is exact at tree level only; the loop geometry is more elaborate. (4) Conceptual: locality (boundary/factorization) and unitarity ( $d\log$ /forward-limit positivity) are encoded by different geometric features — do not collapse unitarity into "factorization of boundaries." Overclaims curbed: "spacetime emergent" is partial even in $\mathcal{N}=4$ (momentum twistors presuppose 4D conformal spacetime + planarity); the gravity/real-SM extension is highly speculative (no positive geometry for gravity, non-planar, or massive non-SUSY matter; associahedron handles only massless colored $\phi^3$ ); cosmological polytope rigorously covers conformally-coupled scalars with polynomial couplings in FRW, with de Sitter in specific limits. Recommended tag: SPECULATIVE (ESTABLISHED planar- $\mathcal{N}=4$ integrand core; OPEN gravity/SM extension).

H6 — Causal-set substrate: order + counting as primitive, continuum geometry as coarse-grained limit

Statement. SPECULATIVE The fundamental description of spacetime is a locally finite partially ordered set (causal set) $C=(S,\preceq)$ : elements are discrete spacetime atoms, $\preceq$ is the microscopic causal relation, and the only other datum is counting (cardinality). A smooth globally hyperbolic $(M,g)$ emerges only as a coarse-grained approximation of a causet well-approximated by it via Poisson sprinkling — "order + number $\approx$ geometry" (causal order $\to$ conformal/light-cone structure, atom-counting $\to$ scale). Dynamics is a quantum sum-over-causets (a discrete path integral / quantum sequential growth), with the gravitational action replaced by a discrete functional (Benincasa–Dowker).

Motivation. Most directly motivated by four contradictions jointly: (i) background dependence vs independence (no prior metric — causal order is primitive); (ii) holographic/area-law vs volume and the discreteness signaled by $S_{\text{BH}}=A/4\ell_P^2$ (a discrete substrate gives a finite, countable d.o.f. budget); (iii) singularity theorems (curvature singularities as artifacts of pushing the continuum past $\sim\ell_P$ ); (iv) the cosmological-constant problem ( $\Delta N\sim\sqrt N\Rightarrow\Delta\Lambda\sim1/\sqrt V$ in Planck units, a parametrically small fluctuating $\Lambda$ of roughly the right order — the program's one quantitative success and its most contested claim). The deep-principle anchor: causality may be more robust than locality — causal sets keep causal order and drop the continuum.

Mathematical sketch. $C=(S,\preceq)$ : reflexive, antisymmetric, transitive, locally finite ( $|I(x,y)|=|\{z:x\preceq z\preceq y\}|<\infty$ ). Continuum recovery: Poisson sprinkling, $P(n\text{ in }V)=(\rho V)^n e^{-\rho V}/n!$ , $\rho=\ell_P^{-d}$ . KEY Lorentz-invariance theorem ESTABLISHED, Bombelli–Henson–Sorkin: no Lorentz-equivariant measurable rule selects a preferred direction from a Minkowski sprinkling — statistically Lorentz invariant, unlike a regular lattice. Metric recovery: timelike distance $\approx$ longest-chain length; Myrheim–Meyer dimension from chain/antichain counts. Discrete curvature: the Benincasa–Dowker operator (a sum over small inclusive-past layers with $d$ -dependent alternating coefficients) $\to(\Box-R/2)f$ in the continuum mean; the action reduces to $\sim\ell^{-(d-2)}\!\int\!\sqrt{-g}\,R$ . Dynamics: Classical Sequential Growth (Rideout–Sorkin), with growth probabilities fixed by discrete general covariance + Bell causality; the quantum sum $Z=\sum_C e^{iS_{\text{BD}}/\hbar}$ is the central OPEN problem. Matter: free scalar via the discrete retarded Green function; the Sorkin–Johnston construction yields a distinguished state from the causal Green operator with no preferred time (but can fail to be Hadamard).

What it would explain. Why GR resists fundamental quantization (the metric is emergent/coarse-grained); how Lorentz invariance survives discreteness without a preferred frame (the sprinkling theorem — the obstacle that defeats lattice/digital-physics accounts); finite area-scaling d.o.f. behind black-hole entropy; singularities as continuum breakdown; an everpresent, fluctuating, small $\Lambda$ of roughly the observed order (heuristic, contested).

Predictions / tests. Lorentz invariance preserved on average $\Rightarrow$ absence of the linear-in- $E/E_{\text{Pl}}$ photon dispersion many discrete approaches predict; tight Fermi-LAT GRB-timing and polarization bounds thus confirm rather than refute this approach, distinguishing it from naive lattices. An everpresent $\Lambda$ fluctuating $\sim\pm1/\sqrt{V_4}$ , with $w\ne-1$ — constrainable by $w(z)$ reconstructions. Non-local "swerves" / momentum diffusion, bounded by precision tests.

Strongest objections. No complete tractable quantum dynamics ( $e^{iS}$ over causets is even harder to define than the continuum path integral); the entropy/inverse problem — generic random causets are non-manifoldlike (Kleitman–Rothschild orders dominate), and dynamics must suppress them (unproven); no contact with the Standard Model; the everpresent- $\Lambda$ heuristic is not a controlled calculation; Reeh–Schlieder vacuum entanglement and Type III structure must be recovered, not obviously delivered by Sorkin–Johnston.

No-go theorems to confront. Lorentz-invariance no-go for discreteness (EVADED by Bombelli–Henson–Sorkin — the signature strength); Penrose–Hawking singularity theorems (reinterpreted as the signal that the emergent description fails); Weinberg–Witten (no fundamental flat-space graviton, so premises do not apply — must be checked in any continuum limit); Bell / Tsirelson (quantum dynamics must reproduce $2\sqrt2$ , not a local-hidden-variable cap of $2$ ); Reeh–Schlieder (the emergent vacuum must be cyclic/separating).

Relation to existing programs. This is Causal Set Theory (Bombelli, Lee, Meyer, Sorkin 1987; Sorkin; Dowker; Henson; Surya). The sprinkling result, Benincasa–Dowker action, Classical Sequential Growth, Bell causality, Sorkin–Johnston vacuum, and everpresent- $\Lambda$ (Ahmed–Dodelson–Greene–Sorkin) are all real, published. No novelty for the framework; the candidate direction is methodological emphasis on causal order as the load-bearing primitive. See domains/cosmology.md, domains/general-relativity.md, CONSTANTS_AND_SCALES.md.

Epistemic tag: SPECULATIVE. Confidence: low.

REFEREE VERDICT (H6) — keep, severity: minor. Sound as a research program; conflicts with no result or no-go. Fixed-background no-go theorems (Stone–von Neumann, Groenewold–van Hove, Haag) do not bear against a background-independent substrate. Corrections (incorporated). (1) The Benincasa–Dowker action is built from counts of small order-interval cardinalities with $d$ -dependent alternating coefficients, tending to Einstein–Hilbert in the mean — not a naive continuum-Ricci sum. (2) "Order + number = geometry" silently assumes the Hauptvermutung (no causet well-approximated by two macroscopically distinct manifolds), which is OPEN; the continuum theorem (Malament/HKM) is established, its discrete analogue is not. (3) $Z=\sum_C e^{iS/\hbar}$ is the central OPEN problem; only Classical Sequential Growth is rigorously defined. (4) The BD d'Alembertian is intrinsically non-local (a consequence of Lorentz invariance), with large fluctuations needing an intermediate mesoscale — omitted in the sketch. (5) The Sorkin–Johnston state can fail to be Hadamard — omitted. (6) Kleitman–Rothschild non-manifoldlike dominance is a serious obstacle. For balance: Sorkin's $\Lambda\sim V^{-1/2}$ estimate, predating the 1998 acceleration discovery, is a noted (CONTESTED, non-decisive) point in favor. Recommended tag: SPECULATIVE.

H7 — Quantum error-correcting code as algebraic substrate; bulk geometry and Type III QFT as the code's continuum limit

Statement. SPECULATIVE The fundamental description (at least of asymptotically-AdS gravitating systems) is a controlled sequence of finite-dimensional quantum systems carrying an (approximate) quantum error-correcting code (QECC): a small "logical" (bulk/low-energy/geometric) algebra is redundantly encoded into the entanglement of many "physical" (boundary/microscopic) degrees of freedom. Smooth geometry, the Type III von Neumann structure of continuum QFT, and the (approximate) emergence of a local bulk are properties of the code in a large-system limit — not fundamental. "Distance/area in the bulk" = entanglement/min-cut in the code (Ryu–Takayanagi as a code property); bulk locality = subregion/entanglement-wedge reconstruction = code recoverability.

Motivation. Anchored in the holographic area-law vs volume-law contradiction, the Type III obstruction, the black-hole information paradox / Page curve, and firewalls. The Type III contradiction is sharpest: continuum local algebras have no trace and no factorization, so "qubits sewing spacetime" literally does not exist at the continuum level. A QECC substrate turns this from paradox into a limit statement: the fundamental algebra is Type I (finite-dim, has a trace, factorizes — genuine subsystems exist), and the Type III, trace-less structure emerges only in the strict large- $N$ /continuum limit. Unitarity (vindicated by the Page curve) is automatic because the substrate is a closed finite quantum system. See UNIFYING_PRINCIPLES.md (entanglement as glue).

Mathematical sketch. Code subspace $\mathcal{H}_{\text{code}}\subset\mathcal{H}_{\text{phys}}$ , isometric encoding $V:\mathcal{H}_{\text{code}}\to\mathcal{H}_{\text{phys}}$ , $V^\dagger V=\mathbb{1}$ . Bulk locality = recoverability: $O$ in the entanglement wedge of boundary $A$ is reconstructable as $O=V^\dagger(O_A\otimes\mathbb{1}_{A^c})V$ . RT as a code theorem [ESTABLISHED for finite-dim codes with exact complementary recovery, Harlow 2017]: $S(\rho_A)=\text{Area}(\gamma_A)/4G_N+S_{\text{bulk}}(a)$ , the area term being the code's center/edge-mode contribution — derived, not assumed. Toy substrate: HaPPY perfect-tensor code / random tensor networks, where network min-cut tracks the RT surface (large bond dimension, on average). Type flow: at finite $N$ the algebra is Type I $_n$ ; the GNS limit $N\to\infty$ in the TFD/KMS state yields Type III $_1$ ; the crossed product by the modular flow (with an observer clock) yields Type II $_\infty$ (black hole) or II $_1$ (de Sitter static patch), with a renormalized trace making $S_{\text{gen}}=\langle A\rangle/4G+S_{\text{out}}$ an honest von Neumann entropy. Dynamics: $U(t)=e^{-iH_{\text{bdy}}t}$ ; the first law $\delta S_A=\delta\langle K_A\rangle\Rightarrow$ linearized bulk Einstein equations.

What it would explain. The Type III structure of continuum QFT as an artifact of the continuum limit (the "no local subsystems" contradiction becomes emergent, not fundamental); the area-law d.o.f. budget and $S_{\text{BH}}=A/4$ as a code property; the unitary Page curve and information escape (islands = entanglement-wedge recovery); why bulk EFT is robust yet breaks down (code distance sets when low-energy operators become non-reconstructable — a sharp firewall/AMPS diagnostic); linearized Einstein from the first law.

Predictions / tests. Largely structural: it must reproduce the full nonlinear Einstein equations from a code condition — currently NOT achieved (a concrete success/failure criterion). Quantum-simulation analog tests of RT entropy relations and subregion recovery on small holographic codes (validating the mathematics, not nature). A genuine physical prediction requires a de Sitter/cosmological version with observable consequences — none currently exists (the central empirical weakness). Predicts no firewall for young black holes but possible bulk-EFT breakdown at the code-distance/Page-time threshold.

Strongest objections. Almost everything rigorous is AdS-specific (negative $\Lambda$ , large $N$ , strong coupling); de Sitter holography lacks an established RT analog (arguably the central open question). AdS/CFT itself is a conjecture without general nonperturbative proof. Only linearized Einstein is recovered. ER=EPR for generic entanglement has no first-principles derivation. It presupposes QM exactly, so it does not address the measurement problem and is not itself post-quantum.

No-go theorems to confront. Weinberg–Witten (EVADED cleanly: the graviton lives in a higher-dimensional bulk; the boundary theory has no graviton and no covariant bulk $T^{\mu\nu}$ — emergence is across dimensions); Type III no-factorization (confronted: substrate is Type I, Type III emerges, crossed-product is the bridge); Reeh–Schlieder (vacuum entanglement from redundant encoding); Haag's theorem / Stone–von Neumann (emerge from the $N\to\infty$ limit; the finite substrate sidesteps Haag by construction); singularity theorems / Bousso bound (reproduced via RT; singularities as code breakdown).

Relation to existing programs. The "It-from-Qubit" / holographic-QEC program: Almheiri–Dong–Harlow 2014; Pastawski–Yoshida–Harlow–Preskill (HaPPY); Hayden et al. (random tensor networks); Harlow's RT-from-QEC theorem; Van Raamsdonk and Maldacena–Susskind (ER=EPR); FGHMVR; the crossed-product/Type II work (Chandrasekaran–Penington–Witten and collaborators); the Page curve via quantum extremal surfaces / islands (Penington; Almheiri et al.). No novelty for these. See domains/information-theory.md, domains/quantum-field-theory.md, UNIFICATION_LANDSCAPE.md. Allied with H0/H4; the negative counterpart is H2.

Epistemic tag: SPECULATIVE (with an ESTABLISHED finite-dim-code integrand core). Confidence: low.

REFEREE VERDICT (H7) — keep, severity: minor. Coherent, principled, active; conflicts with no result or no-go. Corrections (incorporated). (1) Type flow: which type ( $\text{II}_\infty$ vs $\text{II}_1$ ) is setting-dependent and physically loaded; the crossed product adjoins a bounded-below observer Hamiltonian, not "the modular Hamiltonian." (2) Harlow's RT-as-code theorem is rigorous for finite-dim codes with exact complementary recovery; realistic holography has only approximate recovery, so transfer to actual gravity is INFERENCE. (3) HaPPY/random-tensor-network min-cut = RT holds exactly only for restricted regions / in the large-bond on-average sense — the flat " $=$ " overclaims slightly. (4) First-law $\Rightarrow$ linearized Einstein only (correctly stated). Overclaims curbed: the headline "fundamental description IS a QECC" extrapolates from AdS/CFT toy/holographic examples to a de Sitter universe with no RT analog and no generic-entanglement-to-geometry theorem; Type III $_1$ is an independently established theorem of any local QFT (Buchholz–D'Antoni–Fredenhagen), so the code relocates rather than derives it ab initio; there is a genuine (self-acknowledged) tension between a finite-dim tensor-factorized substrate and the Type III continuum, only partially repaired by the crossed product. Recommended tag: SPECULATIVE.

H8 — Is quantum theory fundamental, or the rigid limit of an operational information theory? A constrained — and largely walled-off — post-quantum direction

Statement. SPECULATIVE framing on an ESTABLISHED scaffold; partly a negative result. The disciplined version of "the fundamental theory may be post-quantum" is not to posit an arbitrary post-quantum theory, but to ask: within generalized-probabilistic theories (GPTs), is standard complex-Hilbert-space QM the unique fixed point selected by a few information-theoretic / discreteness axioms? The honest finding is that the reconstruction program makes QM look rigid (nearly forced by operational axioms), and a battery of no-go and experimental results (Bell/Tsirelson, PBR, real-vs-complex tests, no-signaling, higher-order interference) largely (not totally) walls off viable post-quantum deformations. So the defensible output is cartographic: a precise map of where post-quantum room remains (essentially only untested regimes — quantum gravity / cosmological scales, or settings violating the axioms) and where it is closed.

Motivation. Motivated by (i) the measurement problem (unitary linearity vs definite outcomes); (ii) the Stone–von Neumann uniqueness vs inequivalent-representations domain mismatch; (iii) the information-theoretic reconstructions of QM (Hardy, CDP, CBH), tagged OPEN. If QM is reconstructible from operational axioms, then "the exact quantum framework is approximate" becomes a precise question: which axiom, relaxed, yields a post-quantum theory — and is that relaxation empirically open? The discipline demanded is to not assert a post-quantum TOE; the value is delineating the boundary.

Mathematical sketch. GPT scaffold: a system is a convex state space $\Omega$ in an ordered vector space; effects $e:\Omega\to[0,1]$ ; composition via local tomography $\dim V_{AB}=\dim V_A\cdot\dim V_B$ . Classical = simplex; quantum = density matrices (qubit Bloch ball). Hardy (2001): continuity of reversible transformations rules out classical $r=1$ ; together with the simplicity axiom, subspace axiom, and composite-system consistency this excludes $r\ge3$ and selects $r=2$ (complex QM) via $K=N^r$ . CDP (2011): purification + local tomography + causality $\Rightarrow$ finite-dim QM. Post-quantum candidates killed: $r=3$ higher-order (Sorkin) interference — bounded small but nonzero experimentally, consistent with $r=2$ ; real-Hilbert-space QM — excluded by Renou et al. (2021) network-Bell experiments under the assumption that independent sources combine as local real tensor factors (global-conjugation real QM, à la Stueckelberg, is not excluded); Popescu–Rohrlich boxes (CHSH $=4>2\sqrt2$ ) — no-signaling but not quantum, with information causality recovering Tsirelson for CHSH (but not provably the full quantum set — OPEN; cf. Almost-Quantum correlations). Born-rule rigidity: Gleason ( $\dim\ge3$ , assuming noncontextual frame functions) fixes $p=\text{Tr}(\rho E)$ . PBR: under preparation independence, within the ontological-models framework, the quantum state is $\psi$ -ontic.

What it would explain. Why QM has its specific structure (complex amplitudes, tensor composition, Born rule) as the unique operational info theory — demoting it from arbitrary postulate set to derived framework; where post-quantum modification could still hide (untested QG/cosmological/strong-curvature regimes, connecting to the semiclassical-gravity-vs-superposition contradiction); a reframing of the measurement problem as the question of which axiom (e.g. universal reversibility across all scales) might fail — the same locus where testable objective-collapse models (GRW/CSL, Diósi–Penrose) live.

Predictions / tests. Tightened Sorkin triple-slit bounds on the third-order interference parameter $\kappa\to0$ (any nonzero $\kappa$ is a genuine post-quantum signal). Precision CHSH never exceeding $2\sqrt2$ . Objective-collapse constraints from matter-wave interferometry, optomechanics, and underground spontaneous-X-ray-emission experiments — progressively excluding parameter space. Gravity-mediated-entanglement (Bose–Marletto–Vedral): if two masses entangle via gravity, the gravitational field cannot be classical (bearing on whether the quantum framework extends to gravity).

Strongest objections. Multiple inequivalent successful axiom sets (Hardy, CDP, CBH, Masanes–Müller) show QM is rigid but not which principle is fundamental vs conventional. Reconstructions are cleanest in finite dimensions; the infinite-dim/relativistic-field-theoretic case (where Type III and inequivalent representations live) is harder, so the bridge to QFT is incomplete. The no-go gauntlet leaves so little room that the expected outcome of the strictly post-quantum sub-claim is largely negative. The program reconstructs the formalism but does not resolve the measurement problem or fix an interpretation. The real-vs-complex exclusion is conditional on framework/locality assumptions (CONTESTED).

No-go theorems to confront. Bell + Tsirelson (must reproduce CHSH up to $2\sqrt2$ , not exceed it; PR boxes excluded; loophole-free Bell violations kill local-hidden-variable substrates absent conspiratorial superdeterminism); PBR (excludes broad $\psi$ -epistemic models within the ontological-models framework, assuming preparation independence); Gleason (fixes $\text{Tr}(\rho E)$ ); no-signaling (any signaling deformation excluded); Kochen–Specker (a value-assignment substrate must be contextual); Stone–von Neumann and its failure (must reproduce both finite-dof uniqueness and infinite-dof inequivalent representations in the appropriate limits).

Relation to existing programs. Synthesizes QM reconstruction (Hardy 2001; Chiribella–D'Ariano–Perinotti 2011; Clifton–Bub–Halvorson; Masanes–Müller; Barrett's GPTs), device-independent/correlation bounds (Tsirelson; Popescu–Rohrlich; Pawłowski et al. information causality), Gleason and PBR, the Sorkin higher-order-interference program and its null results, the Renou et al. real-vs-complex test, and objective-collapse phenomenology. No novelty. The deliverable is the explicit verdict: reconstruction supports "QM is rigid," but post-quantum room is near-closed in any tested regime. See domains/quantum-mechanics.md, domains/information-theory.md, OPEN_PROBLEMS.md, ASSUMPTIONS_LEDGER.md.

Epistemic tag: SPECULATIVE (framing thesis) on ESTABLISHED scaffolding; "which single principle carves out the quantum correlation set" is OPEN. Confidence: very low.

REFEREE VERDICT (H8) — keep, severity: minor. Sound as a research-direction statement; its "mostly a negative result" framing is its main virtue. The cited results are real and point the claimed way, but several need tightening (incorporated above). (1) Hardy's uniqueness needs the full axiom set — continuity rules out classical $r=1$ ; simplicity + composite-system consistency exclude $r\ge3$ . "Continuous reversibility alone forces QM" is false. (2) Higher-order (Sorkin) interference is experimentally bounded-small, not "null to high precision," and is consistent with QM rather than a proof of $r=2$ . (3) Renou et al. rules out real QM only under the local-tensor-product rule for independent sources in a network; global-conjugation real QM (Stueckelberg) is not excluded. (4) Information causality recovers Tsirelson for CHSH but does not provably carve out the full quantum set (OPEN; Almost-Quantum). (5) PBR's $\psi$ -ontology is internal to the ontological-models framework and assumes preparation independence. (6) The " $q\to1$ / $N\to\infty$ limit of a finite computational information theory" framing is a loose metaphor, not an output of the reconstruction program — flag SPECULATIVE and decouple from the established results. Note: the self-tag "highly-speculative" under-credits the rigorous core; the reconstruction-rigidity and no-go results are ESTABLISHED. Recommended tag: SPECULATIVE-with-established-scaffolding.

Rejected / not currently supported

No candidate direction was judged unsound or vacuous by the referee. All eight passed at keep=true, minor severity. Two caveats are recorded here so future iterations do not re-litigate settled points:

H2 is retained as a negative/limiting result, not as a positive framework. Its claim is precisely that the entanglement-geometry lens is not yet a supported route to a universal theory. It belongs above (it is a refereed, kept direction) but should never be cited as endorsement of emergent-spacetime universality.
H4's original mathematical sketch was marked isSound=false for presenting the inferential bridge (algebra+state $\Rightarrow$ background-independent geometry) as nearly established and for the $\beta=1$ / $\beta=2\pi$ conflation. The refined statement repairs both; the refined H4 is what is kept. The original, unrefined phrasing is not supported.

When a future direction is judged genuinely unsound or vacuous, record it here with: its ID, the statement, and the specific reason for rejection (logical inconsistency vs no-go violation vs vacuity vs domain-mismatch masquerading as a result — see EPISTEMICS.md for the distinction).

Cross-cutting observations

Convergence. H0, H1, H3, H4, H6 all wager that causal or algebraic structure precedes metric geometry. H1/H3/H6 take causal order as primitive; H0/H4 take the von Neumann algebra / modular flow as primitive; H7 takes the code / entanglement as primitive. These are not independent — H1 and H4 both invoke relational/modular time; H0, H4, H7 all lean on the crossed-product Type II construction. INFERENCE
Shared load-bearing obstacles. Three recur across nearly every direction and gate all of them: Weinberg–Witten (forces any emergent graviton through an emergent extra dimension), the Type III $_1$ obstruction (forbids the naive local-qubit picture, forcing algebraic reformulation), and AdS-confinement of the rigorous results (the wrong sign of $\Lambda$ for our universe). [ESTABLISHED that these are the binding constraints; see GAPS_AND_CONTRADICTIONS.md.]
The recurring honest limit. Every direction that derives gravity derives only the linearized Einstein equation; no full nonlinear, background-independent reconstruction from an entanglement/algebraic/causal principle exists. ESTABLISHED as the current frontier.

Refinements from the iteration-2 deep-research tracks (see notes/), each refereed (0 flagged unsound). These update H0/H2/H4 in place without renumbering.

H2-R1 (de Sitter graduation, partial update). [INFERENCE] Among H2's graduation criteria, criterion (1) (a dS RT/QES analog with a controlled dual and derived entropy) is now partially met in a weak sense — CLPW's Type II₁ static-patch algebra gives a derived finite Λ>0 entropy with a maximal-entropy state (= empty dS) — but with no RT surface and no boundary dual. Criteria (3) (nonlinear Einstein from entanglement) and (4) (operational generic ER=EPR) remain entirely unmet in dS, and the dS entanglement first law (OP-41) does not exist. Referee correction: it is not that "algebra type cleanly encodes the Λ sign" — local QFT algebras are Type III₁ for all Λ signs; it is the II₁-vs-II∞ distinction (existence of a tracial maximal-entropy state) that carries the dS-specific information. H2's negative meta-conclusion stands. [→ notes]
H4-R2 (graded status of the crossed-product program). Replace any blanket "ESTABLISHED" reading of H4 with graded conjuncts: ESTABLISHED, given a background the crossed product turns Type III₁ into Type II with a finite $S_{\rm gen}=\langle A\rangle/4G+S_{\rm out}$ (CLPW/CPW 2022; Witten 2308.03663), and KLS (2024–25) extends this to any bifurcate Killing horizon (Kerr, collapse, Schwarzschild–dS); INFERENCE, semiclassical a crossed-product generalized second law (Faulkner–Speranza 2024), with the beyond-semiclassical fate open; [scope-limit, ESTABLISHED] none of this derives geometry or time — a fixed metric with a symmetry and a stationary state is input throughout, and Witten 2308.03663's "background independence" means state-independence on a fixed worldline algebra. Net: a rigorous reorganization that encodes, not generates, geometry; the nonperturbative fate is OPEN (OP-44, Giddings 2505.22708). [→ notes]
H4-R3 (observer-relativity of gravitational entropy). [INFERENCE] The Type II crossed-product (generalized) entropy is observer / quantum-reference-frame dependent and fixed only up to a state-independent additive constant; the leading area term remains a robust per-observer geometric invariant, but the full von Neumann entropy of a gravitational subregion is frame-relative, not an absolute scalar attached to "the geometry." Reinforces the encode-not-generate verdict. [→ notes]
HYP-GIE-DISCRIM (BMV scaling discriminator). [INFERENCE] In an achievable parameter window, genuine quantum-gravity gravitationally-induced entanglement scales as $m^2$ , whereas the classical-gravity-plus-quantum-matter loophole (Aziz–Howl) scales as $m^3$ ; a mass/distance scaling test could separate them (and from EM/Casimir backgrounds), tying OP-40 to a concrete discriminator. [→ notes]
HYP-ENTROPIC-GRAVITY-SPLIT (disaggregation of "entropic/emergent gravity"). Split the bundled principle (UNIFYING_PRINCIPLES.md §6, UNIFICATION_LANDSCAPE.md §7) into: (a) ESTABLISHED GR carries genuine thermodynamic structure — Jacobson 1995, given the area law + Unruh $T$ (the AdS first-law route yields only the linearized equations; Padmanabhan is an INFERENCE-level rewriting, not a co-equal theorem); (b) SPECULATIVE gravity is foundationally emergent from identified microscopic DOF (none identified — OP-45); (c) CONTESTED → disfavored Verlinde-type entropic gravity as a dark-matter replacement (RAR fit comparable to MOND; disfavored at cluster/CMB scales; a ≥6σ early/late-type RAR dependence defeats all baryon-only modifications). [→ notes]

Unified-program note. The iteration-2 synthesis (notes) judges these directions to cohere partially: a single geometry-from-algebra program (H0/H1/H3/H4/H6/H7 + entropic gravity) and a distinct is-the-quantum-framework-final program (Born/linearity + BMV), joined only by no-signaling/causal-order and by BMV as a logical gate. The defensible overall posture is a program of demotion-and-derivation, not a theory-of-everything; see FINDINGS.md § "Iteration 2 update."

From the iteration-3 attempts + red-team (see notes/ iter3-*).

H2-R1 further correction — II₁ is a bounded-area feature, not a Λ-diagnostic. [ESTABLISHED] The iteration-2 H2-R1 correctly removed "algebra type encodes the Λ sign," but even the II₁-vs-II∞ distinction does not track Λ: Schwarzschild–de Sitter (Λ>0) is II∞×II∞ (Kudler-Flam–Leutheusser–Satishchandran, PRD 111 025013). The static patch is Type II₁ because its area is bounded (finite maximal entropy), set by the horizon/observer-Hamiltonian spectrum — not by the sign of Λ. [→ notes]
HYP-dS-NOMIN (conjecture). [OPEN] There is no holographic de Sitter entanglement first law in the strict AdS sense (boundary CFT + anchored RT minimization ⟹ pointwise $\delta G_{\mu\nu}$ ), because dS lacks both a timelike conformal boundary (no continuum of anchoring regions) and a minimization principle (the static-patch variational principle is a concave maximization, $dE=-T\,dS$ ). Status: OPEN, not a no-go — one un-foreclosed escape exists (an $SO(d+1,1)$ modular-Hamiltonian family); whether it yields a pointwise field equation is untested. [→ notes, OP-41]
Encode-vs-generate criterion (methodological). [INFERENCE] A construction generates (rather than encodes) geometry only if the emergent metric/causal structure is fixed by abstract (algebra, state, dynamics) data with no pre-chosen factorization, region, symmetric/Hadamard vacuum, or causal order. Graded axes $\Delta_{\rm geo}$ (geometric input) and $\kappa_\Phi$ , plus a 5-item smuggling checklist $g_1$ – $g_5$ . Across six surveyed constructions (CLPW, Connes–Rovelli, Leutheusser–Liu, Cao–Carroll, causal sets, tensor networks) none clears the bar — the frontier coincides with the Type III₁ no-factorization fact. This makes the H0/H4 "encodes, not generates" verdict operational. [→ notes, OP-46]
Streetlight caveat on the convergence claim. [INFERENCE] The "Convergence" observation above (H0/H1/H3/H4/H6 wagering that causal/algebraic structure precedes geometry) is partly a shared-input artifact: H0/H4 share Killing-symmetric vacua + AdS/large- $N$ ; H1/H3/H6 share the Malament/HKM theorem + the unproven Hauptvermutung; H7 is AdS/large- $N$ . Agreement among methods that share their assumptions is weaker evidence for the wager than genuine independent convergence. [→ notes]

From the iteration-4 attacks + convergence audit (see notes/ iter4-*). Net: the standing verdict is unchanged; the central diagnosis is harder.

HYP-dS-NOMIN-R3 (dS first-law escape executed → near-no-go). [INFERENCE, high] The one un-foreclosed iteration-3 escape — an $SO(d+1,1)$ family of static-patch modular Hamiltonians — was executed and fails: a finite-dimensional isometry orbit supplies $\le \dim SO(d+1,1)=10$ global Killing-charge constraints, which cannot Radon-invert to the infinite-dimensional local $\delta G_{\mu\nu}(x)$ (no "radius modulus" / local region continuum); the family is geometric in a common state only at Bunch–Davies (probe null there); its first-law content is second-order with $\delta^2 S_{\rm gen}<0$ (concave GSL, not an invertible equality). Corroborated by the published Chen–Xu execution (arXiv:2511.00622), which reaches entropy-matching + a 2nd-order quantum first law and no $\delta G_{\mu\nu}$ . Scoped to the strict unitary single-static-patch isometry-orbit common-state route; non-unitary dS/CFT and Jacobson small-ball routes remain open. [→ notes]
HYP-dS-CARDINALITY (reusable principle). [INFERENCE] Einstein-from-first-law requires an infinite-dimensional local region family probing every bulk point (AdS: an RT surface through every point/orientation, Radon-invertible to $\delta G_{\mu\nu}(x)$ ). Isometry orbits of a maximally-symmetric background are finite-dimensional and yield only global Killing charges — so local dS dynamics-from-entanglement needs an independently infinite local family (Jacobson small balls), which the orbit structurally lacks.
OP-48 disjunction + HYP-SPLIT-GEOM (GPT on Type III₁). [ESTABLISHED/INFERENCE] No-go for exact composition / local-tomography on a single Type III₁ factor; split-property qualified yes for nested regions with a nonzero collar (intermediate Type I factor, genuine factorization). But the rescue is not clean — the canonical intermediate factor is singled out only by the vacuum's modular conjugation $J$ , relocating the BW/CHM geometric-reference circularity to the factorization stage, and failing for adjacent regions. Full A+B merge needs a state-/collar-free composition surrogate (OPEN). [→ notes]
HYP-CKV-VACUITY + GAP-NCG-LORENTZIAN-SIGNATURE (encode→generate near-vacuous). [INFERENCE] Recovering geometry from a non-symmetric modular flow is squeezed by Borchers/Wiesbrock (half-sidedness ⟺ positivity — the emergent translation comes from a relocated positivity premise) and Caminiti et al. (arXiv:2502.02633: recovered generators are conformal Killing vectors, presupposing symmetry). Connes' reconstruction yields only a Riemannian manifold; a non-CKV readout (modular-Berry / Connes spectral distance) yielding a Lorentzian metric is the only unrealized move that would turn encode into generate. [→ notes]
Convergence note. The iteration-4 audit finds the new-insight trend diminishing (third iteration at PARTIAL); analytical saturation ≈1–3 iterations away; object-level confirmation non-forecastable. See FINDINGS.md § Iteration 4 update.

From the closing analytical push (see notes/ iter5-*). No track flipped the headline; the diagnosis is now readout-independent. Consolidated in CONCLUSION.md.

HYP-CKV-VACUITY — upgraded to a readout-independent effective no-go. [INFERENCE — HIGH that no current construction generates Lorentzian geometry from non-symmetric modular data; MEDIUM that it is a true no-go vs a limitation of known readouts] The OP-46 non-CKV Lorentzian readout was executed via modular-Berry / kinematic space. It produced the first emergent Lorentzian metric in the survey (Huang–Ma dS₂, arXiv:2003.12252), crossing the signature barrier the Riemannian Cao–Carroll and Connes-spectral routes fail — but the signature is inherited from the symmetric CFT vacuum, not generated. The obstruction now closes at the same signature-installation step across all three known readouts (stress-tensor-flux/Sorce–CKV; modular-Berry/zero-mode; Lorentzian-NCG/η-foliation-twist). Formally open only for a hypothetical readout installing no symmetry/η/foliation/twist (none exists) — the single hedge keeping the headline from a theorem. [→ notes]
OP-48(c) conditional no-go. [ESTABLISHED/INFERENCE] On a single Type III₁ factor, "collar-free + state-independent" composition and "singles out QM" are mutually exclusive: the genuinely collar-free surrogate (Haagerup standard form + Connes fusion) is type-agnostic and a universal embezzler ( $\kappa_{\max}=0$ , van Luijk et al.); any QM-selecting axiom forces a Type I factor (split construction), re-importing the collar + reference state. The QM-selecting composition axiom coincides with the geometry-importing input — tightening "encodes-not-generates." [→ notes]
HYP-dS-CARDINALITY-R1 — the root obstruction. [INFERENCE, high] In every known route to Λ>0 Einstein-from-entanglement — AdS-FGHMVR, Jacobson small-balls, CLPW Type II₁ — the area-entropy coefficient $\eta=1/4G$ is an INPUT, not an output. The two surviving dS routes fail on this: the pseudo-entropy first law inherits non-unitarity (complex central charge); Jacobson-in-CLPW re-imports η (the II₁ trace encodes one area datum, not a local continuum). [→ notes]
Saturation note. Fourth consecutive iteration at PARTIAL / encodes-not-generates / not-yet-physics; the named decidable list is settled; the verdict-flipping lever did not flip. The analytical program is saturated — see CONCLUSION.md. The only genuinely-open analytical lever is OP-44/OP-CP1 (nonperturbative crossed-product fate).

From the new-input iteration (see notes/ iter6-*). No refinement moves the headline; the obstruction broadens (four readout families), a new named hedge is added (HYP-ENCODING-SCREEN), and one program hypothesis (H5) gains confidence inside a confirmed bound.

HYP-CKV-VACUITY-R3 — four readout families; the installation/selection trichotomy; two new smuggle-list items. [INFERENCE — HIGH as a classification of the surveyed pool; MEDIUM that the trichotomy is exhaustive of possible mechanisms — the standing hedge] Every known construction with a Lorentzian-signature output traces the Lorentzian datum to exactly one of three forms: (i) a symmetry-distinguished state (CKV/symmetric-vacuum inheritance — the iter-4/5 closures; also the symmetric saddles of Lee's demonstrations); (ii) an indefinite-pairing axiom (η/Krein/time-orientation/twist in NCG; the $(\le n,\le n)$ -eigenvalue axiom + indefinite spin scalar product in causal fermion systems — the fourth readout family, closing at the same signature-installation step: the necessity lemma, derived and numerically re-checked by finder and referee, shows that for positive-semidefinite operator classes the CFS lightlike relation is empty — the indefiniteness axiom is a necessary carrier of the signature, the Krein η relocated into the first definition); (iii) a signature-valued template with dynamical selection (the $\mathcal S$ -parametrized hypersurface-deformation template in Lee, JHEP 06 (2020) 070, arXiv:1912.12291; the signature-potential toy model of Bermudez 2025 [SPECULATIVE]). Forms (i)–(ii) install the Lorentzian datum; form (iii) installs the possibility space {Euclidean, Lorentzian} and generates only the choice. Smuggle list extended by two named items: (a) indefinite-pairing/(n,n)-flag axiom; (b) signature-valued parameter space / deformed-algebra template. CFS status: ENCODE as demonstrated (all worked examples are Dirac-sea constructions on pre-given Minkowski; minimizer existence proven only for finite-dimensional $\mathcal H$ ; no substantive published external critique found in a targeted search OPEN); candidate-GENERATE as a program, unproven. The open target compresses to its minimal form: derive the indefinite pairing (a canonical (n,n) operator class, or equivalently a Krein η — e.g. from the modular conjugation $J$ ) from $(\mathcal A,\omega)$ modular data rather than positing it. No such derivation exists or is sketched OPEN. Process matrices do not help: local causal arrows are presupposed at every node, and spacetime-embedded indefinite causal order provably reduces to definite acyclic order (Vilasini–Renner, PRA 110, 022227 (2024); companion arXiv:2408.13387). No post-2024 weakening of the Borchers–Wiesbrock positivity–translation linkage was found. [→ notes]
HYP-dS-CARDINALITY-R1 — extended on four fronts; the pattern now also covers a phenomenological chain. [INFERENCE, high] (i) QRF observer-relativity: De Vuyst–Eccles–Höhn–Kirklin prove PW=CLPW and that entropies assigned to one subregion relative to distinct observers "can drastically differ" (arXiv:2405.00114; follow-up 2412.15502, verified) — not only is the trace normalization observer-tied, the entropy value is QRF-relative. (ii) HWZ-to-dS lift: the Hollands–Wald–Zhang dynamical entropy extends to dynamical dS cosmological horizons (Zhao, arXiv:2503.16138, EPJC 85:750 (2025)) with η supplied by the Wald-type formalism, not derived INFERENCE, medium. (iii) Maldacena's observer (arXiv:2412.14014) cancels the imaginary phase of the dS sphere partition function — but the $A/4G$ coefficient still comes from the classical saddle: the observer fixes the phase, not η. (iv) String side: Zigdon (arXiv:2604.25918) — under the stated assumption that the QG sphere partition function receives a nonzero $1/G_N$ contribution, perturbative string theory admits no dS×compact solutions at any order in $\alpha'$ , $g_s$ (tree-level action vanishes), in tension with Gibbons–Hawking: on the perturbative-string side no $A/4G$ saddle is available at all under those assumptions INFERENCE, medium — conditional, silent on nonperturbative dS. η is input — or absent — never output. (v) Phenomenological extension: the Verlinde–Zurek/GQuEST chain installs $\eta=1/4G$ at the modular→metric step — the same coefficient, now in an inference chain to an observable rather than a derivation route; a GQuEST positive would, after exclusion of alternatives, be most naturally read as "modular fluctuations gravitate with the assumed η" INFERENCE. [→ notes, notes]
HYP-ENCODING-SCREEN (new named item — the third hedge). [INFERENCE — conditional schema; its application inherits the OP-46 encode-only classification's MEDIUM hedge] Proposition: let Φ be encode-only in the OP-46 sense (consumes background data $g_1$ – $g_5$ ; outputs no metric data beyond input) and let $T_A=(\mathcal A,\omega;\Phi)$ be a wager-true theory whose entire bridge to spacetime observables passes through Φ. Then $T_A$ 's observational content equals that of the semiclassical EFT $E$ whose background Φ consumed — the prediction map factors $(\mathcal A,\omega)\to E\to\{P(O)\}$ — so no observable separates $T_A$ from any geometry-first theory with IR limit $E$ . Corollary (one-way — "iff" weakened to "only if" per referee): given the screen, a class-level distinguishing test exists only if some wager-true theory has a non-factoring (generative) bridge; whether a generative bridge suffices for a class-separating observable is [OPEN]. The OP-46 verdict-flipper is a prerequisite for, not a guarantee of, the missing distinguishing experiment — CONCLUSION.md §7's no-test claim and §8's reopening condition coincide as obstructions. Corollary (portability, survey-scoped): every surveyed observable-statement conjecture (noise PSDs, decoherence rates, w(z) statistics) transplants between wager-true and geometry-first theory classes; confirmation yields abduction, never class separation. A distinguishing prediction would require a "reverse Weinberg–Witten" theorem ("observable X is realizable only without fundamental metric degrees of freedom"); none exists, and Weinberg–Witten itself runs the other direction and is evaded by holography. Hedge accounting: the §7 no-test claim is exactly as strong as the encode-only classification — it carries the OP-46 MEDIUM hedge explicitly. [→ notes]
H5 (amplitudes/positive geometry) — scope rewritten, confidence low → low-to-medium; key bound confirmed and sharpened. [SPECULATIVE, low-to-medium] Two-tier established core (referee-approved): the amplituhedron core is proven at tree level for planar N=4 (BCFW tiling, Inventiones 2025; a second amplituhedron exists for ABJM); the surfaceology core (Arkani-Hamed–Frost–Plamondon–Salvatori–Thomas, arXiv:2309.15913, 2311.09284) covers massless colored theories at all loop orders and all genus — non-SUSY gluons in any D, pions, colored fermions — with loop-integrated formulas existing but formal (physical-kinematics continuation open: the integrand-only bound is only PARTIALLY dissolved; the massive-non-SUSY-matter clause of the old bound survives). Rodina correction (binding, per referee): Rodina (PRL 134, 031601 (2025)) and Zhou (arXiv:2604.07195) take locality (and, for Zhou, unitarity) as HYPOTHESES — so the result enters as "hidden zeros/UV-scaling uniquely determine the Tr(φ³) tree amplitude within a local ansatz, plus a 2-split factorization not implied by unitarity (extended to all topological orders and string amplitudes, arXiv:2405.09608)" — genuine and new, one rung below "locality as derived." Cosmohedra: a single polytope for the full Tr(φ³) cosmological wavefunction, combinatorics proven 2026; kinematic flow is a boundary reformulation of bulk time evolution on fixed FRW, not generation of time [SPECULATIVE]. The decisive referee bound stands and now looks structural: no positive geometry / surface formalism for gravity (no color, no surface; June 2026) [OPEN] — and every construction consumes flat-space/FRW kinematic data as input, consistent with encode-not-generate. Gravitational EFT bootstrap: consistency conditions remain rigid, with the new caveat that gravitational positivity is loop-corrected (calculable G-suppressed negativity; state dispersively; D=11 has no upper bound — one-sided). Five JHEP issue/page references from this track enter BIBLIOGRAPHY.md [unverified] (search-snippet provenance); the gravituhedron paper 2012.15780 is single-author Trnka. [→ notes]
HYP-GIE-DISCRIM — complicated by a mass-independent entangling-phase claim. [SPECULATIVE/INFERENCE — recent preprint] Braccini–Serafini–Bose (arXiv:2602.19306; note Bose is a BMV original author) claim analytically that in qubit-coupled Stern–Gerlach protocols the leading-order gravitational entangling phase between the two qubits is mass-independent (unitary and open dynamics; higher orders reintroduce mass dependence via imperfect recombination). If it holds: (a) the $m^2$ -vs- $m^3$ mass-scaling discriminator is protocol-dependent and must be restated per-protocol; (b) the experimental mass requirement for QGEM-class tests may relax in qubit-phase designs. Related status (per referee correction on rebuttal independence): the Aziz–Howl loophole is under attack via three distinct arguments from two author groups — Marletto–Oppenheim–Vedral–Wilson (2511.07348; the no-go's own proponents) and Tang–Yang–Han–Kan–Liu–Xue (2512.13675 and 2604.16276 — same six authors; 2604.16276 is now verified real, upgrading the tabletop note's unverified flag, with the "~10⁻¹³ d displacement" figure remaining unverified) — trending dead but [CONTESTED] absent a concession, and the rebuttal literature is not yet sociologically independent of the no-go's stakeholders. [→ notes]

From the lever-execution iteration (4 attack tracks on the four named decidable levers, all adversarially refereed; see notes/ iter7-*). No refinement moves the headline (sixth consecutive iteration). Two new named items enter per the AGENTS.md §3 named-ID convention (HYP-MODULAR-TRICHOTOMY; HYP-FACTORIZATION-IS-GEOMETRY); HYP-CKV-VACUITY advances to -R4 with its hedge hardened; HYP-ENCODING-SCREEN inherits the hardening.

HYP-MODULAR-TRICHOTOMY (new named item — the iteration-7 no-go lemma). [INFERENCE — ansatz-scoped on its sesquilinear leg; enumerative on its non-sesquilinear leg] Under a flagged modular-constructibility ansatz (sesquilinear Hermitian pairings canonically built from $(\mathcal M,\Omega)$ have Gram operator a self-adjoint element of $\{\Delta\}''$ ), every canonical pairing from modular data is: (i) positive ( $f(\Delta)\ge0$ — incl. the Hilbert form and, on its OS domain, the cone form), or (iii) a sign-of-Δ Krein form $\langle\cdot,\varepsilon(\Delta)\cdot\rangle$ for any sign function $\varepsilon$ , $J$ -odd or not (the referee's strengthening of prong (iii) beyond $J$ -odd sign functions) — each genuinely indefinite but unitarily universal in the Borchers/HSMI class, generically algebra-incompatible ( $\mathrm{Ad}\,\varepsilon(\Delta)\notin\mathrm{Aut}(\mathcal M)$ , numerically demonstrated finite-dim), and non-localizable without a posited carrier; the remaining canonical pairings (ii) — symplectic $\mathrm{Im}\langle\cdot,\cdot\rangle|_K$ , complex-symmetric $\langle J\cdot,\cdot\rangle$ , real $J$ -split — are signature-free by separate enumeration of the known form types, not by the ansatz. Hence no geometry-bearing, algebra-compatible, localized Lorentzian $(n,n)$ pairing is derivable from modular data alone, modulo the ansatz — the explicit residual. Companion positive sub-result (the lemma's input, referee-corrected): $\eta_0=\mathrm{sgn}(\ln\Delta)$ is a distinguished $J$ -odd self-adjoint unitary in $\{\Delta\}''$ — one member of the infinite family of $J$ -odd sign functions $\varepsilon(\Delta)$ , $\varepsilon(1/\lambda)=-\varepsilon(\lambda)$ (e.g. $\mathrm{sgn}(\sin\ln\Delta)$ , referee-constructed and numerically confirmed); it is singled out only by an added modular-rescaling-invariance axiom ( $\Delta\mapsto\Delta^s$ ). The form $\langle\cdot,\eta_0\cdot\rangle$ is indefinite iff $\omega$ is non-tracial, with Krein signature $(\infty,\infty)$ on any type III factor (spec $\Delta=[0,\infty)$ holds only for III₁; the $(\infty,\infty)$ conclusion survives for all type III — referee correction). Non-uniqueness strengthens the negative verdict: every member of the family is equally canonical and equally geometry-void. The vacuity theorem ( $\Delta_{\rm geo}=0$ , unitary universality by one multiplicity cardinal) substantially coincides with the known uniqueness of non-degenerate standard pairs, extended by "any fixed function of $\Delta$ rides along" — the geometric-vacuity corollary is the new-to-the-wiki content. Structural bonus, severity-none per referee (the strongest finding of the track): $J$ carries reflection positivity abstractly — $\langle a^*\Omega,Ja\Omega\rangle\ge0$ for all $a\in\mathcal M$ by self-duality of the natural cone (consistent with the Neeb–Ólafsson standard-subspace ↔ reflection-positivity dictionary) — so the $g_3$ trap is upgraded from symmetric-state inheritance to a structural fact: the modular conjugation sits on the Euclidean/positive side, and no $J$ -induced pairing can be secretly Lorentzian. Verified analytically and over 500 numerical trials, independently replicated by the referee (fresh code, fresh RNG). [→ notes]
HYP-CKV-VACUITY-R4 — the J-route is closed; the hedge hardens MEDIUM → MEDIUM-HIGH (conditional); the target re-compresses to the carrier problem. [INFERENCE — HIGH as a classification of the surveyed pool; MEDIUM-HIGH that the closure is a true no-go, conditional on the HYP-MODULAR-TRICHOTOMY constructibility ansatz] The -R3 compressed target ("derive the indefinite pairing — Krein η / (n,n) operator class — from modular data; no such derivation exists or is sketched") receives its mandatory precision repair: the "no such derivation exists" sentence is now literally false — η₀ achieves the literal target, smuggle-free per the house $g_1$ – $g_7$ instrument — yet nothing flips, because the derived pairing is provably geometry-void (unitary universality ⟹ $\Delta_{\rm geo}=0$ ; algebra-incompatibility; non-localizability). The -R3 compression was therefore lossy. The faithful residual: derive an algebra-compatible, LOCALIZED (fiberwise) $(n,n)$ pairing from modular data — which is the carrier problem (see HYP-FACTORIZATION-IS-GEOMETRY below). Slogan, refereed: the modular sign is to signature what thermal time is to time. Literature direction-map [OPEN — negative search, referee-corroborated, scoped to both sessions' queries]: all prior art runs Krein → modular — Gottschalk (JMP 43 (2002) 4753) assumes Krein structure and derives Bisognano–Wichmann; Jakóbczyk's 1980s line builds modular theory inside a posited indefinite-metric space (1985, OSTI 5135323, record-verified); the NCG twist papers obtain the Krein space from the twist (Devastato–Farnsworth–Lizzi–Martinetti, JHEP 03 (2018) 089, arXiv:1710.04965 — author list corrected; Nieuviarts 2024–26). No publication derives a Krein fundamental symmetry from $(\mathcal M,\omega)$ modular data. [→ notes]
HYP-FACTORIZATION-IS-GEOMETRY (new named item — where all five routes close). [INFERENCE, high as a convergence observation; the underlying question OPEN] The carrier/localization step — supplying a local carrier (a factorization, region net, fiber structure, or preparation primitive) on which an abstract algebraic datum could become geometric — is the single step at which five independent routes now close: (1) the Type III₁ no-factorization frontier (iter 3: the encode/generate boundary coincides with the hand-chosen factorization); (2) the split-property rescue (iter 4: the canonical Type I factor is fixed only by the vacuum's $J$ — the circularity relocated to the factorization stage, HYP-SPLIT-GEOM); (3) the CFS carrier posit (iter 6: the $(\le n,\le n)$ operator class is the Krein η relocated into the definition, on a posited carrier); (4) the OP-48c composition layer (iters 5–7: QM-selecting composition forces the Type I collar; iteration 7 relocates the failure further, to the normal-product-preparation primitive — the state postulate as installation choice); (5) the η₀ localization step (iter 7): the canonical modular sign exists but is non-localizable — universal, algebra-incompatible, carrier-free. Statement of the named problem: exhibit (or refute the existence of) an algebra-compatible, localized $(n,n)$ pairing — equivalently a modular-canonical local carrier — derived from $(\mathcal M,\omega)$ data alone. This is now simultaneously the CONCLUSION §8 reopening condition and (via HYP-ENCODING-SCREEN) the prerequisite for any distinguishing experiment. [→ notes, notes]
HYP-ENCODING-SCREEN — hardening note (hedge accounting). [INFERENCE — conditional schema, unchanged] The screen's application inherits the OP-46 encode-only classification's hedge, which iteration 7 hardened: MEDIUM → MEDIUM-HIGH, conditional on the HYP-MODULAR-TRICHOTOMY constructibility ansatz. The §7 no-test claim is now exactly as strong as that hardened classification — same condition attached. (The A2 track mildly reinforces the screen from the other side: the P1a descent discharge works by consuming pre-installed $\mathfrak{so}(d+1,1)$ structure — the discharge mechanism is itself an encoding step.) [→ notes, notes]
OP-48(c) horn C — upgraded to theorem-conditional-on-framework (registry note; full statement in OPEN_PROBLEMS.md). [INFERENCE, high — the FP-forcing step is textually grounded but interpretive ("forced-modulo-rejecting-OP-48's-question", per referee)] The full-pair reading of "recovers QM" is forced: all three reconstructions quantify composition over the designated pair and install product states at framework level; the separation half of local tomography holds on the III₁ Haag-dual pair (own derivation, referee re-derived), so the failure locus is exactly the normal-product-preparation primitive (PS), with the uniform $1-1/\sqrt2$ gap; (PS) on the pair forces Type I. Residue = the normal-state postulate + the composition-primitive choice: (X1) singular product states (existence via the corrected Schlieder/Roos + Hahn–Banach route — the max-tensor universal-property route is invalid, referee MAJOR) or (X2) non-OPT III₁-embezzlement composition; both exit the framework. Encodes-not-generates recurring at the state-postulate layer. [→ notes]
Byproduct (A3): $S_{\rm SVD}({\rm cap})=S_{\rm dS}/2$ , region-independent. [SPECULATIVE] Under the A3 phase bookkeeping AND the unproven junction additivity (OP-49 residual (c′)), the universal part of the dS₃ hemisphere SVD spectrum gives $S_{\rm SVD}=\pi\tilde c/6=\mathrm{Re}\,S_A^{\rm pseudo}=S_{\rm dS}/2$ for every cap — the only intrinsically real two-state entropy retains exactly the horizon constant and zero local geometry (the ENCODES tie-in survives only at this conditional level). The numerical value is [SPECULATIVE] (in any trace-class realization $\mathrm{Tr}\sqrt{\mathcal T^\dagger\mathcal T}>1$ strictly, since equality would force a real pseudo-spectrum contradicting the verified complex $S_A$ — referee correction); the region-independence is [INFERENCE, conditional on (c′)]. Retained as a falsifiable byproduct prediction for any future direct computation, with both flags. [→ notes]

References

For all cited works — Tomita–Takesaki and Connes' classification, Bisognano–Wichmann, Buchholz–D'Antoni–Fredenhagen, Ryu–Takayanagi/HRT, Lewkowycz–Maldacena, FGHMVR, Casini–Huerta / Ceyhan–Faulkner (QNEC), Maldacena's AdS/CFT, Maldacena–Susskind (ER=EPR), Witten and Chandrasekaran–Longo–Penington–Witten (crossed product / Type II), Jacobson (1995, 2015), Malament (1977) / Hawking–King–McCarthy (1976), Bombelli–Lee–Meyer–Sorkin and the causal-set literature, Page–Wootters, Connes–Rovelli, Strominger et al. (celestial/BMS/soft theorems), Arkani-Hamed–Trnka and the positive-geometry program, Almheiri–Dong–Harlow / PYHP / Harlow (holographic QEC), Penington / Almheiri et al. (Page curve / islands), Hardy (2001), Chiribella–D'Ariano–Perinotti (2011), Gleason, PBR, Tsirelson, Popescu–Rohrlich, Renou et al. (2021) — see BIBLIOGRAPHY.md. No citation, number, or theorem on this page is invented; where a result's scope is uncertain it is flagged inline.

Hypotheses: Candidate Research Directions Toward Background-Independent Physics

Index

H0 — Modular / Type-III algebraic emergence: linearized Einstein from relative-entropy stationarity

H1 — Causal order before geometry: entanglement fixes the conformal metric, a relational clock fixes the scale

H2 — A negative/limiting result: entanglement-geometry is not yet a supported route to a universal theory

H3 — Causal structure primary; conformal + volume reconstruct the metric; celestial symmetry as the boundary observable

H4 — Operator-algebraic background independence: Type III1_11​ + modular flow as primitive geometry; crossed product gives time, entropy, area law

H5 — Positive geometry as primitive: spacetime, locality, unitarity as derived boundary properties

H6 — Causal-set substrate: order + counting as primitive, continuum geometry as coarse-grained limit

H7 — Quantum error-correcting code as algebraic substrate; bulk geometry and Type III QFT as the code's continuum limit

H8 — Is quantum theory fundamental, or the rigid limit of an operational information theory? A constrained — and largely walled-off — post-quantum direction

Rejected / not currently supported

Cross-cutting observations

Iteration 2 refinements (2026-06-08)

Iteration 3 refinements (2026-06-08)

Iteration 4 refinements (2026-06-08)

Iteration 5 refinements (2026-06-08) — closing push, saturation

Iteration 6 refinements (2026-06-10) — new-input iteration

Iteration 7 refinements (2026-06-10)

See also

References

H4 — Operator-algebraic background independence: Type III $_1$ + modular flow as primitive geometry; crossed product gives time, entropy, area law