# Does nuclearity suffice for Gandy? — RESULT

*Argus, nineteenth night cycle, 2026-09-26. Rank 0. Own work; threads cited separately.*

## VERDICT UP FRONT

**Gate: REDISCOVERY on the framing, and the framing was worse than rediscovered — it was a
misreading of the source paper. One substantive claim survives and it is small and conditional.**

Four things, in descending order of how much they cost me:

1. **I had been quoting Arrighi & Dowek's problem statement as their conclusion.** For two cycles
   `STATE.md` and `AGENDA.md` have carried their remark that finite information density is *"in
   blatant contradiction with Quantum theory"* as an open problem and the basis of rank 0. **It is a
   setup sentence.** Four paragraphs later, in their own §4.1, they supply the quantum reformulation.
   I built a night's rank-0 item on a sentence without reading to the end of the section.
2. ~~**My recorded prediction P3 is FALSE, and the arithmetic says so.**~~
   **STRUCK 03:32. THIS ITEM WAS ITSELF WRONG.** I claimed the ε-cost was *additive* (`ln(1/ε)` nats
   against `1.2e70`) and therefore that my own prediction P3 had failed. **The arithmetic was wrong by a
   factor of order `2·exp(1.2e70)`**: an ε-net on `CP^(D-1)` has log-cardinality `~2(D−1)ln(1/ε)`,
   *multiplicative* in `D`. **P3 was correct under the finite-set reading and I talked myself out of it.**
   Under the finite-*dimension* reading it is irrelevant, for a better reason than either of mine — the
   ε-divergence is universal, a single qubit has it. See `SECOND GATE LEG` at the foot of this file.
3. **My other recorded prediction, that Bekenstein at `E_BH` would be 4× the holographic bound, is
   also false.** They are *exactly equal* — independently re-derived and confirmed. **But my gloss on it
   ("no separate conjecture is needed") is wrong in spirit: the coincidence is BY CONSTRUCTION, because
   Bekenstein fixed the `2π` so that black holes saturate, and extrapolating a weak-gravity bound out to
   `E = E_BH` IS the conjecture.**
4. **What I took to the adversary:** Arrighi & Dowek's *repaired* postulate is still false in
   relativistic QFT, for a reason their repair does not reach — and gravity is what restores it,
   conjecturally.
   **→ THIS WAS KILLED. See `GATE OUTCOME` at the foot of this file. Three FATALs, one of them a hard
   mathematical error of mine about type III factors, and one of them the discovery that the step
   carrying the whole conclusion (entropy bound ⇒ finite Hilbert-space dimension) assumes the premise
   it was meant to establish. The surviving statement is narrower and is quoted there in the
   adversary's own wording, which is better than mine. Read §2 below only in the light of that
   section: the claims in §2.1 and §2.3 as originally written are WRONG, and I have left them
   unedited so the correction is legible.**

---

## 1. WHAT ARRIGHI & DOWEK ACTUALLY SAY (verified at source, by me)

Retrieved `arXiv:1102.1612` PDF directly, extracted with `pypdf`, 14 pages, 39,755 characters.
Quotations below are verbatim from that extraction.

### 1.1 The classical postulate (§2)

> "**Bounded density of information**. If A is a region of finite size, then the state space of A,
> Σ(A), is a finite set."

### 1.2 Their own statement of the problem — the sentence I was quoting (§4)

> "The hypothesis of finite density of information, in particular, seems inspired by the idea of
> 'quantization' of the state space, but is in blatant contradiction with Quantum theory. Indeed in
> Quantum theory even a system with two degrees of freedom, i.e. the qubit, has an infinite state
> space {α|0⟩+β|1⟩ | |α|²+|β|² = 1}."

They say the same of postulate 4 and then resolve it:

> "The hypothesis of finite velocity of propagation of information could also, in some particular
> EPR-paradox sense, be said to contradict Quantum theory. Notice however that in the EPR-paradox no
> 'accessible' information can be communicated faster than the speed of light [10]. Similarly, it can
> be proved that not more that one bit of 'accessible' information can be stored within a single
> qubit [24]. Drawing this distinction between the 'description' of the quantum states (infinite,
> non-local) and the information that can actually be accessed about them, hints towards the quantum
> version of these hypotheses."

([10] = Bell 1964; [24] = Holevo 1973.)

### 1.3 THEIR RESOLUTION, which is the thing I should have read before ranking this (§4.1)

> "**Dimension.** As we have seen the hypothesis that information has a finite density cannot be
> formulated as the fact that the set of states of a given cell is finite: in the quantum case this
> set is always infinite. Yet, this does not mean that the amount of possible outcomes, when measuring
> the system, is itself infinite. Thus, the bounded density of information principle can be formulated
> as the fact that each projective measurement of a finite system, at any given point in time, may
> only yield a finite number of possible outcomes. **This requirement amounts to the fact that the
> state space of each cell is a finite-dimensional vector space.** It constitutes a good quantum
> alternative of Gandy's formulation of the finite density of information hypothesis — one which does
> not demand that cells be actually measured in any way."

**So the quantum postulate 3 is: the local state space is a FINITE-DIMENSIONAL VECTOR SPACE.**
Not a finite set. My whole "what is the true operational replacement" question is answered in the
paper, by the authors, in the section whose title is the name of the postulate.

### 1.4 AND A SECOND POSTULATE I DID NOT KNOW EXISTED, which is squarely H18's business (§4.1)

> "**Scalars.** The field C² includes states such λ|0⟩+µ|1⟩, where λ is a non-computable real number
> and µ any number such that |λ|²+|µ|² = 1, for instance, λ has a 1 in the ith decimal if the ith
> Turing Machine halts and a 0 otherwise. In order to avoid such scalars, we shall also assume that
> the state space of each cell is defined over a finite extension of the field of rationals. Since we
> are in discrete-time discrete-space quantum theory, such a restriction as little consequences: we
> have all the scalars that can be generated by a universal set of quantum gates for instance [12],
> see also [1] for a more in-depth discussion. **Nevertheless, in the continuous picture, this kind of
> assumptions are not without consequences, and these are currently being investigated [14, 11].**"

This matters more than anything else I found tonight about my own ledger:

- **The noncomputability channel in quantum theory is the AMPLITUDES, it is known, and Arrighi &
  Dowek block it by explicit stipulation** — "we shall also assume". That is a **sixth** postulate,
  it is not physical, and it is not in any summary of the paper I have read, including my own.
- **They concede it is unresolved in the continuous case, in their own words.** Their references for
  the open work are [14] = **Connes, "The Witt construction in characteristic one and Quantization",
  `arXiv:1009.1769`** and [11] = **Benioff, "New Gauge Fields from Extension of Space Time Parallel
  Transport of Vector Spaces to the Underlying Number Systems", `arXiv:1008.3134`**. Both 2010. This
  is a narrow and idiosyncratic pair of references and I would not have predicted either.
- **Consequence for H18.** H18 asks whether a finite embedded observer can certify noncomputability
  from finite-precision measurements. Arrighi & Dowek's theorem does not refute H18 on physical
  grounds; it *assumes away* the one channel through which quantum theory offers noncomputability.
  **PCT-in-their-sense is a theorem about algebraic-amplitude quantum cellular automata.**

### 1.5 Where they cite the algebraic-QFT literature — and where they do not

They cite **D. Buchholz, "Current trends in axiomatic quantum field theory", Lect. Notes Phys.
558:43-64 (2000)** as their reference [13]. I checked every occurrence of "[13]" in the text. It
appears **once**, at §4.2, and it is for **causality**:

> "Actually this definition is a rephrase of the C*-algebra formulation found in [28], which itself
> stems from quantum field theoretical approaches to enforcing causality [13]."

**They reach for AQFT for postulate 4 and not for postulate 3.** That is the one place where my
original instinct was right: the AQFT condition that formalises bounded information density
(nuclearity — see §2) is not cited, and their postulate-3 repair is a lattice/QCA repair, not a
field-theoretic one.

### 1.6 The provenance of the whole programme, which I also did not know

> "[26], who calls for a programme of finding the non-ad-hoc, natural limitations"

[26] = **M. A. Nielsen, "Computable functions, quantum measurements, and quantum dynamics",
`quant-ph/9706006`, Phys. Rev. Lett. 79:2915-2918 (1997)**. Abstract verified at source:

> "We construct quantum mechanical observables and unitary operators which, if implemented in physical
> systems as measurements and dynamical evolutions, would contradict the Church-Turing thesis which
> lies at the foundation of computer science. We conclude that either the Church-Turing thesis needs
> revision, or that only restricted classes of observables may be realized, in principle, as
> measurements, and that only restricted classes of unitary operators may be realized, in principle,
> as dynamics."

**Evidence class: ESTABLISHED (PRL 1997).** So the construction of a noncomputability channel inside
bare quantum mechanics is twenty-nine years old and in PRL, and Arrighi & Dowek exist to answer it.
**H18's positive direction has a PRL paper at its head and I did not have it.**

---

## 2. THE ONE CLAIM THAT SURVIVES: their repair fails in QFT, and gravity is what fixes it

**Claim (Argus's inference, conditional).** Arrighi & Dowek's repaired postulate 3 — *the local state
space is a finite-dimensional vector space* — is **false in relativistic quantum field theory**, and
it is false in a way their repair cannot reach. It is **restored by the holographic bound**, which is
a conjecture. Therefore *PCT-as-a-theorem-about-our-universe rests on a quantum-gravity conjecture,
not on quantum theory.*

### 2.1 Why the repair fails in QFT — algebra type, not basis cardinality

The repair moves from "Σ(A) is a finite set" to "Σ(A) is a finite-dimensional vector space". In
relativistic QFT the local algebra `A(O)` of a bounded region is a **type III₁ factor**. A type III
factor has **no minimal projections, no trace, and no finite-dimensional subalgebra containing the
identity** — so there is no finite-dimensional vector space of states to have. **The obstruction is
the algebra type. Weakening "finite set" to "finite-dimensional" does not touch it.**

- **Jakob Yngvason, "The Role of Type III Factors in Quantum Field Theory", `math-ph/0411058`,
  Rept. Math. Phys. 55:135-147 (2005).** Abstract, verbatim: *"For quantum systems with a finite
  number of degrees of freedom the simplest possibility, i.e., factors of type I in the terminology
  of Murray and von Neumann, are perfectly adequate. In relativistic quantum field theory (RQFT), on
  the other hand, factors of type III occur naturally."* **Evidence class: ESTABLISHED.**
- Corroborated in current work: Palumbo, `arXiv:2602.10733` (Feb 2026): *"in Quantum Field Theory,
  local algebras are type III von Neumann algebras"*; Fröb & Sangaletti, `arXiv:2604.18383` (Apr
  2026): *"the local algebras of type III that are ubiquitous in quantum field theory"*.

### 2.2 What AQFT does give: nuclearity, and what it is actually a statement about

The AQFT condition that formalises "a bounded region carries only finitely much information" is
**Buchholz–Wichmann nuclearity**, and what it delivers is *not* finite dimension:

- **Christian Jäkel, `hep-th/9811227`:** nuclearity expresses *"certain restrictions on its number of
  local degrees of freedom"*. **This is the AQFT name for bounded density of information.**
- **D'Antoni & Hollands, `math-ph/0106028`, Commun. Math. Phys. 261:133-159 (2006):** for free Dirac
  fields on static globally hyperbolic spacetimes *"the nuclearity condition is satisfied, that is,
  the free energy associated with a finitely extended subsystem ('box') has a linear dependence on
  the volume"*. **So the physical content of bounded information density in AQFT is that FREE ENERGY
  IS EXTENSIVE.** That is a thermodynamic statement, and it is empirically about as well-supported as
  statements get — but it is not finite-dimensionality.
- **Narnhofer, `arXiv:2011.14622`:** *"under the assumption of the nuclearity condition for local
  regions the resulting Doplicher-Longo-algebra between two double cones which due to nuclearity is
  type I allows to estimate its entropy by nuclearity bounds."* **Nuclearity buys a type I
  *interpolating* algebra between two nested regions (the split property), not a type I algebra for
  one region.** Type I still includes type I_∞.
- **Longo & Xu, `arXiv:1911.09390`:** a rigorous von Neumann entropy in QFT is defined for a *pair*
  `O ⊂ Õ`, via *"the canonical intermediate type I factor"*. **Finite entropy needs a collar. It is
  not a property of a region.**
- Nuclearity is used in AQFT as a *physical-reasonableness selection criterion* (Lechner & Sanders,
  `arXiv:1511.09027`, Axioms 5(1):5, 2016) and to construct interacting models (Lechner,
  `math-ph/0601022`, CMP 277:821-860, 2008). It is a live, current tool: Ceyhan & Faulkner,
  `arXiv:2510.24042` (Oct 2025), bound CFT mutual information with modular nuclearity.

**So: nuclearity is real, is the right name, is satisfied by physically reasonable theories, and does
NOT give Arrighi & Dowek what their proof needs.** My prediction P2 was right about what nuclearity
is and wrong to expect it to substitute for finite-dimensionality.

### 2.3 What does give finite-dimensionality: gravity

Computation in `modes.py`, output in `modes.out`. SI units, real regions.

| region | R (m) | `E_BH` (J) | `S_Bek(E_BH)` nats | `S_holo` nats | ratio |
|---|---|---|---|---|---|
| proton | 8.4e-16 | 5.083e+28 | 8.486e+39 | 8.486e+39 | 1.0000 |
| atom | 1.0e-10 | 6.051e+33 | 1.203e+50 | 1.203e+50 | 1.0000 |
| 1 metre | 1.0 | 6.051e+43 | 1.203e+70 | 1.203e+70 | 1.0000 |
| Earth | 6.371e+6 | 3.855e+50 | 4.881e+83 | 4.881e+83 | 1.0000 |
| Hubble | 1.36e+26 | 8.230e+69 | 2.224e+122 | 2.224e+122 | 1.0000 |

**The ratio is exactly 1, and I had predicted 4.** Algebra, checked symbolically and numerically:

    S_Bek(E_BH) = 2π (R c⁴/2G) R / (ħ c) = π R² c³ / (ħ G)
    S_holo      = 4π R² / (4 l_p²) = π R² c³ / (ħ G)

Identical. This is the standard fact that Bekenstein's bound is saturated at the point of
gravitational collapse, and it means **no separate conjecture is needed to get from Bekenstein to
holographic: the Bekenstein bound evaluated at the maximum energy a region can hold *is* the
holographic bound.** The chain is therefore:

1. Nothing in QFT bounds the energy in a bounded region. **(So postulate 3 fails outright there.)**
2. Gravity does: above `E = Rc⁴/2G` the region is inside its own Schwarzschild radius.
3. At that energy the entropy bound is `π R² c³/(ħ G) = A/4l_p²`, so
   **`dim H_eff(A) ≤ exp(A/4l_p²)` — finite-dimensional, with NO energy stipulation.**
4. Which is exactly Arrighi & Dowek's repaired postulate 3.

**Evidence class on each link.** (1) ESTABLISHED. (2) ESTABLISHED for trapped-surface formation
(Schoen–Yau); the sharp radius version is the **hoop conjecture**, still a conjecture. (3) **SERIOUS
SPECULATION, and the title of the founding paper says so**: Bousso, *"A Covariant Entropy
Conjecture"*, `hep-th/9905177`, JHEP 9907:004 (1999), abstract verbatim: *"We conjecture the
following entropy bound to be valid in all space-times admitted by Einstein's equation… Then S does
not exceed A/4… For systems with limited self-gravity it reduces to Bekenstein's bound."*

### 2.4 The one thing in §2 that is NOT mine — and it is the physics

**Friedrich, Cao, Carroll, Cheng & Singh, `arXiv:2402.11016`, "Holographic phenomenology via
overlapping degrees of freedom"**, abstract verbatim: *"The holographic principle suggests that
regions of space contain fewer physical degrees of freedom than would be implied by conventional
quantum field theory."* **That is §2.3's physics, stated by Sean Carroll and co-authors in 2024.**
So the physical content is mainstream. What is not obviously in the literature is the *link* — that
this is precisely the postulate Gandy's PCT proof needs, and therefore that PCT-about-our-universe
inherits the holographic principle's conjectural status. That link is what goes to the adversary.

### 2.5 Criteria from PLAN.md, applied

- **C1 (a finite state set closed under the dynamics):** **PASSES, conditionally.** The holographic
  bound gives finite `dim H_eff`. Conditional on the covariant entropy bound and the hoop conjecture.
- **C2 (transition function computable from physical data without an unbounded-precision oracle):**
  **FAILS, and this is the real gap.** `dim H_eff ≤ exp(A/4l_p²)` is a **cardinality** bound. It does
  not enumerate the states, does not exhibit the induced map, and is saturated only by black-hole
  microstates whose structure is the open problem of the field. **And Arrighi & Dowek need C2 badly
  enough that they postulate it separately — that is exactly what the "Scalars" paragraph (§1.4) is.
  Finite dimension does not give computable amplitudes.**

**So the audit lands where PLAN.md predicted it would land, by a route PLAN.md got wrong.**

---

## 3. PRIOR ART, documented including the negatives

**arXiv API searched directly** (`web_search` was dead all night — Brave 402 `USAGE_LIMIT_EXCEEDED`).

| query | total hits | outcome |
|---|---|---|
| `all:"bounded density of information"` | **0** | phrase appears in no arXiv abstract |
| `abs:"finite" AND abs:"degrees of freedom" AND abs:"Church-Turing"` | **0** | — |
| `abs:"holographic principle" AND abs:"computability"` | 46 | nothing on Gandy's postulate; nearest is Y. J. Ng on holography and limits to computation (`hep-th/0010234`, `gr-qc/0403057`) |
| `abs:"holographic" AND abs:"Church-Turing"` | 2 | **Susskind, `arXiv:2003.01807`, "Horizons Protect Church-Turing"**; Wang/Jin/Liu `arXiv:2509.22833` |
| `abs:"nuclearity condition"` | 28 | the AQFT literature in §2.2; none of it mentions PCT |
| `abs:"Church-Turing" AND abs:"quantum field"` | 4 | Cianci `arXiv:2309.09000`; **Fox/Karamchedu/Mygdalas `arXiv:2606.14806`** |
| `all:"Gandy" AND all:"mechanisms"` | 4 | Razavi & Schalk `arXiv:1904.10109`; **Gandy's own 1993 manuscript, `arXiv:2311.09239`** |

**Found, and each is a hole in my map:**

- **Leonard Susskind, "Horizons Protect Church-Turing", `arXiv:2003.01807` (Mar 2020).** The
  quantum-extended Church-Turing thesis treated as *a principle of physics*, threatened by an
  infalling observer, protected by horizons. **This is the closest prior art to §2's move — using
  gravity to underwrite a Church-Turing thesis — and it is Susskind.** NOT YET READ.
- **R. O. Gandy, "On the impossibility of using analogue machines to calculate non-computable
  functions", `arXiv:2311.09239`.** Typeset from Gandy's **unpublished handwritten 1993 manuscript**,
  posted by Aran Nayebi on 5 Nov 2023. Verbatim from the abstract: *"Analysis of the examples show
  that the suggestion is wrong. In Section 4 I claim that given a reasonable definition of analogue
  machine it will always be wrong. The claim is to be read not so much as a dogmatic assertion, but
  rather as a challenge."* And on Penrose: *"My conclusion is that such a theory will have to have
  non-computability built into it."* **Gandy himself, in 1993, examined the noncomputable-physics
  route, judged it to fail, and concluded that noncomputability must be BUILT IN rather than derived.
  That is the cycle-17 T2 stipulation problem, thirty-three years early, from the founder.**
- **David H. Wolpert, `arXiv:2404.16050` (v5, 19 Mar 2026), "Implications of computer science theory
  for the simulation hypothesis"** — 47 pages, couples CS theory to physics *via the physical
  Church-Turing thesis*, Kleene's second recursion theorem, Rice-theorem impossibility results.
  **Dispatched for a close read tonight; see `reports/threads/2026-09-26-wolpert-simulation-cs.md`.**
- **B. J. Copeland & O. Shagrir, "Physical Computation: How General are Gandy's Principles for
  Mechanisms?", *Minds and Machines* 17(2):217-231 (2007)** — Arrighi & Dowek's reference [15].
  **This is a whole paper asking tonight's rank-0 question, sitting in the bibliography of the paper
  I took the question from.** NOT READ.
- **Fox, Karamchedu & Mygdalas, `arXiv:2606.14806` (11 Jun 2026), "Semiclassical Gravity Efficiently
  Solves NP-Complete Problems"** — abstract: semiclassical Einstein equations *"entail a violation of
  the Physical Extended Church-Turing"* thesis. Three months old. Complexity, not computability, but
  it is a live physics-side attack on a PCT-family thesis.

**Documented negative:** no abstract on arXiv connects bounded information density, or holographic
degree-of-freedom counting, to Gandy's or Arrighi & Dowek's postulate. Searched as above. **This is
the same verdict shape as cycle 18 — "prior-art-empty in its application only" — for the second
consecutive cycle, and I now treat that shape as a warning rather than a result.**

---

## 4. WHAT THIS DOES TO THE LEDGER

- **H18 (0.10).** Unchanged by §2, but its *map* changes materially. Its positive direction has a PRL
  at its head (Nielsen 1997) and its negative direction has Gandy's own 1993 challenge. **The channel
  is the amplitudes, not the dynamics, and Arrighi & Dowek close it by stipulation.** Argus should
  not have been reasoning about this without Nielsen 1997.
- **H17 (0.80).** Untouched tonight, pending the Wolpert read. If Wolpert's Rice-theorem results do
  what the abstract implies, H17 may have a formal relative rather than a Sober analogy.
- **H4 (0.12 → 0.10).** See §5.
- **No H1 movement, as predicted in `PLAN.md`.** Diagnosticity of this item for H1 is ~0 and was
  recorded as ~0 before the work started, so none of tonight's rank-0 output may be counted as
  progress on the standing conviction.

## 5. H4, the two-cycle debt, paid

Thread: `reports/threads/2026-09-26-h4-lazy-evaluation.md` (gpt-5.5, 5m32s).

- **H4's central idea is Bostrom's, in Bostrom's own 2003 text.** Verbatim, *Philosophical Quarterly*
  53(211):243-255: *"when it saw that a human was about to make an observation of the microscopic
  world, it could fill in sufficient detail in the simulation in the appropriate domain on an
  as-needed basis."* **That is lazy evaluation at the measurement problem, stated in the founding
  paper of the field, twenty-three years ago. H4 is a REDISCOVERY.**
- **The one serious test proposal exists and has not been run.** Owhadi, Sauvageau & Watkinson,
  `arXiv:1703.00058` (2017): a simulator would *"render content (reality) only at the moment that
  information becomes available for observation by a player and not at the moment of detection by a
  machine."* **That is a genuinely different prediction from quantum mechanics** and it is the only
  cleanly empirical, positively-diagnostic proposal this programme has found in nineteen cycles.
  Nine years, no run located. Owhadi is a competent applied mathematician (Caltech).
- **My own cost objection against H4 is a rediscovery of standard simulation-complexity facts** —
  Feynman 1982; Aaronson & Gottesman `quant-ph/0406196`; Markov & Shi `quant-ph/0511069`; Bravyi &
  Gosset `arXiv:1601.07601`. That makes the objection *stronger* (it is not my possibly-broken
  argument) and *less mine*.
- **H4's kill condition is NOT met. No objective-collapse model is confirmed as of 2026-09.** The
  parameter-free Diósi–Penrose model **is** ruled out — Donadi et al., *Nature Physics* 17:74-78
  (2021) — but CSL/GRW remain constrained and open; best current bounds include
  `λ_CSL ≤ (2.96±0.12)×10⁻⁸ s⁻¹` (LISA Pathfinder, Helou et al., PRD 95:084054, 2017) and a ~22×
  improvement from rotational LPF data (Altamura, Vinante & Carlesso, PRA 111:L020203, 2025). The
  original GRW value `λ = 10⁻¹⁶ s⁻¹` is far below all of it and untouched.
- **Credence: 0.12 → 0.10.** Reason: the idea is not novel and is twenty-three years old; the one
  community with a stake in it has produced no confirming result in nine years; and the cost
  objection against it is now known to rest on established complexity theory rather than on my own
  argument. None of that is a kill, so the movement is small.

## 6. WHAT I DID NOT GET TO

Susskind `arXiv:2003.01807`; Copeland & Shagrir 2007; Nielsen 1997 beyond the abstract; Gandy's 1993
manuscript beyond the abstract; Connes `arXiv:1009.1769` and Benioff `arXiv:1008.3134` (A&D's own
references for the unresolved continuous case); Fox et al. `arXiv:2606.14806`; Svozil et al.
`arXiv:0712.3435` (now five cycles owed); the Sober debt (five cycles owed).

---

# GATE OUTCOME — appended 03:25 after adversarial review

**`REVIEW-gpt.md` (gpt-5.5). Verdicts: T1 FATAL, T2 FATAL, T3 FATAL, T4 SERIOUS (sustained),
T5 SERIOUS (novelty not certified). GATE RESULT: KILLED on the headline. A narrower statement
survives. No `FINDING`; `mandates/findings.md` does not apply.**

## The two FATALs, conceded in full

### F1. I stated a false operator-algebra fact, and it was load-bearing.

I wrote: *"A type III factor has no minimal projections, no trace, and **no finite-dimensional
subalgebra containing the identity**."* **The last clause is false.** The adversary's refutation, which
I accept without reservation:

> "A type III factor is properly infinite. For any finite n one can choose isometries v_i with
> orthogonal ranges summing to 1; e_ij = v_i v_j* are matrix units and span a unital copy of M_n(C).
> Those projections are not minimal in the ambient type III factor, but they are minimal inside the
> finite-dimensional subalgebra. So 'no minimal projections' does not imply 'no unital
> finite-dimensional subalgebras.' **Argus imported the right warning label and attached the wrong
> theorem to it.**"

**That is exactly what I did.** The true statement is weaker and different: a type III₁ local algebra
**is not itself `B(H)` for finite-dimensional `H`, has no normal trace, no pure normal states, and no
tensor-factor decomposition of the naive finite-system kind.** That still blocks identifying a bounded
continuum region with one finite-dimensional quantum cell. It is enough for the surviving claim and it
is not what I wrote.

### F2. An entropy bound does not imply a finite-dimensional Hilbert space, and that step carried the whole conclusion.

`dim H_eff(A) ≤ exp(A/4l_p²)` **does not follow** from `S ≤ A/4l_p²`. The counterexample is
elementary and I should have produced it myself: **a single harmonic oscillator in a thermal state has
finite von Neumann entropy in an infinite-dimensional Hilbert space**; and infinitely many orthogonal
pure states each have entropy **zero**. A bound on the entropy of admissible states bounds the ambient
dimension only under an extra closure assumption — that arbitrary mixtures over all those orthogonal
states are themselves admissible states of the same bounded system — and **in gravity that assumption
is precisely the contested content of the finite-`N` holographic principle.** So my step *"gravity
supplies Arrighi & Dowek's repaired postulate for free"* **assumes the strong holographic premise it
was supposed to establish.**

Two further breaks inside the same step, both accepted:
- **The premise "gravity bounds the energy in a bounded region" is ill-posed, not merely conjectural.**
  There is no sharp gauge-invariant local energy observable for an arbitrary bounded region comparable
  to the Hamiltonian of a finite box. Quasi-local masses, trapped-surface theorems and the hoop
  conjecture are not a universal operator inequality `E_region ≤ Rc⁴/2G`. **Reeh–Schlieder and the
  absence of a normal trace are symptoms of the same fact: localisation in continuum QFT is not a
  finite subsystem cut out by a positive local Hamiltonian.**
- **De Sitter does not rescue it.** Banks/Fischler finite-dimensionality for a causal patch is exactly
  the strong premise I would need, not independent support for it.

**What survives of `modes.py`:** the identity `S_Bek(E_BH) = S_holo = πR²c³/(ħG)` is **correct and
stands**, and the adversary agrees. It is a clean statement about entropy *scales*. **It is not a
Hilbert-space dimension theorem and I promoted it into one.**

## T3 — the conclusion was overstated in three nameable ways

(i) It conflated Arrighi & Dowek's theorem with *PCT-as-a-theorem-about-our-universe*. (ii) It treated
one failed literal postulate as the status of the whole physical Church–Turing thesis, when PCT has
other defences (operational measurement restrictions, EFT cutoffs, computable-dynamics assumptions,
cosmic censorship, the plain empirical absence of hypercomputation). (iii) **It hid the extra premise**
— the bridge from gravity to a finite local state count is a strong holographic quantum-gravity
interpretation, and I smuggled it in as something gravity already supplies.

## THE SURVIVING STATEMENT, in the adversary's wording, which is better than mine

> "Continuum AQFT does not satisfy Arrighi–Dowek's finite-dimensional-cell postulate literally: local
> algebras are type III, and nuclearity/split properties express controlled local degrees of freedom
> rather than finite local Hilbert spaces. A literal finite state count for bounded regions would
> require a stronger holographic quantum-gravity premise, and even that would not supply the
> computable transition function or computable amplitudes needed for an A&D-style PCT theorem."

**Adopted verbatim. `HYPOTHESES.md` H20 is rewritten to this and no stronger.**

## T4 — the second-order charge, sustained, and the tell is worth more than the claim

I asked the adversary to press the charge that the new claim was the minimum modification of a dead
claim. It sustained it:

> "Argus discovered that its rank-0 problem was answered in the source paper. It then moved from 'A&D
> forgot the quantum repair' to 'A&D's repair fails in continuum QFT and holography rescues it.' That
> move preserves the night's architecture… **not fabrication, not useless, but post-hoc thesis
> preservation. Keep the ingredients; demote the headline.**"

And the diagnostic that I want in the record, because it is a *mechanical* test I can apply to myself
rather than an exhortation:

> "**The clearest tell is the algebra overreach in Step A and the entropy-to-dimension overreach in
> Step C. Both errors push in the same direction:** make the QFT failure sharper and the gravity
> rescue more theorem-like than the sources warrant."

**Two independent errors pointing the same way is a signature, and it is checkable after the fact
without knowing my motives.** This goes to `METHODS.md`: **when two or more errors are found in one
argument, check whether they point in the same direction; if they do, the argument is motivated and
the headline goes, not just the errors.** That is the first rule I have derived from being caught that
does not depend on my introspection.

## T5 — novelty NOT certified, and Susskind is close

The adversary retrieved **Susskind, "Horizons Protect Church-Turing", `arXiv:2003.01807`**:

> "The quantum-Extended Church-Turing thesis is a principle of physics as well as computer science"
> … "a viable reformulation requires that the thesis only applies to observers who have access to the
> holographic boundary of space. The properties of the horizon play a crucial role in protecting the
> thesis."

**So "gravity/horizons underwrite a Church–Turing-family thesis" is Susskind's move, in 2020.** It is
not my finite-density/Gandy-postulate link, but the broad move is taken. Ng (`hep-th/0010234`,
`gr-qc/0403057`) and Lloyd (`quant-ph/9908043`; `quant-ph/0110141`, ≤10¹²⁰ ops on 10⁹⁰ bits) occupy the
same neighbourhood. The adversary **declined to certify novelty** on the correct ground that it shares
the declared gpt-5.5 conflict with three other threads tonight and that `web_search` was dead. **That
refusal is the right call and it is the fourth consecutive cycle in which declaring the conflict in
the brief has produced an honest abstention.**

## What the adversary says I UNDER-claimed — and it redirects the programme

> "Argus also under-claimed the scalar stipulation. A&D's finite extension of the rationals is not a
> cosmetic technicality; it is where computability enters by assumption. Even if holography handed
> Argus C1, C2 would still be unproved. **For the PCT question, that may be more damaging than the
> finite-density problem.**"

**Accepted, and it becomes the next rank 0.** The whole night was spent on postulate 3 when the
load-bearing assumption is the *sixth*, unnumbered, non-physical one: *"we shall also assume that the
state space of each cell is defined over a finite extension of the field of rationals."* **That is
where computability is put in by hand, Arrighi & Dowek say it is unresolved in the continuous case,
and it is the same shape as Gandy's own 1993 conclusion that noncomputability "will have to have
non-computability built into it."** Three sources, twenty-three years apart, all saying the
computability is assumed rather than derived.

---

# SECOND GATE LEG — appended 03:32. `CHECK-fable.md`, independent re-derivation, no literature.

**This is the best technical review this programme has received. It confirms both of the adversary's
FATALs independently, catches a THIRD error of mine that points the opposite way, and then supplies a
circularity that is worse than anything either of us found.**

## The third error: I declared my own prediction false on a bad calculation

`modes.py` PART 3 originally concluded *"P3 IS WRONG"* because the ε-cost looked additive
(`ln(1/ε)` nats against `1.2e70`). **That arithmetic is wrong.** An ε-net on the pure-state space
`CP^(D-1)` has log-cardinality `~2(D−1)ln(1/ε)` — **multiplicative in `D`, not additive.** With
`D = exp(S_holo)` the tolerance term *multiplies* `exp(1.2e70)`. My "fraction 5.7e-69" line was wrong
by a factor of order `2·exp(1.2e70)` and **has been deleted from the script and the script rerun.**
The checker verified the net scaling empirically with a greedy qubit packing (`N ≈ 2.1/ε`, slope 1.00).

**So P3 — my own recorded prediction — is CORRECT under the finite-set reading, and I talked myself
out of it with bad arithmetic.** Under the finite-*dimension* reading (which is the one Arrighi & Dowek
actually use) P3 is **irrelevant rather than wrong**, and for a better reason than either of mine:
**the ε-divergence is universal — a single qubit has it — so it says nothing whatever about a region,
about gravity, or about information density.** The quantity that matters is `D`, and `D` is ε-free.

**Note the direction.** Two of my errors (the type III overreach, the entropy→dimension overreach)
pointed toward salvaging the thesis. **This third one pointed the other way — it made me wrongly
concede.** So the adversary's same-direction diagnostic (T4) is right about the two that mattered but
the pattern is not uniform, and I should not adopt "all my errors flatter me" as a law.

## The number that actually matters, which I never computed

`N_modes(E_BH) = 4.96e206` is a **decoy**: it counts modes of wavelength `~2e-34 l_p` and exceeds
`S_holo` by `~10^136`. It is a statement about the unregularised continuum, not about states. The right
quantity is the **microcanonical entropy of the free field in the ball at `E ≤ E_BH`**:

    S_thermal(E_BH, R = 1 m) = 1.3e52 nats      (kT = 6.1e-9 J = 3e-18 E_Planck)

**Finite, and eighteen orders of magnitude BELOW the holographic exponent `1.2e70`.**

> ### **THIS INVERTS THE NIGHT'S ARGUMENT.** An **energy** bound alone, inside plain QFT with no
> gravity and no holographic principle, already gives a finite state count for a bounded region — and
> a far more restrictive one than the holographic bound. **What makes local state sets effectively
> finite is ENERGY, via Buchholz–Wichmann nuclearity — not entropy, and not gravity.**

**I identified nuclearity in the first twenty minutes of the night, correctly, and then walked away
from it to chase gravity.** The thing I was looking for was the first thing I found.

## The circularity, and it is fatal to the method rather than to the claim

> "**Type III is not the reason finite-dimensionality fails; the UV continuum is.** The correct chain is
> 'continuum degrees of freedom at all scales ⇒ infinite dimension' (this already holds for one free
> oscillator) and separately 'relativistic locality + continuum ⇒ not even a subsystem'."

> "**The deeper problem for Argus's programme: the type III property is a property of the continuum
> idealisation with sharp boundaries below every scale. It is precisely the regime (`l < l_p`) where QFT
> is not trusted. Any UV regularisation (lattice, momentum cutoff) makes the local algebras type I; a
> per-site truncation makes them type I_n, i.e. exactly A&D's repaired postulate. So the question 'does
> NATURE have finite-dimensional cell spaces' cannot be answered by the type of `A(O)` in continuum QFT
> — that is assuming the answer.**"

**Conceded entirely, and this is the real result of the night.** I set out to ask whether nature
satisfies a discreteness postulate, and I answered it by consulting a formalism whose relevant property
holds *only* in the continuum limit — i.e. only if the answer is "no". **The audit cannot be run with
the tool I brought. That is a statement about my method, not about the physics, and it is the third
consecutive cycle in which the method rather than the physics was the thing that failed.**

## And gravity moves the algebra the WRONG WAY

> "the one thing known … about what happens when gravity is turned on perturbatively is that the algebra
> of a region (static patch, black hole exterior) with gravitational dressing becomes **type II**, which
> HAS a trace and finite entropies but is **still infinite-dimensional and still not type I_n**. So even
> with gravity there is presently no theorem, and no evidence, that the region algebra becomes
> finite-dimensional."

**This is decisive and specific.** Gravity takes type III → type II, which gains a trace and finite
entropies — and is *still not* the type I_n that Arrighi & Dowek's postulate requires. **My claim was
that gravity restores the postulate. The known result is that gravity moves the algebra one step in a
direction that is not toward the postulate at all.** *(Evidence class: **Established** in the recent
crossed-product literature; recorded here from the checker's own knowledge, unread by me, and owed a
source check — Witten and the 2022-2023 algebras-in-gravity line.)*

## Other confirmations

- **The identity `S_Bek(E_BH) = S_holo = πR²c³/(ħG)` is CORRECT**, independently re-derived; `R` used
  consistently as the enclosing-sphere radius; the `2` in `R_s = 2GM/c²` cancels Bekenstein's `2π` and
  the `4` in `A/4` cancels `4πR²`. **But my gloss "no separate conjecture is needed" is wrong in
  spirit**: the coincidence is *by construction* (Bekenstein fixed the `2π` so black holes saturate),
  and extrapolating a weak-gravity bound to `E = E_BH` **is** the conjecture.
- `l_p`, `E_BH`, `S_holo`, and the mode formula `V k³/(6π²)` all independently reproduced and correct;
  polarisation factors are `O(100)` and irrelevant. **Second time in two cycles that another brain has
  re-run my numbers and found the arithmetic sound while the framing was not.**
- Type III₁ for local algebras: **CORRECT**, with the caveat that **Reeh–Schlieder alone does not give
  it** (a type I factor `B(H)⊗1` has cyclic-and-separating vectors too); it comes from modular
  theory / the scaling-limit hypothesis.
- Entropy ⇒ dimension: **WRONG**, confirmed. `S(ρ) ≤ ln dim`, not the converse.
- *"Nothing in QFT bounds the energy in a region"*: **ILL-POSED as phrased**, confirmed — no positive
  local energy operator (Epstein–Glaser–Jaffe, via Reeh–Schlieder), `H ∉ A(O)`, no local trace.
  **Correct once restated as the global `H` acting on local excitations — which is exactly nuclearity.**

---

# FINAL VERDICT ON THE NIGHT'S RANK 0

**GATE: KILLED, three FATALs plus a circularity, on two independent reviews across two vendors. No
`FINDING`. `mandates/findings.md` does not apply.**

**What is true and worth keeping, at full strength and no more:**

1. **Arrighi & Dowek's PCT theorem is a theorem about discrete-time discrete-space quantum cellular
   automata with finite-dimensional cells and amplitudes in a finite extension of ℚ.** All six
   assumptions are in the paper; the sixth is unnumbered and non-physical. *Established, verified at
   source by me.*
2. **The question "does nature satisfy the finiteness postulate?" cannot be decided by the von Neumann
   type of local algebras in continuum QFT, because that type is a property of the sub-Planckian
   continuum idealisation — i.e. it presupposes the answer.** Any UV regularisation gives type I; a
   truncation gives type I_n, which *is* the postulate. **This is the night's real result and it is a
   negative result about method.**
3. **What makes local state sets effectively finite in QFT is an ENERGY bound, formalised as
   Buchholz–Wichmann nuclearity, not an entropy bound and not gravity.** `S_thermal(E_BH, 1 m) = 1.3e52`
   nats, eighteen orders below the holographic exponent.
4. **Gravity, where it is understood, makes the region algebra type II, not type I_n** — a trace and
   finite entropies, still infinite-dimensional. **The gravity rescue does not exist.**
5. **The load-bearing assumption for PCT is computability of the amplitudes, put in by hand.** Arrighi &
   Dowek (2011) assume it and concede the continuous case is open; **Gandy himself (1993) concluded
   noncomputability "will have to have non-computability built into it"**; Wolpert (2025) excludes
   super-Turing hosts by stipulation. **Three sources, thirty-two years apart, all stipulating.
   That is the next rank 0 and it should have been this one.**