Reviewer3 (AI review)

October 4, 2026

AI Review (reviewer3.com)

Comments (8)

1. Bipartite assumption prevents extension to the sphere — Critical

The base is a trivalent network on S 2 , the base of the Hopf fibration S 3 → S 2 , locally a honeycomb with sublattices A and B; machine runs use a honeycomb torus; the ribbon linking of App. A, rule 10, is computed on a Goldberg sphere, which is not bipartite, so there only the ribbon structure is used, not the strand labels. Each vertex is a six-way

The abstract and introduction claim that the model routes causal flow on the base of the Hopf fibration (the sphere $S^2$). However, the definition of the two conserved strand densities ($n_1$ and $n_0$) and the resulting $\psi$ measure—which are essential for reading out charge and electromagnetism—relies on the existence of a bipartite sublattice division (sublattices A and B). Because any trivalent discretization of $S^2$ (such as a Goldberg sphere) must contain exactly 12 pentagons by Euler's formula, it cannot be bipartite. Consequently, the global strand labels and the $\psi$ measure cannot be defined on $S^2$, limiting the scope of the electromagnetic and charge readouts to bipartite surfaces like the flat torus.

2. Unexplained exception in collective speed scaling — Minor

free: inside matter the measured one-tick collective speed c kin , current over density, is c / √ 1 + σ , with σ the buffered share of §4.4 and c 2 kin (1 + σ ) = 0 . 12 – 0 . 13 ≈ STEP 2 on 29 of 30 substrate and rule

The authors state that the relation between the collective speed and the buffered share holds on 29 of 30 substrate and rule cases. However, they do not discuss or explain the single case that did not satisfy this relation. To ensure the scope and limitations of the claim are fully understood, the authors should identify this exceptional case and explain why the relation failed or deviated, or temper the claim to reflect this boundary.

3. Inconsistent spatial isotropy in light bending derivation — Minor

At a readout point, the time share puts a tick on the distant clock at

The derivation of the post-Newtonian parameter γ = 1 and the resulting light bending angle relies on an isotropic effective speed of light. However, the physical justification provided states that the depth share only scales the radial proper length, which implies an anisotropic spatial metric where only the radial component is modified. An anisotropic metric would yield a different light bending angle than the standard isotropic General Relativity value. The authors should resolve this inconsistency by either justifying how the transverse spatial components are also scaled, or recalculating the bending angle using the anisotropic metric implied by the radial-only depth share.

4. Origin dependence of C-odd moment in sub-regions — Minor

in lattice constants. From the routing tables alone, each A-type reversed vertex adds exactly − √ 3 and each B-type + √ 3 , with no net C-odd momentum, so the result does not depend on the origin.

The authors derive that the C-odd moment of any region is exactly given by a simple formula independent of the origin, asserting this holds because there is no net C-odd momentum. However, for an arbitrary sub-region of a larger network, boundary currents and net flows crossing the boundary can generate a non-zero net momentum. This would make the calculated moment dependent on the choice of origin, violating the claimed exactness and independence. The authors should specify the boundary conditions or symmetry requirements under which this formula holds for a sub-region, or restrict the claim to closed manifolds.

5. Vague mathematical behavior of fractional charge readouts — Minor

A single reversed vertex would give ± 1 3 , which no line bundle on S carries: in readout space a fraction has no Gauss-law reading, and it exists only on the base, as the point holonomy (2). Only integer charges therefore have a Coulomb field: the particles of §3.6 carry

The authors assert that fractional charges (such as a single reversed vertex) have no Gauss-law reading in readout space because no line bundle on S can carry a fractional degree. However, the incidence transform is a mathematically defined integration over the sky of an observer, which must yield some concrete output when applied to a configuration with a single defect. Simply stating that it has 'no reading' is mathematically vague. The authors should explicitly calculate and describe what the incidence transform actually produces for a single reversed vertex, and explain how this output is mathematically regularized or discarded in the readout theory.

6. Discrepancy in numerical verification of exact holonomy — Major

On the machine, the frame turning of all 80 contractible cycles around placed closures equals the enclosed sum, and for random discs of radius 6–16 the enclosed holonomy follows it with slope 1.00–1.02.

The authors state that the discrete Stokes theorem for point curvature is exact at every scale. However, the numerical verification on random discs of radius 6–16 yields a slope of 1.00–1.02 rather than exactly 1.00. If the relation is a topological identity, there should be no systematic deviation in the slope. The authors should explain whether this deviation arises from the discrete boundary definition of the random discs or from numerical approximation, and clarify how the boundary holonomy is computed to resolve this discrepancy.

7. Circular derivation of boost covariance and Lorentzian signature — Major

Every observer’s split of the axis is even, since no split can see

The derivation of the Lorentzian signature and boost covariance relies on the postulate that a change of observer must preserve the front lines r = ±ct. By assuming that the transformation must keep these front lines invariant, the Lorentzian metric structure is effectively assumed from the outset rather than emerging naturally from the underlying 2D network dynamics. To make the derivation robust, the authors should show how the invariance of the front lines arises from the physical properties of the network or the incidence transform itself, rather than defining the change of observer to satisfy this condition.

8. Insufficient statistical sampling for localization scaling — Minor

Table 3: Strip transfer on uniform substrates: η = ξ W / W from the smallest to the largest width measured ( W = 16 – 256 , eight seeds per point, two to four at

The phase localization and critical scaling are analyzed using a very small number of disorder realizations, specifically eight seeds per point, and only two to four seeds for the largest system size of W = 256. In statistical studies of localization transitions, such small sample sizes introduce significant statistical noise, which can severely affect the reliability of the estimated critical boundary and the conclusion that no stable diffusive phase exists. The authors should increase the number of independent seeds, especially for larger system sizes, to provide robust error bounds and confirm the scaling behavior.

Author response

Thank you for the close reading. Brief replies, point by point.

1. Sphere. Right: by Euler's formula a trivalent sphere cannot be a bipartite honeycomb everywhere; like a fullerene it needs a fixed number of base defects, around which the strand and arm labels acquire holonomy. This does not conflict with the machine, whose network is made of defects everywhere: on a base of cosmological size their total deficit, 4π, is spread over the whole sphere, and locally the network is the flat honeycomb that the torus runs sample. The next version says this in §1.1, and the abstract will state that the runs use periodic patches of the base.

2. The exception. Accepted. It is p = 0.6, ρ = 0.3 under the earlier per-axis buffer rule (0.101), and the count is 31, not 30; the next version names it.

3. Light bending. A change of the radial component alone is the Schwarzschild metric in areal coordinates; in isotropic coordinates the same field takes the isotropic form, and the bending, being independent of coordinates, is the same. The next version adds this step.

4. C-odd moment. The moment is computed from the routing tables, not from the flow, so no boundary current enters. The exact statement is per vertex: each reversed vertex contributes ∓√3 with no net C-odd momentum, and a region's moment is the sum over its vertices. The next version says "per vertex".

5. Fractional charge. In readout space a loop around a point is contractible, and the flux through a closed surface is the degree of a line bundle, an integer; ±1/3 has no reading there and exists only as a point holonomy on the base. The next version separates whether an isolated vertex can exist (it can) from whether it can be read (only at the scale at which it closes, with two other vertices of its type, into a unit of charge). The ψ channel itself has not been evaluated on the machine; this is listed as open.

6. Slope 1.00–1.02. The exact statement is the one for the 80 cycles, with no deviation. The slope 1.00–1.02 comes from a regression of route holonomy against the number of defects inside a geometric disc, whose boundary differs from the region a route encloses; the next version labels it as a regression.

7. Lorentzian signature. The invariance of the front lines is not assumed. It follows from the even split of the time–depth axis, which follows from a machine fact, that the two ways of assigning identities at a collision swap leave the same occupancy, and the principle that no reading may depend on that assignment. Version 1 graded the even split as a postulate, which made the chain look circular; the next version grades it as derived and adds a claim-dependency table.

8. Sampling. The conclusion that no stable diffusive phase exists rests on the conductance of coarse cells (L = 1024, b = 8–256), not on the strip-transfer table, whose boundary is a finite-size contour. More seeds at W = 256 are left open for further work.

xPeerd (AI review)

October 4, 2026

AI Review (xPeerd — DAReview)

Summary

This manuscript, On a Routing Network of Causal Flow: Incidence Transform from the Hopf Bundle to Four-Dimensional Spacetime, proposes a deterministic routing machine on a trivalent network and argues that, after coarse-graining plus an incidence transform, the model yields readouts interpreted as light, gravity, and charge in four-dimensional spacetime. The paper’s main strengths are its internal ambition, unusually explicit claim-labeling scheme ([E], [M], [D], [P]), and some careful separation between exact machine properties, measured simulation outcomes, derived consequences, and postulates. The abstract and introduction clearly state the intended program and architecture of the model.

Its main weaknesses are more serious. The manuscript repeatedly converts postulates into physically weighty conclusions without sufficient external validation, reproducibility closure, or demonstrated uniqueness of the readout. Several central bridges—from machine dynamics to 3D isotropy, from time–depth equipartition to Lorentzian structure, from delay offsets to gravity, and from frame holonomy to observable charge/interference—remain asserted, schematic, or explicitly deferred as open problems. The paper itself admits that some key computations are “not evaluated on the machine” and that several derivations are still open.

Potential Major Revisions

1. Serious concern about the evidential gap between formal machine properties and physical interpretation

The central interpretive move is extremely strong relative to the evidence shown. The manuscript states that “light, gravity and charge in four-dimensional spacetime are nonetheless read out of it” and that “the reading is one transform.” Yet many of the physically consequential steps are labeled [P], meaning they are postulates or framework identifications rather than results. The manuscript itself defines [P] as “a physical identification made by the framework, stated as a definition rather than a result,” and says a claim resting on [P] alone is written as a definition.

This creates a major epistemic problem: the paper often presents high-level physical consequences in prose that sound evidentially stronger than their grading permits. For example, the manuscript claims that “the two shares are equal, which gives the four-dimensional reading its Lorentzian signature and fixes the post-Newtonian parameter at 1+γ = 2.” But the equality of shares is not empirically derived here; rather, the paper later says, “Third, each collision swap credits half to time and half to depth: time–depth equipartition,” explicitly marking this as [P]. This means the emergence of Lorentzian signature and γ = 1 is not established by the machine alone, but depends on an adopted interpretive rule. That distinction is crucial and should be much more prominently and repeatedly constrained in the main claims.

2. Reproducibility concern: several core readouts remain schematic or unevaluated

A number of key outputs are not fully computed on the machine. The paper states, “This step is schematic: the counts are not evaluated line by line here.” It further says, “This step too is schematic: the winding exists only in the L2 reading, and Rψ has not been evaluated on the machine.” These are not peripheral omissions. They directly affect the central claimed route from microscopic dynamics to gravitational and charge-type readouts.

Likewise, the open-problems section acknowledges deferred tasks including “evaluating the counts on each flow line and Rψ on the machine,” “the scale of G,” and “the stationary-phase concentration of the walk sum.” Since the incidence transform is the paper’s unifying mechanism, leaving its most physically distinctive ψ-channel only partially formalized weakens the manuscript substantially. At minimum, the authors should either reduce the strength of the headline claims or provide completed computations for these readouts.

3. Insufficient demonstration that the proposed transform is unique, necessary, or physically non-arbitrary

The manuscript’s physical story depends on a particular incidence transform with two measures, one positive and one signed on a double cover. But the paper does not adequately show that this transform is uniquely forced by the machine rather than selected because it yields desired analogies. For example, in §5.3 it introduces the two channels “Rρf = ∫Σo fρ(ℓ) dµo(ℓ), Rψf = ∫Σo fψ(ℓ)e^{iθ(ℓ;o)} dµo(ℓ)” and then assigns gravity-type readings to ρ and charge-type readings to ψ.
However, the manuscript does not adequately compare this mapping against plausible alternative transforms or gauge choices, nor does it show why the chosen observer/readout architecture is physically privileged.

This is especially problematic because the manuscript frequently moves from structural possibility to ontological interpretation. For a foundational physics paper, the burden is not merely to show that one mathematically coherent reading exists, but that the reading is robust, constrained, and not merely a selective interpretive overlay.

4. Major concern about the use of postulates to obtain relativistic-gravitational consequences

The manuscript writes: “Light bending and Shapiro delay” with “γ = 1, the value of general relativity for the post-Newtonian light-bending parameter, following from the equipartition postulate, not from a measurement.” This sentence is admirably candid, but also reveals the central difficulty: the paper’s apparent agreement with GR in this sector is not a nontrivial prediction here. It is produced by the chosen postulate.

Similarly, the text states, “The potential is the static delay offset counted in periods of that clock” and “With the static offset of §4.4 and Φ = GM/(c²r), the clock cancels from the coupling.” Yet the identification of delay with gravitational potential is itself framework-imposed, not independently validated against a broader class of observables. The manuscript should therefore sharply distinguish: a) exact machine results, b) numerically measured coarse-grained regularities, c) analytic derivations conditional on assumptions, d) physical identifications imposed by interpretation.

At present, these categories are labeled, but the narrative still risks overstating what is actually shown.

5. The manuscript signals unresolved foundational issues that undermine present claims of completeness

The open-problems section is extensive and touches central rather than peripheral matters. Examples include: • “The phase renormalization of the static half” • “The coupling of the two halves” • “Readout computations deferred” • “The spin reading”

This is not merely a healthy research agenda; it indicates that several pillars of the proposed theory are still unfinished. In particular, the coupling between static and transport halves appears essential to the paper’s physical claims, yet the manuscript admits, “A derivation, rather than a fit, is open.” A paper may certainly present a partial program, but then its title, abstract, and conclusion should more explicitly reflect that it offers a conjectural readout framework rather than a robust derivation of spacetime physics.

6. Potential paper-mill/fabrication screening outcome: low risk of generic fabrication, but elevated risk of theory inflation

I do not see classic paper-mill traits such as citation spam, meaningless terminology chains, or obviously inconsistent sectioning. The manuscript is internally structured, technically stylized, and repeatedly flags open issues. That argues against routine fabricated text.

However, I do see a different risk profile: theory inflation. The paper makes very large claims—emergence of light, gravity, charge, Lorentzian signature, equivalence, light bending, Shapiro delay, charge as half vortices—while many critical bridges remain postulated, schematic, or deferred. For instance, “The half vortex is observable” is followed by an interference argument derived from the assumed ψ-phase interpretation rather than a completed machine-level demonstration. This is not fraud by itself, but it is a warning sign for overextended interpretation relative to demonstrated evidence.

7. The model’s empirical relevance remains weakly tested

The manuscript reports several measured internal behaviors, such as “We find no stable diffusive phase,” logarithmic fall of conductance, and a crossover/localization interpretation reminiscent of Chalker–Coddington physics. These are interesting within the machine. But the connection from these simulation observables to real physics remains mostly analogical. Agreement with known forms is often invoked at the level of shape or interpretation, not stringent quantitative prediction. For a paper making foundational claims, this is insufficient. The manuscript needs a dedicated section listing concrete falsifiable predictions that differ from GR, gauge theory, causal set approaches, lattice models, or topological toy models.

8. AI-content disclosure raises editorial and authorship questions that require careful handling

The manuscript openly states that “Claude models implemented and ran the machine and its analyses, carried out derivations, audited claims against the machine’s readouts and retrieved related mathematics and physics,” and also “corrected English grammar and wording, improved fluency, drew the figures and edited the LATEX source.” This is unusually extensive AI involvement. The author states that all content was reviewed and that the author takes responsibility, which is helpful. However, given the paper’s already interpretive and foundational nature, this level of AI participation increases the need for independent reproducibility assets, transparent code auditability, and human-verifiable derivation appendices. Otherwise, the manuscript risks containing elegant but insufficiently stress-tested reasoning chains.

Potential Minor Revisions

1. The abstract and introduction should more consistently distinguish results from postulates

The abstract states that the transform “reads both halves as light, gravity and charge in four-dimensional spacetime.” Because some of these readings depend on [P]-level identifications, the abstract should qualify that these are proposed readouts within the framework, not fully machine-derived observables.

2. Several phrases are rhetorically stronger than the evidence warrants

Examples include: • “The reading is one transform.” • “The half vortex is observable.” • “The reading is Lorentzian.”

These should be softened unless accompanied by direct machine evaluation or theorem-level proof under clearly stated assumptions.

3. Some terminology may impede accessibility even for technical readers

Phrases such as “content carries no identity,” “time–depth axis,” “the point curvature of the reversed routers reads as charge,” and “observer’s split of that axis into time and depth” are distinctive, but the paper would benefit from a compact glossary or a boxed map between machine objects and claimed physical readouts.

4. Formatting and readability issues

The extracted text shows several encoding/hyphenation artifacts such as “three￾dimensional,” “post￾Newtonian,” “Gen￾eral,” and “eval￾uating.” Some line breaks also interrupt readability in equations and references. These may partly reflect PDF extraction, but they should still be checked in the source and final typeset version.

5. Some claims would benefit from immediate cross-reference to their evidential grade

The grading scheme is a strong feature. But because the manuscript alternates between [E], [M], [D], and [P], it would help readers if especially consequential statements in the introduction and conclusion carried explicit parenthetical grade reminders.

6. AI content analysis for this post-2021 work

Estimated percentage of AI-generated or AI-assisted content: substantial AI assistance is explicitly disclosed, but a reliable percentage cannot be inferred from the manuscript alone. The text itself confirms extensive AI involvement in implementation, analyses, derivations, language editing, figure drawing, and source editing.

Highlighted AI-associated section: The disclosure paragraph near the end is explicit that AI models “implemented and ran the machine and its analyses,” “carried out derivations,” “audited claims,” and “retrieved related mathematics and physics.”

Assessed epistemic impact: This disclosure does not invalidate the work, but it raises the verification threshold. Because the manuscript is conceptually ambitious and inferentially layered, extensive AI assistance heightens the need for independent human audit of derivations, code, and numerical analyses. In this case, the epistemic impact is material rather than superficial.

Recommendations

1. Reframe the paper more modestly as a partial foundational program unless the deferred computations are completed. The present title and abstract imply a stronger derivational closure than the manuscript supports.

2. Add a dedicated “Claim Dependency Table” showing, for each headline conclusion, whether it depends on [E], [M], [D], [P], or combinations thereof. This would significantly improve transparency and prevent interpretive overreach. The current grading system is promising but not yet fully operationalized in the paper’s rhetoric.

3. Complete or substantially extend the machine-level evaluation of the incidence transform, especially the ψ-channel. The manuscript itself identifies this as unfinished; without it, the charge/interference sector remains under-supported.

4. Separate clearly the three levels of contribution: a) exact structural results about the routing machine, b) coarse-grained numerical phenomenology, c) speculative physical identifications. Right now these are present, but too interwoven.

5. Provide direct reproducibility materials: code repository, parameter files, simulation scripts, and derivation notes sufficient for third-party verification. This is especially important given the explicit AI-assisted derivation and analysis workflow.

6. Introduce a falsifiability section listing at least three concrete predictions that would distinguish this framework from standard GR, conventional lattice-gas analogies, causal-set models, and topological toy models.

7. Tighten the discussion of gravity and relativity claims so that readers cannot mistake postulated identifications for independent predictions. The sentence stating that γ = 1 follows “from the equipartition postulate, not from a measurement” is commendably honest; the rest of the manuscript should be equally explicit whenever a major physical consequence relies on [P].

Overall, my review is skeptical but not dismissive. I see a technically imaginative manuscript with some nontrivial internal structure, but in its current form it overreaches relative to what has been established. The work would benefit from a substantial narrowing of claims or a substantial expansion of completed derivations and machine-level demonstrations.

Author response

Thank you for the review. Its central point, that [P] carries too much in the text, is right about the grading, and the next version changes the grading itself:

- The grade [P] is split. [P] keeps only the few principles of the construction. Definitions and the dictionary entries between machine and physics become [R], readings: each names a machine quantity after the physical quantity whose defining property it has, and that property is graded on its own ([M] or [D]).
- Time–depth equipartition was graded [P] in version 1, which understated it. It is derived from an exact machine fact, that the two identity assignments at a collision swap leave the same occupancy, and the principle that no reading may depend on that assignment. The Lorentzian signature and 1 + γ = 2 follow from it.
- A claim-dependency table is added as an appendix, as recommended.

On the other points:

- Schematic steps (M2): acknowledged. The ψ channel has not been evaluated on the machine, and the next version says so in the introduction.
- Uniqueness (M3): the double fibration is fixed by the geometry of great circles on S³, and the two measures by the even and odd combinations of the two strands under their exchange.
- Falsifiability (M7): γ = 1 comes with no free parameter, so any measured deviation would contradict it; quantitative departures from general relativity need the scale of G and of mass, which are open.
- Reproducibility (M8, recommendation 5): the scripts, readouts and a smoke test behind every number are deposited with the paper (doi:10.5281/zenodo.23124125; App. D).
- The broken words (minor 4) are hyphenation lost in text extraction.

XSci (AI review)

October 4, 2026

Referee Report

Manuscript: "On a Routing Network of Causal Flow: Incidence Transform from the Hopf Bundle to Four-Dimensional Spacetime"

Recommendation: Major revision

Summary and Recommendation

This manuscript asks whether light, gravitation and charge can be recovered as readings of a substrate that stores none of them. The substrate is a deterministic routing automaton on a trivalent honeycomb network whose only structure is a static per-vertex reversal flag written once at a "lock," and whose only content is register occupancy together with a routing letter at reversed vertices. Coarse-graining is argued to split the machine into a static half, in which the integers $N_A$, $N_B$ and $N_{AB}$ compose exactly under cell doubling and fix a first-Chern-type charge $c_1$, and a transport half, which retains only two conserved densities and their diffusion. A single incidence transform over a double fibration relating flow-line space to readout space then converts these data into four-dimensional quantities: the collective front reads as light at $c = \mathrm{STEP} = 0.354$ lattice constants per tick, the static offset of a buffer-induced delay field reads as a $1/r$ gravitational potential, and the point curvature $\kappa_v = \pm 2\pi/3$ at reversed vertices reads as charge obeying an intersection-number form of Gauss's law. Numerical support comes from automaton runs at $L \leq 1024$.

The work is ambitious and organized with unusual care. Its claim-grading scheme — [E] exact, [M] measured, [D] derived, [P] postulate — is a device much work in this area would benefit from, and the author applies it even where doing so is uncomfortable. The exact composition results of §3.2, the recognition that reversed vertices behave as order-3 orbifold points so that the total space is a Seifert fibration rather than a circle bundle, the degree-theoretic reading of Gauss's law, and the candid methodological disclosures of Appendix B are substantive strengths. The difficulties nonetheless concentrate where the conclusions are drawn, and fall into three groups: the numerical evidence does not yet discriminate the claims it is asked to support; the geometry named in the title is not the geometry on which anything is computed; and several foundational commitments announced at the outset are harder to sustain than the text allows, with the grading scheme applied less consistently in the headline statements than in the body. These appear addressable, and I recommend major revision.

Major Concern Category 1: The numerical evidence and the claims it is asked to support

1.1 The $1/r$ law is not discriminated, and the delay chain it anchors is untested

Equation (6) is the quantitative pillar of the gravitational reading, yet the evidence for it is the ratio $K \equiv 4\pi\sqrt{D_bD_z}\,r\,\mathcal{G}(r)$ reported in $0.80$–$1.06$ over $r = 9$–$92$: a band of roughly $\pm 13\%$ over one decade, with the hypothesized power already divided out and no fitted exponent or uncertainty given. Appendix B records that an earlier estimator produced an apparent $r^{-1.3}$ later diagnosed as an artifact, so the observable is demonstrably sensitive at this level, and with runs on $L = 96$ and $192$ tori extending to $r = 92$ periodic images cannot be neglected, since the image sum of a three-dimensional kernel saturates rather than falling as $1/r$. A log–log fit with errors at three or more sizes, with images controlled, would convert a consistency check into a measurement.

1.2 The delay field predicted for a mass has not been measured where it could be seen

Equation (7) is a linear-response ansatz whose supporting measurement shows how much flow the buffers hold rather than that they act as a local capacitance, and nothing in the chain after it is checked against the machine: the propagator actually measured is a tracer Green's function on defect-free flow, a medium with no buffers at all. The one relevant measurement, in §4.4, was made on a torus, where the mass's buffering is shared by all bits so that a static spatial profile cannot be separated from a uniformly growing background, and the reported constancy of inter-radius differences is therefore consistent with a static offset but equally with none. The same paragraph records a front "2–5 ticks late (2–4 s.e.)" at $r = 25$ while §6.2 relies on the front not being delayed. Measuring the delay profile around a buffered source in an open or absorbing geometry, and resolving that excess, would be the single most valuable addition to the paper.

1.3 Transport results are reported under two different versions of the buffer rule

Section 4.2 states that the normal-diffusion test used an earlier buffer rule and that the rule of Appendix A "changes the rates by up to 44% at $\rho = 0.3$, not the shape," and Appendix E makes the division consequential: the wavenumber test, the cell machines, the locality test, and the buffer-source and clock-rate measurements all carry the dagger of the superseded rule, while Table 1 does not. Yet the clock rate is what §6.2 promotes to every observer's clock and what enters both definitions of (13), and the buffer source fixes the capacitance in (7). The invariance of the shape under the rule change is asserted rather than shown. Table 1's parametrization needs attention too: the text reads as though $\beta$ in $1/D_n = a(p) + \beta\rho^2$ were a constant, but the tabulated values imply slopes of roughly $1.7$, $15$, $27$ and $43$ at $p = 0.2$, $0.4$, $0.6$ and $0.8$, and four rows have only three populated columns.

1.4 The localization claims and the attribution to a quantum universality class

Section 3.3 concludes that no stable diffusive phase exists, from a conductance falling as $\ln b$ with $dg/d\ln b = -0.036$ to $-0.044$: a total change of about $0.14$ over five octaves, reported without uncertainties or realization counts, in a paragraph that also notes that the sparsest substrates "first rise up to $b \approx 32$–$64$ and then fall." That non-monotonicity no single slope describes, and the strip boundary of Fig. 3 is characterized, correctly, as "the line where the localization length matches the measurement size," which makes it a finite-size contour rather than a phase boundary. More consequential is the interpretive step: the two-parameter flow and the Chalker–Coddington model derive their criticality from interference of complex amplitudes, whereas this machine routes occupancies deterministically with no amplitude at all, so the reported phenomenology is that of classical chiral percolation, with a different mechanism and different exponents. If the correspondence is an analogy of phase-diagram topology, saying so costs nothing; if an identification of universality class, the mechanism must be exhibited and exponents compared.

1.5 Isotropy is established only at second moment and only without defects

For lattices with the symmetry of the honeycomb base and the cubic lift, the second-rank velocity tensor is isotropic by symmetry alone while the fourth-rank tensor generically is not, which is the familiar reason lattice-gas hydrodynamics requires an explicit fourth-order check; and the claim here is stronger than the hydrodynamic one, since residual fourth-order anisotropy would appear in the readout as a direction-dependent front speed. Section 5.4 concedes "lattice corrections at higher order" without quantifying them. Moreover $D_b = D_z$ is verified to $0.02$–$0.5\%$ only on defect-free flow, precisely where the routing buffers are absent and the drag of reversed vertices identified in §4.2 as the base-momentum channel does not operate, so isotropy inside matter — where the gravitational reading needs the lift distance of (6) to be meaningful — is untested. Measuring $D_b/D_z$ at nonzero $p$, and the front speed along inequivalent directions, would settle both points.

Major Concern Category 2: The geometry that is named and the geometry that is used

2.1 The Hopf structure is invoked throughout but never instantiated

The Hopf fibration appears in the title, the abstract and the opening of §2.1, but every transport, localization, propagator, island and cascade measurement is on a honeycomb torus; §4.3 states that "the torus runs use a periodic patch of the base; the global Hopf structure enters only on the sphere," and Appendix E shows the sole spherical result to be the ribbon-linking check for rule 10. On the torus, as §3.1 itself observes, the bundle is "a Seifert fibration over the torus with Euler number $(N_A - N_B)/3$, never $S^3$," so the global constraint $\sum\kappa_v \in 2\pi\mathbb{Z}$ is never active. The sphere raises a sharper difficulty the manuscript notes and sets aside: the Goldberg sphere "is not bipartite, so there only the ribbon structure is used, not the strand labels" — yet strand conservation, the collision-swap rule, the A/B curvature distinction and the whole $\psi$ channel depend on those labels. A revision should resolve the resulting dilemma, either by running on a bipartite spherical triangulation or by arguing that the torus results transfer.

2.2 The fiber is compact where charge needs it and non-compact where gravity needs it

Charge is defined as boundary holonomy "in units of a full turn," counted as total turning rather than modulo a turn, and the force of the orbifold observation — that $\pm 2\pi/3$ is not a full turn and three vertices make a unit — depends on the fiber being a circle of definite circumference. Section 4.3 meanwhile states that the lift uses the unwrapped height and that "no result here depends on its circumference." If the circumference is genuinely irrelevant, the fiber height is an unbounded record of strand history and the third direction announced in §1.1 is a diffusive coordinate rather than a geometric one; if the fiber is a circle, equation (6) should be a cylinder Green's function with a crossover to a two-dimensional logarithm, exactly the crossover invoked in §4.4 and §6.4 when the manuscript warns that a source spread over all fiber heights would give a logarithm. Stating the circumference in lattice units and locating the fitted window relative to it would resolve the ambiguity.

2.3 The charge spectrum, and the resolution-dependence of the objects carrying it

Equation (3) gives $c_1 \in \frac{1}{3}\mathbb{Z}$, and §6.3 argues that an isolated $\pm\frac{1}{3}$ has no Gauss-law reading because no line bundle on $S$ carries that degree — which settles what can be read, not what can exist. Nothing restricts the defect pattern, §3.1 notes that the index law holds "a single reversed vertex included," and a lone reversed vertex would satisfy the island criterion of §3.6 while still contributing boundary holonomy at every radius and a $\mp\sqrt{3}$ term to the $C$-odd moment; no argument is offered for why three same-type vertices should bind. The carriers are moreover resolution-dependent, since "about half of the islands at one resolution revert to non-islands one level coarser" and "fusion is a change of resolution." The conserved integer is exact while its decomposition is not, so whether two charges are one particle depends on the observer's resolution, and since the mass of (13) is proportional to a source's buffered content, mass inherits that dependence.

Major Concern Category 3: Foundational commitments and the grading scheme

3.1 A global tick and a globally shared schedule phase

The first declared non-assumption is that there is no shared time and no global now, yet Appendix A specifies a synchronous update in fixed order in which every register advances together, a globally defined simultaneity surface in the ordinary sense; the collision-swap rule, requiring two bits to cross one edge in the same tick, suggests the synchronization is essential rather than incidental. More concretely, the tie rule admits a globally shared quantity into the local dynamics, since when several entry pairs qualify "the first in the router's turning sense from axis $t \bmod 3$ is captured." The appeal to Margolus-neighbourhood partitions is apt as far as it goes, but those make a reversible local rule well defined without a global parameter entering the physics, whereas here the phase arbitrates which pairs are delayed, and the delay field is the origin of the gravitational reading; ties occur in 6–7% of captures at $\rho = 0.3$. A run with a tie-break using no global parameter, compared against the synchronous results, would turn a definition into a demonstration.

3.2 The erasure clock counts the erasure of information the machine does not hold

Section 4.5 credits $\ln 2$ to each collision swap for "which outgoing bit is which," and §6.2 adopts the resulting rate as every observer's clock. But §5.4 establishes, graded [E], that the two ways of assigning identities leave the same occupancy, and draws from this the principle that no physical reading may depend on how identities are assigned. If identity is not state, the $\ln 2$ is not information the configuration lost but the entropy of a labelling the machine does not maintain — and this propagates, because $\Gamma_e$ fixes both $\Phi$ and $M$ in (13), while alternative conventions would change it by order-unity factors whose effect is unreported. A related circularity attends the check in §6.4, where $\Gamma_s$ is compared with $\mathrm{STEP}^2/D_z$: §4.3 already used that relation to interpret $D_z$, so the agreement is internal arithmetic rather than independent corroboration. The genuinely striking fact in §4.5 — a rate varying by only 7–20% across media whose event mixes differ three- to fourfold — deserves foregrounding rather than a single sentence.

3.3 Mass is a property of the environment, and the sources cannot move

Equation (13) makes rest mass proportional to buffered content times clock rate, and both factors are properties of the ambient flow: the mean buffered share is $\sigma_0 \approx 1.05\,p\rho$ and the clock rate scales roughly as $\rho^{0.8}$, so the same three reversed vertices would read as different masses at different background densities. The three-layer principle of §6 is evidently designed for this and is applied convincingly to charge, where the $C$-odd moment per unit of $N_A - N_B$ is exactly $-\sqrt{3}$, but no corresponding ratio-layer construction is supplied for mass. Second, reversed vertices never move, appear or disappear, so the objects identified as particles are permanently immobile and there is no geodesic equation, orbit or free fall. This bears on the equivalence claim of §6.2, whose two supporting checks concern the propagation of flow rather than the motion of a source, whereas the weak equivalence principle is a statement about the universality of test-body acceleration.

3.4 The grading scheme is applied less consistently in the headline claims than in the body

The abstract states that the equality of the two shares "fixes the post-Newtonian parameter at $1 + \gamma = 2$," while §6.2 states that $\gamma = 1$ follows "from the equipartition postulate, not from a measurement," and that postulate carries [P]; by the manuscript's own rule that a claim resting on [P] alone is written as a definition, this should not appear as a result, and the Cassini bound $\gamma - 1 = (2.1 \pm 2.3)\times 10^{-5}$ cannot test a framework predicting $\gamma = 1$ identically with no mechanism for deviation. The same looseness affects [E], defined as "a theorem, or a fact verified bit by bit," which are two different warrants: the index law is quantified over all defect patterns on the torus but supported by thirteen, and the swap geometry rests on a test torus asserted to cover "every local case." Relatedly, §5.3 concedes that the walk sum concentrates on straight flow lines "by a stationary-phase argument not proved here," so all of §6 is conditional on a step whose conditionality is invisible there.

Minor Issues

1. Reconciling the swap rate with the propagator. Section 6.4 reports $\Gamma_s \approx 0.11$ per bit per tick at $\rho = 0.15$ and calls this "within 5% of $\mathrm{STEP}^2/D_z = 0.118$"; the printed values differ by about 7%. Additional significant figures or a corrected tolerance would remove the discrepancy, and the swap counts of 265, 927 and 2893 per tick are quoted without a normalization convention.

2. The discrepant case in the light-speed test. Section 6.1 reports $c_{\mathrm{kin}}^2(1+\sigma) \approx \mathrm{STEP}^2$ "on 29 of 30 substrate and rule cases," but the thirtieth is neither identified nor discussed. Since medium-independence of the front speed is the whole content of the claim that light is medium-free, the outlier and its probable origin should be reported.

3. Fit windows and the crossover scale. The grey band $r < 9$ in Fig. 6 lies outside the fitted range without a stated criterion, while the $L = 192$ result is quoted for $r \geq 11$; the exclusion rule and its effect on the amplitude should be given. Separately, $b^* \approx 14$ is measured at one parameter point, yet the two-level architecture used throughout §§5–6 is defined relative to it.

4. Statistics at coarse resolution and in Table 3. The island fraction is said to stay near $0.10$ from $b = 4$ to $256$, but at $L = 1024$ and $b = 256$ a substrate holds sixteen cells, so that fraction is one or two islands. Table 3 carries eight seeds per point but two to four at $W = 256$, where the falling-versus-flat classification is drawn. Realization counts and error bars are needed in both places.

5. Non-monotonic density dependence at low defect density. Table 1 gives $D_n$ at $p = 0.2$ rising from $1.852$ to $2.031$ over $\rho = 0.05$–$0.1$, roughly 10% and beyond the quoted errors, before falling. The monotone form $1/D_n = a(p) + \beta\rho^2$ cannot represent a maximum, so either the fit is restricted to a subrange, which should be stated, or the low-density behaviour needs comment.

6. Definitions and conventions that need tightening. The quantity called conductance is the number of routes crossing a cell per face, whereas comparison with a two-parameter flow presumes a dimensionless conductance. The boundary convention for $\Theta$ in equation (4) should be stated, and the printed random-placement limit of (5) contains a typographical error for $p_Ap_B + (1-p_A)(1-p_B)$, an identity that does check out on substitution.

7. Claims whose grading and wording disagree. Section 3.3 says regions of different winding meet along "exactly $|\Delta h|$ co-directed chiral routes and no more" while marking the sentence measured. Lemma 1 is proved per bit although later arguments concern pairs. The spin-structure identification used in §5.3 and §6.3 is listed as open in §7 and should be flagged as conditional where it is used.

8. Loose tolerances and unsupported extrapolations. Locality is concluded from a $\pm 20\%$ match on striped substrates whose layers differ roughly threefold in $D_n$. The effective-medium result, a 3–10% reduction at $p = 0.03$–$0.07$, is extended to "for real matter it is negligible" without the area-fraction estimate that inference requires, and Appendix B's removal of the observer's own buffering excludes an effect §5.2 says is always present.

9. Quantities whose constancy deserves a check. The strand-difference coefficient is reported as $D_P \approx 0.07$–$0.10$, nearly independent of $p$ and $\rho$, while $D_n$ varies ninefold, and §7 lists this as an open puzzle. Confirming that $D_P$ is not limited by estimator resolution, for instance by the wavenumber-independence test already applied to $D_n$, would precede treating the near-constancy as physics.

10. Scope, postulates and reproducibility. Neither the torus-only scope nor the fact that the $\psi$ channel was never evaluated is visible from the abstract. A table of the minimal postulate set would help, since [P] attaches to at least eleven identifications. Seeds, versions and runtimes should be recorded, the README label crosswalk included as a table, and $\overline{h}$ and the form of $G$ restated self-containedly rather than imported from a companion deposit.

Missing and Inadequately Integrated Literature

Cited but Inadequately Integrated

Several works are invoked for a single feature while their substance bears directly on claims the manuscript makes. Wang and Cohen (1995) and Ruijgrok and Cohen (1988) establish the rotator and deterministic Lorentz lattice-gas setting, but neither their measured transport coefficients nor their classical trajectory-closure mechanism is compared with what is found here, although both bear on whether the logarithmic conductance drift is a localization signature or a known feature of rotator dynamics on this lattice. Chalker and Coddington (1988), Khmelnitskii (1983) and Pruisken (1984) are cited to assert a quantum-Hall phase structure without reproducing the predicted weak-coupling slope, which could be set against the measured $-0.036$ to $-0.044$. Helgason (2011) is cited for the Radon and Funk family without the inversion and support theory that determines whether a reading loses information, and Penrose (1967) supplies the double-fibration analogy and the correct observation that the real version yields signature $(2,2)$ without drawing the implication that the Lorentzian reading is imposed by the observer split rather than supplied by the geometry. Kibble (1976) supplies a defect-formation picture but not the statistical predictions bearing on whether the lock produces same-type triples, and Frisch, Hasslacher and Pomeau (1986) is cited only for six channels with exclusion when its central lesson here is the lattice-symmetry requirement for isotropic hydrodynamics. A distinct issue concerns the Acknowledgments, which state that seven cited works' shapes are "used here without being cited in the text," among them causal sets, which bear on the first non-assumption of §1.1, and holography, which bears on the identification of fiber height with holographic depth. Placing each citation where its idea is used would let readers see which element derives from where.

Truly Missing Papers

The most consequential omission concerns the universality-class attribution. Cardy (2010) is devoted to the relationship between quantum network models and classical localization problems and is explicit about when a quantum model's observables reduce to those of a classical walk and when they do not, making it the natural authority for deciding whether this construction belongs to the Chalker–Coddington class; Beamond, Cardy and Chalker (2002) supplies the technical derivation, mapping the transition onto one for self-avoiding classical walks identified as percolation-cluster hulls. On the transport side, Kong and Cohen (1989) shows that in deterministic lattice-gas wind-tree automata a non-Gaussian displacement distribution can coexist with a mean-square displacement linear in time, so the two tests used here can both be passed by a non-Gaussian walk and the distribution should be reported directly. Three further works would situate the manuscript relative to the problems it addresses: Collins, Perez, Sudarsky, Urrutia and Vucetich (2004), which argues that a discrete substrate with a preferred frame generically induces percent-level Lorentz violation absent fine tuning and is the obstacle the claim that boost covariance is "a property of the reading" implicitly answers; Ambjørn, Jurkiewicz and Loll (2004), the standard account of four-dimensional geometry emerging from a discrete causal substrate, which offers the template of treating dimension as a measured, scale-dependent observable; and Barceló, Liberati and Visser (2005), which surveys analogue models yielding an effective Lorentzian metric and is explicit about which features of general relativity they reproduce and which, notably dynamical field equations, they do not. The first four of these were verified to exist as cited.

Ambjorn, J., J. Jurkiewicz, and R. Loll. 2004. Emergence of a 4D world from causal
    quantum gravity. Physical Review Letters 93:131301.
Barcelo, C., S. Liberati, and M. Visser. 2005. Analogue gravity. Living Reviews in
    Relativity 8:12.
Beamond, E. J., J. Cardy, and J. T. Chalker. 2002. Quantum and classical localization,
    the spin quantum Hall effect, and generalizations. Physical Review B 65:214301.
Cardy, J. 2010. Quantum network models and classical localization problems.
    International Journal of Modern Physics B 24:1989-2014.
Collins, J., A. Perez, D. Sudarsky, L. Urrutia, and H. Vucetich. 2004. Lorentz
    invariance and quantum gravity: an additional fine-tuning problem?
    Physical Review Letters 93:191301.
Kong, X. P., and E. G. D. Cohen. 1989. Anomalous diffusion in a lattice-gas wind-tree
    model. Physical Review B 40:4838-4845.

Conclusion and Path Forward

The manuscript pursues a question worth asking, and the apparatus assembled to pursue it is substantial. Its organizational discipline is real: the grading scheme makes the construction legible in a way comparable work rarely is, the disclosures of Appendix B inspire confidence rather than diminishing it, and several combinatorial results — the exact composition of the static register, the orbifold reading of reversed vertices, the degree-theoretic form of Gauss's law — are clean contributions in their own right. The concerns raised here are not that the construction is wrong but that the distance between claim and warrant is larger than the text acknowledges, and that the gap falls where the conclusions are drawn.

Three changes would do most to close it. Empirically, the $1/r$ consistency band should become a measured exponent with an uncertainty at three or more system sizes with periodic images controlled, the delay field should be measured around a buffered source in a geometry capable of separating a static profile from a uniform background, the front excess at $r = 25$ should be resolved, and the transport suite brought onto a single buffer rule or reported under both. Geometrically, the machine should either be run on a bipartite spherical triangulation, where the global constraint can act and charge quantization can be tested against it, or the results stated plainly as torus results with the title and abstract adjusted; the fiber's circumference should be given; and the status of an isolated fractional vertex settled rather than deferred. Finally, the grading discipline applied so carefully in the body should extend to the abstract and introduction, so that $1 + \gamma = 2$ appears as a consequence of an adopted postulate, finite verifications are distinguished from theorems, and the conditionality of §6 on the unproven concentration result is visible where it matters. Addressing these would leave the central ambition intact while making the achievements easier to see and to credit.

Author response

Thank you for a careful and constructive report. Brief replies follow; section numbers refer to the current version.

Category 1: the numerical evidence

1.1 The 1/r law. The next version fits the exponent at L = 48, 96, 192 and 384: n = 0.966–0.969 (s.e. ≤ 0.014) with no trend in size, and the shortfall from 1 is a 1/r² correction, K = A(1 + c/r²) with A = 0.998. Periodic images do not enter: displacements are unwrapped on the covering plane, so no image sum is formed and r may exceed the torus (App. B). The displacement distribution is reported directly; its kurtoses approach the Gaussian values.

1.2 The delay field around a mass. Agreed that this is the most valuable measurement. It is not in the next version; the linear-response chain and the excess at r = 25 are listed as open.

1.3 Two buffer rules. The results marked † are rerun under the rules of App. A. The wavenumber test and the clock-rate test come out as before; the buffer source had already been measured under both rules. On striped substrates, diffusion across the stripes now follows the harmonic mean of the layers to 6%, and along them the longest mode lies within 13% of the arithmetic mean. The exception is the cell machines, which the next version reduces to one sentence, since how an L2 machine is specified is a separate question from what coarse-graining keeps. Table 1 will write β(p), state the fit subrange at p = 0.2 and mark the cells not measured.

1.4 Universality class. Accepted. The correspondence is an analogy of phase-diagram topology; the mechanism is classical, since routes carry no amplitude, and the natural class is classical localization on directed networks (Beamond, Cardy and Chalker 2002; Cardy 2010), with exponents left open. "Conductance" will be named as a route count; errors and realization counts for the slopes are left open for further work.

1.5 Isotropy. Both points are measured in the next version. In free flow the fourth-rank anisotropy of the propagator decays as 1/t, and for r ≥ 20 the propagator agrees in every direction to within 4%. Inside matter the lift is not isotropic (D_b/D_z falls from 0.99 at p = 0 to 0.17 at p = 0.6), because reversed vertices take up base momentum while the fiber climb is only paused; D_b = D_z holds in the free flow between islands, where the 1/r propagates.

Category 2: the geometry

2.1 The Hopf structure. The closed base follows from an axiom of the framework that every postulate presupposes: the total information of the universe is finite, so the base is a closed surface and the space of flow lines is compact, the great circles of S³ fibered over S² by the Hopf bundle. The machine's results do not depend on the closure: transport, the propagator and the islands are local and need only the local honeycomb, which the torus runs sample on scales far below that of the base. The closure enters through the global constraints of the readout: Gauss's law on the closed readout space makes the total charge zero and pairs the odd charges. A closed trivalent base cannot be bipartite everywhere: by Euler's formula it needs base defects, around which the strand and arm labels acquire holonomy. This fits a network made of defects everywhere; on a base of cosmological size their total deficit, 4π, is spread over the whole sphere. The next version states the axiom and the base defects in §1.1, and the torus scope in the abstract.

2.2 The fiber. It plays both roles. Compact, it carries holonomy, with a whole turn as the unit of charge; unwrapped, it carries diffusion and the 1/r. The defects make the paths differ: around them routes close locally, while elsewhere the flow runs close to great circles, which close only on the scale of the base. The circumference is of that scale, far beyond the fitted window (r ≤ 92), so the crossover to a logarithm lies at the scale of the base. As for the first horn: the third direction is indeed the history of the flow, its climb along the fiber (§1.1). It reads as a geometric direction because, with identities erased at swaps, steps along it cannot be told from steps along the base (D_b = D_z).

2.3 Charge spectrum and resolution. An isolated reversed vertex can exist; whether it can be read depends on the scale at which it closes, with two other vertices of its type, into a unit of charge, and one that closes only on astronomical scales cannot be told from the global flow. That the conserved integer is exact while its split into particles depends on resolution is intended: fusion is a change of resolution. Held content, which sets the mass, adds over vertices in the same way (a one-type island of twelve vertices holds 3.69–3.94 times the content of a three-vertex closure), so the total does not depend on how it is grouped.

Category 3: foundational commitments and grading

3.1 The global tick. The tick is a common step of flow-line space: it fixes the order of updates, and every schedule that respects that order gives the same outcome. The next version shows this on the machine: asynchronous schedules in which one router runs up to 10 ticks ahead of another, each router breaking ties from its own update count, reproduce the synchronous run exactly, and negative controls fail. The tick is not a time in readout space.

3.2 The erasure clock. Accepted, and it changed the paper. Away from the buffers the update is a bijection, swaps included; the only merges are releases, which forget the held pair's entry axis. A swap therefore loses only which bit is which, a branching of the reading, and the arrow sits at the merges. The clock is redefined as the time–depth axis's own step rate, Γₐₓ = c²/D_z, which is measured rather than counted; it varies by 13–17% between media while the mix of events changes three- to fourfold, and the next version foregrounds this. The comparison in §6.4 is now called what it is, the relation of §4.3.

3.3 Mass and motion. Mass is a rate-layer reading and depends on the medium. On the machine the held content of a source varies about twentyfold across six media, while the ratio of two sources stays within 4–5%; the scale of mass is listed as open. Particles do not move in flow-line space, which is a single frame; they move relative to one another in readout space, and free fall is an inertial frame of the reading. The equivalence check concerns test bodies: an object's stationary-phase path depends on the delay field alone, not on its frequency.

3.4 Grading. The next version splits [P] into a few principles and [R], readings; adds a claim-dependency table; derives the even split from the principle that content carries no identity; distinguishes theorems from exhaustive finite checks within [E]; and states at the start of §6 that the readings assume the stationary-phase concentration. On Cassini we see it differently: γ = 1 with no free parameter is exposed to any measured deviation beyond about 10⁻⁵.

Minor issues

1. Γₛ = 0.112 per bit per tick, each bit counting the swaps it takes part in, against STEP²/D_z = 0.118.
2. The exception is p = 0.6, ρ = 0.3 under the earlier rule (0.101); the count is 31.
3. The grey band is the fit exclusion, within one to three free paths of a bit between swaps; starting the fit at r = 20 instead gives n = 0.99, and the 1/r² correction accounts for the difference. b* is measured at one point only.
4. Realization counts and error bars for the island fraction at b = 256 and for the strip transfer at W = 256 are left open for further work.
5. The fit subrange is stated (1.3).
6. "Conductance" is renamed (1.4) and the convention for Θ stated. Eq. (5) reads correctly in the source; the error came from text extraction.
7. "Exactly … and no more" is removed; Lemma 1 is applied to each member of a pair; the spin-structure reading is flagged as open where it is used.
8. Locality is restated with its numbers (1.3): across the stripes the match is 6%; along them, where the estimator is less precise, it is 13% for the longest mode. The extrapolation to real matter is removed, and App. B states that the clean source is a measurement design.
9. Whether D_P is limited by the estimator's resolution is left open for further work; the paper reports its near-constancy as a measurement and draws nothing from it.
10. The torus scope is in the abstract and the unevaluated ψ channel in the introduction; the postulates are in the dependency table, and seeds, versions and run times in the deposit.

Literature

Cardy (2010) and Beamond, Cardy and Chalker (2002) are now cited where the class is discussed; with the class classical, the weak-coupling slope of the quantum two-parameter flow is not the comparison to make. The other suggested works are not added, since the paper cites works whose results its text uses. For Kong and Cohen (1989), the next version reports the displacement distribution directly (1.1). For Collins et al. (2004), our present answer at the level of the machine is the measured anisotropy (1.5); whether interactions in a field-theoretic reading would carry lattice effects back into the low-energy readings is not addressed here. Like the analogue models surveyed by Barceló, Liberati and Visser, the paper reads out static weak-field gravity, not field equations.

On Penrose (1967), the paper draws the implication you point to: the real double fibration gives signature (2,2), and the Lorentzian signature comes from the observer's split of the time–depth axis, a split itself derived from the even split (§5.4). The lesson of Frisch, Hasslacher and Pomeau is taken up under 1.5. The questions raised through Wang and Cohen, Ruijgrok and Cohen, Helgason and Kibble are real ones that our line of research has not yet reached.

The shapes named in the acknowledgments were borrowed as abstract concepts, and the ontological choices behind them differ from those made here. In causal set theory the order of events is spacetime; here the causal order is that of updates in flow-line space, and time is a reading of it. In holography a boundary carries the degrees of freedom of a bulk; here depth is the history of the flow along the fiber, read through a change of chart. Citing them where the text uses a result would suggest a dependence that is not there, so they are acknowledged instead.

XSci (AI review)

October 4, 2026

Referee Report

Manuscript: "On a Routing Network of Causal Flow: Incidence Transform from the Hopf Bundle to Four-Dimensional Spacetime" Recommendation: Major revision required

Summary and Recommendation

This manuscript asks whether light, gravity and electric charge in four-dimensional spacetime can be read out of a deterministic lattice-gas automaton that stores no spatial coordinate, no light cone and no field. The machine is a fixed trivalent routing network on which anonymous bits circulate through six-way routers, with a quenched population of reversed routers acting as frozen defects. Two microscopic events carry everything: a reversible collision swap on edges, which loses only which bit is which, and a routing buffer at reversed vertices, which holds a pair briefly and so writes delay. The author argues that coarse-graining separates into a static half, whose topological integers survive exactly at every block size, and a transport half, which retains only two conserved densities and their diffusion. A cubic lift gives each bit a fiber climb alongside its base motion, producing an isotropic three-dimensional walk whose time-integrated propagator is measured to fall as $1/r$ over $r = 9$–$92$. An incidence transform on a double fibration, modelled on Penrose's twistor correspondence and on Radon-type transforms, is then proposed as the single operation that reads both halves as physics: the collective front as light at $c = \text{STEP} = 0.354$ lattice constants per tick, the buffer-induced delay as a Newtonian potential, and the point holonomy $\pm 2\pi/3$ at reversed vertices as charge. Equipartition of a single time–depth axis is claimed to fix both the Lorentzian signature and the post-Newtonian parameter at $1 + \gamma = 2$.

The work has real strengths and the author deserves credit for several of them. Deriving the routing table from ribbon topology rather than assigning it is elegant, and Lemma 1 — that a collision swap followed by a plain router is indistinguishable in exit arm from no swap followed by a reversed router — is a genuinely pretty structural identity that earns its later use. The exact $-\sqrt3$ C-odd moment per reversed vertex, the index law, and the sealing invariant are crisp and checkable in a way this literature rarely manages. Recasting Gauss's law as the degree of a restricted line bundle, with the consequence that $\pm\tfrac13$ is the degree of no bundle on $S^2$ and a lone reversed vertex therefore has no Coulomb reading, is an appealing resolution of what would otherwise be an embarrassment. The claim-grading scheme, the dependency table of Appendix E, and the script-to-result index of Appendix D together set a standard of auditability that most work of this kind does not approach, and the candour of Appendix B, which retracts an earlier estimator and names its artifacts, builds genuine trust. My recommendation is nonetheless major revision, because the inferential chain from machine to physics has three weak joints that the present text does not close: a single postulate asserted in prose carries nearly every physical conclusion; the empirical results that appear to support the gravitational reading are either generic to any three-dimensional diffusion or, where they test the construction directly, null; and the statistical and verification apparatus, strong in design, is reported in a way that does not yet let a reader judge how much it establishes. Each is addressed below, along with what I take to be a tractable path through it.

Major Concern Category 1: The Inferential Chain from Machine to Physics

1.1 One postulate carries the physical readings, and its consequences are asserted rather than derived

Section 1 announces that there are few postulates, and that the chief one — that content carries no identity (§5.4) — fixes the isotropy of the lift and the time–depth axis with its even split, from which follow both the Lorentzian signature and $1 + \gamma = 2$. Appendix E confirms the concentration: under the single heading "[P] Content carries no identity" sit the isotropic lift, the time–depth axis, the even split, the signature, $\gamma = 1$ with the Shapiro delay and light bending, the E/B apportionment, and the clock. Parsimony is a virtue and this is not in itself a flaw, but it means the postulate must bear a correspondingly heavy load, and the manuscript's justification does not yet carry it. The first step is convincing: that the two identity assignments at a swap leave the same occupancy is a verifiable machine fact, and the averaging argument that $q = \tfrac12$ is the only identity-independent matching follows cleanly. The later steps are of a different kind. That the occupancy fails to record whether a bit's progress came from delay or from turning does not by itself entail that no readout can separate time from depth; it entails only that a readout doing so would need information the occupancy lacks, and the static register — retained exactly at all scales, and invariant under nothing in particular — is precisely such a reservoir. The equipartition argument compounds this, since the symmetry invoked is a $\mathbb{Z}_2$ exchange of two outgoing bits while the quantity constrained is the apportionment of a continuous share between time and depth, and the map between these is not constructed. Because the signature, $\gamma = 1$ and the equal E/B split all descend from exactly this step, the reader is entitled to a formal statement: the space of admissible splits, the group action upon it, and a uniqueness proof for the fixed point. At present §5.4 delivers four numbered consequences in prose and is among the shortest subsections in the paper, which is the inverse of the weight it carries.

1.2 The incidence transform, presented as the unifying mechanism, is evaluated in neither channel

The introduction promises that the reading is one transform, and §5 defines it with care. What §6 then delivers is obtained by other means. The section opens by conceding that every reading rests on "the stationary-phase concentration of §5.3, which is assumed here rather than proved," and the worked example of §6.4, admirably explicit about its own limits, confirms the pattern: the counts are "not evaluated line by line here," the $\rho$ reading is obtained by substituting the separately measured propagator of §4.3 into the delay integral rather than by evaluating $R_\rho$, and $R_\psi$ "has not been evaluated on the machine." The assumed analytic step deserves particular scrutiny, because it is delicate in exactly the regime at issue. At $\lambda = 0$, the $\rho$ channel, there is no phase at all, so no stationary-phase mechanism operates and the concentration on straight lines must come from the ballistic character of routes between islands — a fact about the machine, not a property of the transform. At $\lambda \neq 0$ the phase accumulates along paths the manuscript itself describes as diffusive inside matter, and a diffusive measure does not generally concentrate on geodesics, particularly where §4.3 records $D_b/D_z$ falling to $0.17$. A further gap sits beneath both channels: the walk measure $d\mu_o$ in Eq. (11) is characterized only as real and positive and is never constructed, yet the transform's output amplitude — and through Eq. (14) the coupling $G$ — depends on its normalization. The remedy here is unusually concrete and appears to lie within reach of the deposited code. Evaluating $R_\rho$ directly on the single-closure configuration of §6.4, by sampling flow lines through a readout point and integrating the counts over the sky, and comparing the result against the propagator-based value, would convert the transform from a framing device into a tested mechanism. Evaluating $R_\psi$ around a closure to confirm the predicted $\pi$ advance would do the same for the charge half. Either outcome would be informative, and the author already lists both among the open problems.

1.3 The agreement with general relativity is fixed by construction and cannot discriminate

Section 6.2 derives $\gamma = 1$ and is admirably clear about its provenance, noting that the value follows from time–depth equipartition "not from a measurement." The manuscript then offers the Cassini bound, $\gamma - 1 = (2.1 \pm 2.3)\times10^{-5}$, as consistent with the structural $\tfrac12$, and adds that a split departing from $\tfrac12$ by more than about $10^{-5}$ would contradict it. This last clause is the paper's only stated falsification condition, and it does not function as one. The split is not a measured property of the machine that might have come out near but not at $\tfrac12$; it is postulated to be exactly $\tfrac12$, and §5.4 argues further that equipartition is a fixed point of coarse-graining, so nothing can move it. What the Cassini result tests is whether the construction is viable at all, and on that it returns a pass that general relativity itself and every scalar-tensor theory in its GR limit would also return. The honest statement is that the construction's internal consistency is a real achievement, since an emergent model could easily have produced $\gamma \neq 1$, while empirical support is a separate matter the comparison does not supply. What would supply it is a second parameter, and the most natural candidate is $\beta$, the nonlinearity parameter constrained by perihelion precession and by lunar laser ranging. Here the manuscript's position is genuinely unclear and worth clarifying, because Eq. (8) and its Appendix C derivation are explicitly first order in $\sigma$: if a second-order calculation is feasible it yields a prediction that could disagree with observation, whereas if the construction is intrinsically linear then it cannot reproduce general relativity beyond the first post-Newtonian order, and that limitation should be stated prominently rather than left implicit. The same gap affects the preferred-frame parameters, which bear directly on §5.4's concession that diffusion singles out the rest frame of the medium.

1.4 The gauge sector has topology but no dynamics

Section 6.3 establishes that the flux through a closed surface equals the enclosed $\sum c_1$ independently of the surface's shape and size, and the five-step derivation in Appendix C is correctly structured. But electromagnetism is not exhausted by Gauss's law, and what follows it is absent. No Coulomb force is derived, so nothing says how a test charge responds to another's field and no coupling analogous to the $G$ of Eq. (14) is identified; there is no Ampère or Faraday law and hence no wave equation for the field. The E/B assignment is formal in a related way: a boost does not simply reapportion the two in equal shares, since the Lorentz transformation of the field strength mixes them with velocity-dependent coefficients, and the manuscript asserts the right shape without writing the mixing law. One consequence of the present assignment deserves direct attention, namely that for a static charge the depth share is identified with a magnetization of $\mp\sqrt3$ per reversed vertex. This says a static charge carries a definite magnetic moment fixed by its charge, which is not obviously compatible with electrodynamics, where a static charge has no magnetic field in its rest frame; either the moment is not to be read magnetically, in which case the sentence should be rewritten, or it is, in which case it is a prediction whose magnitude and observational status should be given. The spin-structure claim needs similar care, since the half vortex, the winding $l = c_1/2$, and the integer total angular momentum of a unit charge all descend from stipulating that the relevant bundle is the spin structure rather than the frame bundle — a point §7 correctly lists as underived. Until it is settled, these results should be presented as conjecture rather than in the declarative form currently used in §6.3 and in the abstract.

Major Concern Category 2: The Empirical Foundation of the Gravitational Reading

2.1 The $1/r$ propagator is generic to three-dimensional diffusion and does not discriminate

The empirical centerpiece of the gravitational argument is §4.3, which measures the time-integrated identity propagator on defect-free flow and reports $K \equiv 4\pi\sqrt{D_bD_z}\,r\,\mathcal{G}(r) = 0.80$–$1.06$ over $r = 9$–$92$, with the emphasis that the $1/r$ "needs no defects and no tuning." That is true, and it is the difficulty. The time-integrated Green's function of any isotropic three-dimensional diffusion is $1/(4\pi Dr)$, as the manuscript's own citation of Helgason in §5.3 acknowledges, so measuring $1/r$ in a system constructed via the cubic lift to be an isotropic three-dimensional walk is a successful check of the simulation and the estimator rather than evidence that the quantity is a gravitational potential. What would make the reading discriminating is a second, independent signature, and the manuscript attempts this through $\gamma$ — which, as above, is fixed by postulate. Both legs of the gravitational case thus consist of a generic diffusion result and a postulated parameter, with no measurement that could have come out otherwise, and the author should say what observation on the machine would have falsified the reading. The quantitative reporting also needs strengthening, since the $K$ range represents roughly a 26% spread around the ideal $K \equiv 1$ with no error bars on $K$ itself, and the fitted exponents of $0.966$–$0.969$ with standard errors at or below $0.014$ sit two to three standard errors below unity without comment. The proposed correction $K \approx 0.998(1 - 7/r^2)$ amounts to $8.6\%$ at the inner edge of the fit window, comparable to the deviation it explains, and whether its form was chosen before or after inspecting residuals is not stated. Because Eq. (14) identifies $1/(4\pi n_{\text{vis}}\sqrt{D_bD_z})$ with $G\bar h/c^4$, the amplitude of $\mathcal{G}$ is not a detail but the coupling constant, and a systematic error budget together with a justification of the $r < 9$ exclusion is needed before Eq. (14) can be read as more than dimensional bookkeeping.

2.2 The machine's direct tests of gravitational time dilation return null results

Section 4.4 reports two measurements that bear directly on the gravitational reading, and both come out negative. A static mass on a torus leaves no permanent local difference of clock rate, with the excess delay growing at the same rate at every radius from $r = 6$ to $35$ and the differences between radii constant over $8\times10^3$ ticks. Outside the mass the front is not delayed within $\pm0.3$ ticks out to $r = 13$, and the 2–5 tick excess at $r = 25$ is reported at two to four standard errors and deferred as unsettled. The manuscript's response is to relocate the effect, noting in §6.2 that the Shapiro delay is a statement in readout coordinates and that the static offset is a value at each point rather than something accumulated along a path. The move is internally coherent but carries a cost the text does not acknowledge: it removes the machine from the evidentiary chain for the paper's most distinctive predictions, so that the propagator runs, the clock-rate runs and the front-delay runs are no longer supporting the gravitational claims they were built to test. Two things would help. The torus null result has an obvious candidate explanation the manuscript notes only parenthetically — on a finite periodic domain every bit samples the buffering region over long times — and testing whether the uniform growth rate of $5\times10^{-4}$ per tick scales with $L$ would distinguish a finite-size artifact from a genuine absence of local clock shift. Separately, the $r = 25$ excess is the paper's only direct measurement of anything resembling a Shapiro delay, and stating whether its sign and magnitude are compatible with, or an order of magnitude from, the predicted value would materially change a reader's assessment even with large uncertainty.

2.3 The propagator and its source are measured in mutually exclusive regimes

The gravitational reading needs two ingredients that must coexist: a source of delay, which is held content at routing buffers, and a propagator carrying it as $1/r$. Each is measured, but never together. The propagator is measured on defect-free flow, and the isotropy $D_b = D_z$ that makes the lift three-dimensional holds only there; §4.3 reports that as soon as defects appear the ratio falls to $0.44$ at $p = 0.1$ and to $0.17$ at $p = 0.6$. Yet buffers exist only at reversed routers, so held content requires $p > 0$ by construction, and the configuration in which the delay field has a source is precisely the one in which its propagator is not isotropic. The manuscript answers with a refraction analogy and with §3.6's assertion that the space between islands has $p = 0$ at the finest resolution, so that sources are compact and the medium between them clean. This picture is plausible but asserted, and the one run that tests it directly is the null result discussed above. What is missing is the intermediate measurement that would settle it: the time-integrated propagator from a compact defect source into clean flow, as a function of $r$, compared against $1/(4\pi\sqrt{D_bD_z}r)$ with the clean-flow values of $D$. This is neither the §4.4 front measurement, which tracks a wave front rather than an integrated field, nor the §4.3 propagator measurement, which has no source; and since §4.4's effective-medium paragraph reports a field per unit source lowered by 3–10% by sub-resolution islands, some version of the measurement appears already to exist, in which case its radial dependence is the single most relevant unreported number in the paper. A terminological matter compounds this. The symbol $p$ carries two incompatible meanings: in §§2–4 it is the fraction of reversed vertices on a uniform substrate, with runs from $0.2$ to $0.8$, while §3.6 retroactively reinterprets a nonzero coarse $p$ as the density of sub-resolution islands and the uniform substrates as mean-field coarse cells. On that reading, Table 1 tabulates transport inside densely packed matter rather than in anything resembling free space, which governs the meaning of every number in §4 and should be stated at the start of that section rather than three sections earlier.

2.4 Lattice-scale tensions with the Lorentzian reading are measured but never confronted together

Three facts bear on how the construction sits with Lorentz invariance, and the manuscript reports each in isolation. First, §6.1 notes that single bits move along the cubic axes at $\sqrt{3/8}$, which is $\sqrt3$ times the front speed $c = \text{STEP}$, so the microscopic constituents propagate at roughly $1.7$ times the speed the readout calls light. Section 5.4 addresses the analogous question for diffusion and concludes it stays within the light cones, but says nothing about ballistic single-bit motion, which is not diffusive and not bounded by $c$; if anonymity prevents such motion from carrying a signal that is a substantive argument and should be made, and if it does not, the readout admits superluminal signalling. Second, §5.4 reports that the time-integrated propagator is raised along the cubic axes by up to 44% at $r = 5$, with a residual 4% directional spread persisting for $r \geq 20$. Since experimental bounds on Lorentz violation reach extraordinary precision in several channels, a construction whose isotropy is restored only asymptotically must say what fixes the lattice constant and the tick in physical units — and the manuscript provides no such bridge anywhere, leaving the scale of $G$ explicitly open. Third, §5.4 concedes a preferred frame, since diffusion singles out the rest frame of the medium. The distinction drawn between a frame chosen by the state and one chosen by the law is the standard and legitimate move for a broken symmetry, but a dissipative medium with a rest frame produces observable effects that go unaddressed. These three combine with a fourth point already noted: because observers are themselves defect patterns, every actual observer sits where $D_b \neq D_z$, so the manuscript owes an account of why the signature is Lorentzian where readings are in fact taken.

Major Concern Category 3: The Statistical and Verification Apparatus

3.1 Several headline numerical conclusions rest on trends whose uncertainties are unreported

Section 3.3 concludes that there is no stable diffusive phase, inferring this from a conductance falling as $\ln b$ with $dg/d\ln b$ between $-0.036$ and $-0.044$. The conclusion is strong — the absence of a phase — and three things are needed to evaluate it. The absolute scale of $g$ is never given, so a reader cannot tell whether the total change of about $0.15$ over the measured range of $b$ is a one-percent or a twenty-percent effect; no statistical error accompanies any individual slope, the quoted range evidently being the spread across densities, which is a different quantity; and the number of disorder realizations is unstated, which matters because at $b = 256$ on an $L = 1024$ system only sixteen coarse cells exist, so the sample is smallest exactly where the asymptotic claim is made. The physical reading is likewise underdetermined, since a slow logarithmic fall is the expected signature of several distinct mechanisms, among them the two-parameter flow the manuscript invokes, a simple crossover, and a finite-size effect in which the measurement window rather than the physics is running; the author half-acknowledges the last in observing that what looks diffusive at one resolution is a crossover between resolutions, but does not separate them. Reporting $g(b)$ with error bars across several $L$ at fixed $b/L$ would do so. The subsequent localization-length formula $\xi \approx b\,e^{g/0.04}$ inherits these difficulties and amplifies them: it integrates the measured flow and exponentiates, so that a ten-percent uncertainty in a slope measured across 3.5 $e$-folds propagates to a factor of $e^5$ or more in $\xi$. Either the propagated uncertainty should be stated, making clear that $\xi$ is determined only to within many orders of magnitude, or the formula should be presented qualitatively. The strip-transfer boundary raises a related issue that the author states honestly in §7, namely that a line defined by $\xi_W \sim W$ at the largest accessible width is a crossover locus rather than a phase boundary, yet Figure 3 plots it as a candidate boundary with points classified by whether $\eta$ falls "significantly" — a threshold never defined, and one that Table 2 shows to be doing real work, since a change from $1.68$ to $1.62$ is classified as a slow fall while seed-to-seed scatter on $\eta$ is nowhere reported.

3.2 The coarse-graining closure, which determines the theory's content, lacks a quantitative criterion

Section 4.1 establishes which variables survive coarse-graining and therefore what the transport half of the theory contains, making it pivotal. The argument is that a linear Markov model on six direction counts mispredicts the cells' own correlations and closes once each cell carries leaky integrals of its bits' base displacement and fiber climb, with forgetting times from 8 to 512 ticks. Three things are missing. No quantitative criterion for closing or mispredicting is given, so neither the discrepancy of the simpler model nor the residual of the augmented one can be assessed, and the paper's subsequent architecture rests on this judgement. The forgetting-time family spans six octaves, which raises a question of fitting capacity relative to the number of independent correlation functions matched; the right answer is the out-of-sample test the manuscript does report, namely that an L2 machine reproduces per-cell fluctuations on substrates it was not fitted to, but that result is relegated to an appendix and marked as having used the superseded per-axis buffer rule, so re-running it under the current rules would be valuable. Finally, the crossover scale $b^* \approx 14$ is determined at a single parameter point and then used throughout as the boundary between ballistic and diffusive behaviour and as the definition of level L2; since $D_n$ varies ninefold across Table 1, $b^*$ should be expected to vary too, and a short scan over the same grid would settle where the two-level structure actually sits.

3.3 The exactness grade is applied inconsistently, weakening the scheme the paper relies on

Section 1 defines [E] as a theorem or a fact verified bit by bit by exhausting the finite cases it covers, and in practice the label covers at least four different things. Some [E] claims are theorems with proofs supplied, as with Lemma 1 and the discrete Stokes theorem. Some are genuine exhaustions over a delimited case set, as with the swap geometry checked on every $\chi_1$ register of a test torus, where locality of the rule means exhausting local configurations really does establish the general statement. But the merge structure is graded [E] on "exhaustive checks on small tori" without saying what was exhausted; confluence, which underwrites the axiom that the tick is a causal rather than a temporal order, is graded [E] on six asynchronous schedules at $L = 8$ at a single parameter point over 400 ticks, which is a well-designed test with good negative controls but is a sample rather than an exhaustion; and §5.3 grades [E] a claim about the fundamental group of complements of line families in $S^2\times S^2$ that is neither proved in the paper nor a finite check, despite the requirement that odd charges appear in even numbers resting upon it. The cost is specific: the scheme's value is that a reader can scan for [E], take those claims as secure, and concentrate on the [M] and [D] layers, and that value evaporates if the grade must be re-interpreted case by case. The remedy is cheap. Reserve [E] for theorems with proofs and for exhaustions where the completeness argument is stated, introduce a separate marker for finite verification at specific sizes with the size and parameter point reported inline, and move the topological claims into Appendix C alongside the other derivations.

3.4 The lock is postulated and consequential but never simulated

Section 3.6 attributes the entire static half to a primordial first-order transition that writes the defect pattern and a single ribbon hand and freezes them, with defects forming where bubbles meet as in the Kibble mechanism and the resulting distribution a scale-free fractal. Three major results depend on it: particles exist and are distributed as they are, the arrow of time points as it does, and the single hand makes chirality global. Yet the lock is never modelled — there is no simulation of a nucleating transition, no measurement of the defect distribution it would produce, and no corresponding script in the Appendix D index. What the runs use instead is stated plainly: uniform random substrates described as mean-field coarse cells, and multiplicative cascades with a freely chosen spread parameter. A multiplicative cascade is not a Kibble distribution; the latter makes specific predictions relating defect density to the correlation length at freeze-out and specific defect correlations, none of which the cascade reproduces by construction and none of which are checked. The dependence is not incidental, since §3.5's central result on island persistence varies strongly with the cascade spread, so the spatial coherence that produces integer islands is an input rather than an output. The manuscript's framing that "how the lock writes it is open" understates the gap, which is not merely that the mechanism is unknown but that the stand-in substrates are unconstrained by it. One consequence also deserves explicit treatment: taken with Appendix A's rule that A–B annihilation belongs to the lock itself, §3.6 implies that particle number has been fixed since the lock and that nothing since can create or destroy a particle. This sits poorly with observed physics and with cosmology, and whatever the author's view of it, it belongs in §7 among the open problems, where it does not currently appear.

3.5 The role of the Hopf structure, and the identity of readout space, are left ambiguous

The paper is titled for the Hopf bundle and the abstract has it entering through the geometry of the readout and its global constraints, but tracing where it actually operates is instructive. Section 3.1 establishes that with reversed vertices present the total space is a Seifert fibration over an orbifold rather than $S^3$, and that on the torus it is "never $S^3$" since a genuine circle bundle would require the holonomies to sum to a multiple of $2\pi$. Every transport, localization, propagator and clock measurement runs on that torus, while the single sphere computation uses only ribbon framing and discards the strand structure on which the rest of the paper depends. The Hopf structure thus appears to enter only as the scaffold for flow-line space in §5.1, and there only locally, through the statement that a flow line is a direction and a transverse position — four parameters that any three-manifold would supply. This matters because the global structure is where the most interesting claims must live: §6.3's conclusion that the total charge vanishes relies on readout space being closed, and §5.2 defines readout space as the $S^3$ of the scaffold, the very identification §3.1 shows fails whenever $N_A \neq N_B$, which is exactly the charged case. These need reconciling, either by arguing that the scaffold is a fixed $S^3$ on which the Seifert structure does not affect the incidence geometry, or by redoing the degree arguments of Appendix C on a Seifert manifold. A related point concerns the identification of flow-line space with the space of oriented great circles. The mathematics of that identification is standard and correctly cited, but §5.1 itself notes that causation is diffusive inside matter and ballistic only between islands, with straight incidence an idealization of the latter; since a diffusive path is not a line and the transform integrates over the sphere of lines through a point, the magnitude of the idealization error, and whether the $\rho$ reading inside matter is meaningful at all, should be addressed.

Minor Issues

1. Reference [5] is truncated mid-year, reading "JETP Lett. 38, 552 (198"; the correct year is 1983. Reference [4] is missing its terminal period. More consequentially, reference [16] renders the second author of the Cassini paper as "L. less" rather than L. Iess, which matters because this is the manuscript's only citation to experimental data. All three should be corrected at proof stage.

2. Section 4.5's expression for the mean held ratio is uninterpretable as printed, giving $\sigma_0 = \langle\sigma\rangle \approx 1.05\,\rho\rho$ with the symbol duplicated. From the surrounding statement that it follows the defects, $1.05\,p\,\rho$ appears to be intended. The quantity feeds §6.1's front-speed check, so a reader cannot currently verify those numbers.

3. Equation (5)'s random-placement limit is garbled, printed as $p_{APB} + (1-p_A)(1-p_B)$ with subscripts evidently collapsed from $p_Ap_B$. Since this limit is the only stated check on Eq. (5), the equation cannot be verified as it stands.

4. Appendix E contains a substantive error, listing the boost as "a change of vibration" where §5.4 has "a change of fibration." Because the dependency table is offered as the paper's audit trail, errors there carry more weight than elsewhere, and the whole table would repay a careful pass against the main text.

5. The Appendix D index appears shifted by one row in at least five entries, so that the isotropy result points to the buffer-source readout, the buffer-source and clock-rate results point to the axis-rate readout, and so on through the front-speed row. As the paper's reproducibility interface this table deserves careful rather than mechanical repair.

6. Agreement between $D_b$ and $D_z$ is quoted to five significant figures with no stated error. Section 4.3 gives $1.0585$ and $1.0587$ at $\rho = 0.15$ and describes agreement to $0.02$–$0.5\%$ across delays. If the agreement is exact by construction of the matching procedure, saying so would be clearer than the numerical presentation; if it is measured, an uncertainty is needed.

7. The propagator fit window is justified loosely and its sensitivity is unreported. The excluded band $r < 9$ is described as within one to three mean free paths, a wide range, and since the fitted exponents are $0.966$–$0.969$ rather than unity, the effect of moving the cut to $r = 12$ or $15$ is material and should be stated.

8. The scaling $S \propto \rho^{0.8}$ in §4.5 is inferred from three points with no uncertainty on the exponent. Similarly, the effective-medium correction of 3–10% is reported at two endpoint densities, so the claim that it is set by the area fraction of lattice-scale islands cannot be checked against the data given.

9. Capture ties reach 6–7% of captures at $\rho = 0.3$ and are resolved by the schedule phase $t \bmod 3$, a parameter of the rule rather than a state of any site. At that frequency a rule-level convention is shaping a non-negligible fraction of events, and whether the reported quantities shift under an alternative tie-breaking convention should be checked and stated.

10. Daggered results used the superseded per-axis buffer rule, and while §6.1 offers a useful comparison between rule sets, the cell-machine results of §4.2 are entirely daggered with no comparison given. A statement of whether the rule change is expected to affect them, or a re-run, would close this.

11. The quantum of action $\bar h$ is introduced in Eq. (13) without discussion, simply as the action of one tick. Whether this quantity has any of Planck's constant's properties — appearing in a commutation relation, an uncertainty relation, or a quantization condition — is not addressed, which leaves the mass formula's physical content unclear.

12. Locating the arrow of time at defect merges has a consequence given only a sentence. Since §4.5 establishes that merges occur only at releases, and §3.6 that the space between islands has $p = 0$ at the finest resolution, the construction appears to say that most of space has no thermodynamic arrow. Whether this is intended, and what it implies for the readout, deserves more than the single sentence in §6.2.

13. The commitment to a closed universe is not connected to cosmological observation. Section 1.1 derives a closed base from finite information and assigns its base defects a total deficit of $4\pi$, "the curvature of the sphere spread over that scale." This is a cosmological statement, and current data favour spatial flatness; the relation deserves at least a paragraph.

14. The deposited archive shares the paper's own DOI with no version or commit identifier, and while seeds are said to live in the module docstrings, none appear in the paper. Given how strong the reproducibility apparatus otherwise is, a versioned archive reference and the seeds for the headline results would complete it.

15. The acknowledgments list six bodies of work "used here without being cited in the text." Influences that shaped specific claims should be cited at those claims so that the debts are traceable, rather than collected in an acknowledgment where a reader cannot tell which idea came from where.

Missing and Inadequately Integrated Literature

Cited but Inadequately Integrated

The manuscript's engagement with the lattice-gas literature stops short of its most useful content. Frisch, Hasslacher and Pomeau (1986) [1] is cited for the six-channel exclusion structure, but the central lesson of that programme — that lattice gases reproduce hydrodynamics only after careful attention to spurious conserved quantities, and that unnoticed invariants can silently invalidate a long-wave description — goes unaddressed. The present machine introduces several new exact invariants in §2.3, and the claim that coarse-graining retains only $n$ and the fiber pair depends on those invariants being complete; an explicit argument that they exhaust the conserved quantities of the update would close a real gap. Ruijgrok and Cohen (1986) [4] and Wang and Cohen (1995) [3] are cited to place the route network as a deterministic Lorentz lattice gas, but the detailed results in that literature on orbit statistics, loop ensembles and sensitivity to scatterer arrangement bear directly on §3.2's loop-soup measurement and on its claim that two arrangements sharing three integers have the same phase despite transfer lengths differing by up to 30%.

The localization literature is similarly invoked more than used. Khmelnitskii (1983) [5] and Pruisken (1984) [6] supply the two-parameter flow whose shape the phase diagram is said to have, and that framework makes a quantitative prediction for the logarithmic running of the diagonal conductance — precisely the slope the manuscript has measured. Comparing the measured value against the prediction, even to show disagreement, would convert an appeal to shape into a test. Chalker and Coddington (1988) [7], Beamond, Cardy and Chalker (2002) [8] and Cardy (2010) [9] come with established exponents for network models and for their classical reductions, which is the class the present model explicitly occupies; since the author has strip-transfer data in hand, measuring an exponent and comparing would be more informative than leaving the question entirely open in §7.

On the geometric side, three citations deserve fuller use. Gluck and Warner (1983) [11] is cited for the Grassmannian identification, but its substantive content concerns which great-circle fibrations of $S^3$ exist and how they deform — directly relevant to §5.4's claim that a boost is a change of fibration, which currently has no concrete realization. Helgason (2011) [13] supplies inversion formulas, support theorems and range characterizations for Radon-type transforms, which bear on whether the manuscript's reading is invertible; that is arguably the most natural question to ask of an integral-transform account of emergence, since invertibility would determine how much of the machine the four-dimensional description captures. Penrose (1967) [12] is invoked for the double fibration's shape, but the twistor programme's power lies in relating cohomology on twistor space to solutions of field equations, and since the manuscript has a double fibration but no field equations, stating what the counterpart output would be — even as a target — would clarify the programme considerably. Finally, Kibble (1976) [10] is the model for the lock, and as discussed above the integration currently stops at the name.

Truly Missing Papers

Weinberg, S., and E. Witten (1980): Not cited. This paper proves that a theory with a Lorentz-covariant conserved stress tensor admits no massless particle of spin greater than one, composite or elementary, and with a Lorentz-covariant conserved current no charged massless particle of spin greater than one half. It is the standard no-go result confronting programmes that derive gravity as an emergent phenomenon, and a manuscript claiming to read gravity out of a lattice automaton should state where it sits with respect to it — most plausibly by arguing that its emergent Lorentz covariance is approximate rather than exact, which is itself a substantive claim about the lattice scale and connects directly to the anisotropy measurements of §5.4. [Verified via WebSearch.]

Kadanoff, L. P., G. R. McNamara and G. Zanetti (1989): Not cited. This study tested a lattice-gas automaton against its predicted hydrodynamics and found a logarithmic discrepancy in the viscosity that was resolved only after three previously unnoticed conserved quantities were incorporated. The parallel to the present paper's logarithmic conductance, and to its reliance on having enumerated the machine's invariants, is close enough that the episode should be discussed. [Verified via WebSearch.]

Will, C. M. (2014): Not cited. The standard review of experimental tests of general relativity and of the parametrized post-Newtonian framework, within which the manuscript addresses one of ten parameters. It would let the author state precisely which parameters the construction fixes, which it leaves free, and which current data would falsify, converting the present single-parameter comparison into a phenomenological programme. [Verified via WebSearch.]

Wolfram, S. (2020): Not cited. The most extensive recent treatment of emergent space, time and relativistic structure from a discrete rewriting system, including an explicit treatment of causal invariance — the property that different update orderings yield the same causal structure, which is exactly what Appendix A's confluence check establishes here. The overlap in both aims and technique is close enough to warrant discussion, if only to distinguish the two approaches. [Verified via WebSearch.]

Hardy, J., O. de Pazzis and Y. Pomeau (1973): Not cited. The HPP model is the original deterministic lattice gas with exclusion and collision rules and the direct ancestor of the cited FHP model. It is also the canonical cautionary case, since its square lattice carries a spurious anisotropic invariant that prevents it from reproducing hydrodynamics — the most relevant precedent available for a construction that introduces new exact invariants and depends on their completeness.

Frisch, U., D. d'Humières, B. Hasslacher, P. Lallemand, Y. Pomeau and J.-P. Rivet (1987): Not cited. The detailed companion to reference [1] and the standard source on the lattice-symmetry conditions required for isotropic transport tensors. Section 5.4's measurement that the propagator is raised along the cubic axes at short range, with a fourth-rank anisotropy decaying as $1/t$, is a statement about exactly these tensors, and the established criteria would let the author say whether the cubic lift satisfies them by construction or only asymptotically.

Levin, M., and X.-G. Wen (2005): Not cited. String-net condensation is the best-developed mechanism by which an emergent U(1) gauge structure, with its Gauss law and with propagating excitations, arises from a local lattice model. Given that the manuscript's charge sector is a lattice-based emergent U(1) possessing a Gauss law but no dynamics, this is the natural comparison class.

Jacobson, T. (1995): Not cited. The derivation of the Einstein equation as an equation of state is the standard benchmark for obtaining gravitational dynamics, rather than a Newtonian potential, from a non-gravitational substrate, and would help situate how far the present construction has to travel. Sakharov (1968), likewise uncited, is the earliest proposal that Newton's constant is a derived stiffness rather than a fundamental parameter, and Eq. (14) — which writes $G$ as $c^4$ over a conductance of the flow — is an instance of the same move.

Reference list

Frisch, U., D. d'Humieres, B. Hasslacher, P. Lallemand, Y. Pomeau, and J.-P. Rivet. 1987.
    Lattice gas hydrodynamics in two and three dimensions. Complex Systems 1:649-707.
Hardy, J., O. de Pazzis, and Y. Pomeau. 1973. Time evolution of a two-dimensional model
    system. I. Invariant states and time correlation functions.
    Journal of Mathematical Physics 14:1746-1759.
Jacobson, T. 1995. Thermodynamics of spacetime: The Einstein equation of state.
    Physical Review Letters 75:1260-1263.
Kadanoff, L. P., G. R. McNamara, and G. Zanetti. 1989. From automata to fluid flow:
    Comparisons of simulation and theory. Physical Review A 40:4527-4541.
Levin, M., and X.-G. Wen. 2005. String-net condensation: A physical mechanism for
    topological phases. Physical Review B 71:045110.
Sakharov, A. D. 1968. Vacuum quantum fluctuations in curved space and the theory of
    gravitation. Soviet Physics Doklady 12:1040-1041.
Weinberg, S., and E. Witten. 1980. Limits on massless particles. Physics Letters B 96:59-62.
Will, C. M. 2014. The confrontation between general relativity and experiment.
    Living Reviews in Relativity 17:4.
Wolfram, S. 2020. A class of models with the potential to represent fundamental physics.
    Complex Systems 29:107-536.

Conclusion and Path Forward

The construction assembled here contains more that is worth keeping than the length of this report might suggest. An automaton with a topological register that composes exactly under every doubling of block size, a routing table derived from ribbon topology rather than assigned, an exact C-odd moment per defect, a measured transport law with a demonstrated locality property, and a recasting of Gauss's law as an intersection number that cleanly excludes fractional charge from the readout — these are substantive and they are, for the most part, well established by the evidence presented. The verification apparatus around them, graded claims and a dependency table and a complete script-to-result index with a deposited archive, is better than the norm. The difficulty is that the bridge from this machinery to four-dimensional physics rests on joints that the manuscript has not yet secured: one postulate whose consequences are asserted in prose rather than derived, a transform that is defined but evaluated in neither channel and whose connecting analytic step is assumed, a $1/r$ law that any isotropic three-dimensional diffusion would reproduce, a pair of direct machine tests that returned null results and were then relocated to the readout, and a set of numerical conclusions reported without the uncertainties needed to assess them.

A revision could take either of two routes, and both seem viable. The narrower one would scope the paper to what has been demonstrated — an automaton with an exact topological layer and a measured diffusive transport half, with the physical readings presented explicitly as a programme — which would make a solid contribution to the statistical-mechanics literature without the burden of the four-dimensional claims. The broader one would keep the present scope and supply the missing core, and much of what that requires appears tractable with the apparatus already built. A formal uniqueness argument for the even split would turn §5.4 from the paper's shortest subsection into its technical center. Evaluating the transform directly on the single-closure configuration of §6.4, in both channels, would test the paper's unifying claim rather than framing it. Measuring the integrated propagator from a compact defect source into clean flow would join the two halves of the gravitational argument that are currently established in incompatible regimes. Supplying error bars and realization counts throughout, splitting the exactness grade, and stating plainly where the construction is linear and what that costs it in post-Newtonian reach would bring the reporting into line with the care evident elsewhere. The question the work asks — whether local causal structure, gravity and charge can be read off a fixed combinatorial substrate by a single integral transform — is a serious one, and the author has built enough apparatus to ask it precisely. That is not a small thing, and it is worth the additional work needed to answer it convincingly.

Author response

Thank you for a thorough second report. Brief replies follow; section numbers refer to version 2, and the changes described will appear in the next version.

Category 1

1.1 Equipartition. The next version will state the uniqueness formally: a swap credits a share s of its tick to time and 1 − s to depth on one outgoing bit, and the reverse on the other; exchanging the identities of the two outgoing bits maps s to 1 − s, so a reading that does not depend on the assignment is invariant under this involution, whose only fixed point is s = ½. The static register cannot separate time from depth: it is a function of the defect pattern alone, fixed after the lock, and holds no history of any bit.

1.2 The transform. Both channels have now been evaluated on the machine as one walk sum of the flow. For ρ, with the walk measure written out, R_ρ is the delay integral itself, and the delay field around buffered sources has been measured directly (2.3). For ψ, each walk that ends at a readout point is weighted by the frame holonomy it sweeps around a closure. With half turns the coherence falls to 0.02–0.09 on the line behind the closure and stays at 0.88–0.93 in front: the marked flow line of Gauss's law. With whole turns it is 1 everywhere. Neither reading needs the stationary-phase concentration, which stays open only for phases that vary along a line. The half-turn reading has also been derived: relative to the free flow, each passage through a reversed vertex is a strand exchange, and a closed route of turning number T passes n ≡ T (mod 2) of them, so ψ changes sign around an odd charge. Both will be in the next version.

1.3 Agreement with general relativity. It is structural, not fitted, and the next version will say so. The data are linear: held ticks add, and two sources at once leave each other's held content and propagation unchanged. The reading is not: each link of a causal chain is read against the clock the earlier links have set, and each reading recombines the even split of time and depth. Read this way the static field is the exterior Schwarzschild metric in isotropic coordinates at every order; in the post-Newtonian language γ = β = 1, and α₁ = α₂ = 0 because readings are relational and the structure never moves. What could have come out otherwise is the machine's part: the linearity of the data, D_b = D_z, and K_src = 1 around sources.

1.4 The gauge sector. Version 2's reading of the depth share as a static magnetization was wrong. The C-odd moment of §3.4 is a circulation of the causal flow on the base, and a quantity of flow-line space has no reading in R³ until the incidence transform is applied; version 2 skipped the transform. The next version will read a static charge entirely as E, with B = 0 in its rest frame; B appears only when a boost re-splits the time–depth axis, and E and B then mix by the Lorentz transformation of the field strength, B = v × E/c² to first order. The force on a test charge, the coupling and the field equations beyond Gauss's law will be listed as open: this paper treats the topological half.

Category 2

2.1 Discrimination. The 1/r is generic to isotropic three-dimensional diffusion; what the machine contributes is that the walk is three-dimensional and isotropic at all, with the fiber climb as the third direction and D_b = D_z exactly at the even matching. Outcomes that would have falsified the reading: D_b ≠ D_z at the even matching, K_src ≠ 1 around a buffered source, a second source changing the first one's field, or a ψ sign around a charge with whole turns. The fit window will be reported: moving the lower bound from r = 9 to 12, 15 and 20 leaves A at 0.995–0.997 (L = 192), while the bare exponent rises from 0.966 to 0.992 as the 1/r² correction fades.

2.2 Time dilation on the machine. The uniform growth on the torus is a finite-size effect: 2.0 × 10⁻³, 5.2 × 10⁻⁴ and 1.4 × 10⁻⁴ per tick at L = 48, 96 and 192, falling as L⁻². The excess at r = 25 came from eight realizations; with 64 (L = 96) and 16 (L = 192), the front is late by 0.1–0.3 ticks at every radius, as much as the hold that the arriving bits carry out of the source.

2.3 Source and propagator together. Measured as suggested: each release is read as a point of the lift, and the released bits are followed into clean flow. K_src = 4π√(D_b D_z) r G_src(r), averaged over r ≥ 20, is 0.98–0.99 for a closure at three densities and 0.98 for an island at two, with the released bits' own D; with the D at which they agree with the identity propagator, it is 1–3% lower, and a longer record now running will settle that difference. At the lowest density an island gives 0.93 ± 0.02, which is not resolved. Section 4 will open by stating that its uniform substrates are mean-field cells of matter and that the free flow between islands is their p = 0 limit.

2.4 Lattice scale. A single bit carries no signal past the front: it outruns the front only between collisions, by about two lattice constants at ρ = 0.15, below the scale at which readings begin; and bits are anonymous, so the bit that leaves a collision is not the one that entered. On the preferred frame: the structure never moves, relative frames appear only after the transform, and readings are relational, so no reading depends on a velocity relative to the medium, whose rest frame is a property of the state. That observers are defect patterns is not a difficulty but the condition for reading at all: only matter, where the flow is held and released, interacts causally in a nontrivial way, and a reading carries exactly that hold, its delay and the split of the time–depth axis. Reading and causal interaction are therefore one thing by definition. The reading is of the flow lines through the observer, which arrive through the free flow; the observer's matter sets its clock rate, a rate-layer quantity, while the signature comes from the even split of the swaps in the flow. The lattice constant and the tick in physical units remain open, with the scale of G.

Category 3

3.1 Statistics. Error bars and realization counts for g(b), the propagated uncertainty of ξ and the seed scatter of η need reruns at L = 1024; they remain open and are left for later work.

3.2 The coarse-graining closure. A quantitative closure criterion, a rerun of the cell machine under the current rules and a scan of b* across the grid remain open and are left for later work.

3.3 Grades. In the next version [E] will be reserved for theorems and for checks that exhaust every case the claim covers, and finite-size bit-exact checks will be graded [M]. Confluence keeps [E] with a proof: the tick decomposes into local units, each register is written once per tick by one unit, each unit is deterministic, and the tie phase is the unit's own tick count, so by induction along the dependency order every schedule that respects it gives the same values. The claim on the Z₂ bundle on flow-line space will be proved in App. C: for one source the lines through it form a sphere with a dual sphere, so the complement is simply connected; for several, Alexander–Lefschetz duality gives H₁ = Z^k/Z(1,…,1), so the sign can flip only around an even number of them.

3.4 The lock. Out of scope; it will be listed as open.

3.5 The Hopf structure and readout space. The Seifert structure belongs to the static half, the routing and its point holonomies; readings belong to the dynamic half, whose geometry is the fixed S³ on which reversed vertices are data. The transform, readout space and the degree argument of App. C live on that S³ and need not be redone on a Seifert manifold. Diffusive paths are not lines, and the next version evaluates both channels as full walk sums rather than on straight lines.

Minor issues

1–5. These do not occur in the source and appear to come from text extraction of the PDF: the references read correctly, σ₀ ≈ 1.05 pρ, p_A p_B, "a change of fibration", and the rows of App. D are aligned.
6. The agreement of D_b and D_z is measured, 0.02–0.5% across delays, and will be reported as such.
7. See 2.1.
8. The wording will be revised: the exponent of S stated as approximate, and the effective-medium correction reported as measured at the two densities, without a claim about its cause.
9. Whether the reported quantities shift under another tie-breaking convention remains open.
10. See 3.2.
11. ħ is used only as the action of one tick, to fix units; no quantum property is claimed for it here.
12. The free flow between islands makes no merges and so no order of its own; but every observer is a defect pattern, so every reading is made where merges happen, and the free flow carries order between readings without making it.
13. A closed base of cosmological size reads as flat on every scale much smaller than its own; the observed flatness bounds its size from below rather than excluding closure.
14. App. D will give the commit of the deposited archive and the seeds of the main results.
15. We cite in the text the work in direct lineage with a claim. Work from which we borrowed an abstract concept, or parallel programmes that are ontologically different, is acknowledged rather than cited at a claim, since citing it there would suggest a dependence that does not exist.

Literature

The lesson of the lattice-gas programme on spurious invariants applies directly, and the machine has them: a route's circulation is the strand difference on its track and collision swaps conserve it, so in defect-free flow the strand difference is sealed route by route. The next version will state this. The long-wave description of the strand difference holds inside matter, where reversed vertices exchange circulation among routes; it is not carried by the free flow. Weinberg–Witten does not apply: the programme has no graviton. Gravity is the static potential reading of the ρ channel, a positive magnitude with no winding to quantize, unlike the 2π winding that ψ carries. The other suggested references are not added.

XSci (AI review)

October 5, 2026

Referee Report

Manuscript: "On a Routing Network of Causal Flow: Incidence Transform from the Hopf Bundle to Four-Dimensional Spacetime" Field: Theoretical and mathematical physics (deterministic lattice gases, emergent spacetime, integral geometry) Recommendation: Major revision required

Summary and Recommendation

This manuscript proposes that light, gravity and electric charge in four-dimensional spacetime can be read out of a deterministic lattice-gas automaton running on a fixed trivalent network on the base of the Hopf fibration. The machine has six entry registers per vertex, a fixed routing table, a sparse population of reversed routers whose table is mirrored, and two events: a collision swap on an edge and a hold in a reversed router's one-pair buffer. Coarse-graining is argued to split into a static half, whose topological integers compose exactly at every resolution, and a transport half, which keeps two conserved densities. A single incidence transform then reads the collective front as light at $c = 0.354$ lattice constants per tick, the static offset of the delay field as the Newtonian potential, and the point holonomy $\kappa_v = \pm 2\pi/3$ of reversed vertices as charge, with $\gamma = \beta = 1$ and the exterior Schwarzschild metric recovered for a single source.

The manuscript has genuine strengths. Appendix A specifies the machine completely enough to be reimplemented; the confluence argument is clean and is supported by two negative controls that fail as they should; Lemma 1 is elegant and does real work; the index law, the discrete Stokes theorem of Eq. (3) and the exact C-odd moment $-\sqrt{3}(N_A-N_B)$ are attractive independently of the physics read onto them; and the $\mathbb{Z}_2$ bundle computation in Appendix C is a correct argument with a non-obvious conclusion. Appendix B's disclosure that an earlier estimator produced a spurious $r^{-1.3}$ law increases confidence in the authors' care. The difficulty is that the headline claims concern the emergence of relativistic spacetime, and at every place where the construction turns from lattice statistics into physics, the supporting argument is markedly weaker than the surrounding apparatus.

Major Concern Category 1: The derivation of relativistic structure

1.1 The equipartition fixed point presupposes what it produces

The paper's most consequential claim is that the Lorentzian signature, and with it $1+\gamma = 2$, rest on no postulate but follow from the anonymity of bits. The argument in §5.4 lets a swap credit a share $s \in [0,1]$ of its tick to time and $1-s$ to depth on one outgoing bit and the reverse on the other, concluding that $s = \frac{1}{2}$ is the flip's only fixed point. That step presupposes an affine space of time and depth, a tick divisible between them, and a sum rule tying the shares together, none of which follows from the premise. Appendix E concedes the companion step, listing the boost as "linear, reciprocal" in the reading column, which is where the invariance of the interval actually comes from; that entry and §1.1's claim that the signature rests on no postulate cannot both stand.

1.2 Light is read from a parabolic transport law, and single bits outrun the front

Section 6.1 identifies light with $\sqrt{\langle(\mathbf{v}\cdot\hat{n})^2\rangle}$ and obtains $c = v_{\mathrm{route}}/\sqrt{2} = \mathrm{STEP}$. As Appendix C makes explicit, this is a root-mean-square velocity component of individual bits, a kinetic-theory moment rather than a front velocity; no wave equation is derived and no sharp front is exhibited. The transport half is meanwhile diffusive, with Eq. (7) parabolic and sound overdamped beyond $b^* \approx 14$, so a coarse-grained observer governed by that equation has a Green's function of unbounded support and no light cone at all. Section 6.1 further concedes that single bits move at $\sqrt{3/8} \approx 0.612$, about $\sqrt{3}$ times $c$, defended only by a short free path.

1.3 Schwarzschild follows from an ansatz, and no prediction distinguishes the construction

Equation (17) is reached by adopting a conformally flat static form, fixing its exponent by requiring vanishing scalar curvature, requiring a harmonic lapse-times-conformal-factor, and assigning the two available harmonic functions to the two shares of the even split. This is a classical uniqueness statement; the machine contributes only $U \propto 1/r$ and the even split, so any construction supplying those reproduces Eq. (17) by the same route. The admission that follows is sharper: for several sources the same factors "satisfy the two scalar equations but not the static ones," and §7 lists both the field equations and the multi-source reading as open. Wherever the construction touches observation it matches established theory and defers every possible divergence, leaving nothing on which it could be found wrong.

1.4 The fiber is assigned incompatible roles and the dimension bookkeeping does not close

Depth is defined twice in consecutive sentences of §5.2, first as the fiber height, a single global direction shared by all observers, and then as the radial distance $r$ of Eq. (6), which the same sentence describes as combining base displacement and fiber height and which is by construction observer-dependent. Section 5.4 commits to the first, declaring that depth is "the body diagonal shared by all three" cubic axes, while §6.2's spherically symmetric line element commits to the second; a single global direction cannot be the radial coordinate of a metric centred on an arbitrary source. The dimension counting inherits the confusion, since §5.4 accounts for signature $(1,3)$ first as two sky directions plus depth plus time and then as three cubic axes plus time.

Major Concern Category 2: The evidential base

2.1 Replication is too thin for the precision claimed

Many quoted numbers rest on one to four realisations while being reported to three or four significant figures and compared with theory at the percent level. Table 1's caption states "one to two records per point, jackknife errors 3–8%," but a jackknife over a single record is undefined and over two yields one degree of freedom, yet the table carries entries such as 1.852, 1.941 and 2.031. The fit $1/D_n = a(p)+\beta(p)\rho^2$ is applied to rows of which three of seven have only three measured entries. The pattern recurs where the physics is read: the two-source test licensing $U = \sum_i U_i$ is paired over four realisations, and the $\psi$ walk sum, the only machine evidence for the charge reading, rests on four each.

2.2 The $1/r$ law survives only after a fitted correction and an assumed tail

Appendix E lists this result as resting on machine facts alone, so it warrants scrutiny. The measured $K$ spans 0.80 to 1.06 over $r = 9$–92, a band many near-$1/r$ forms would produce. A bare power fit gives exponents of 0.966 to 0.969, and unity is recovered only through $K \approx A(1-7/r^2)$ with $A = 0.998$, whose coefficient is fitted rather than predicted. More seriously, Appendix B completes the tail as $t^{-3/2}$ and §4.4 reports that 9 to 24 percent of the source field at $r \geq 20$ is supplied that way, in a form implied by the hypothesis under test. Where the measurement visibly fails, around a twelve-vertex island at $\rho = 0.075$, the field averages $0.87 \pm 0.01$ and the matter is left unresolved, although islands are what §3.6 reads as particles.

2.3 The phase diagram reports a finite-size crossover as a phase boundary

Section 3.3 concludes that no stable diffusive phase exists, on the strength of slopes $dg/d\ln b$ of $-0.036$ to $-0.044$ over $b = 8$ to 256, three and a half $e$-foldings and a total predicted change near 0.15. No uncertainty accompanies any slope and no value of $g$ is quoted, so the change cannot be judged against the scatter, and the report that the sparsest substrates first rise to $b \approx 32$–64 before falling is characteristic of a quantity short of its asymptotic regime. The candidate boundary of Fig. 3 is more directly problematic, since the paper itself describes it as "the line where the localization length matches the measurement size." Such a locus moves as $W$ grows and is not a phase boundary.

2.4 The specification and its verification leave three gaps

Section 4 notes in one sentence that results marked with a dagger used "the earlier per-axis form of the buffer rule," and Appendix D reveals the cell-machine analysis among them. That analysis establishes that the cell state composes exactly under $b \to 2b$, the demonstration that coarse-graining closes and hence the premise of the two-halves architecture, and it also locates $b^* \approx 14$; a per-axis buffer gives different capture statistics and held ratio, so how far the result carries over is unknown. Second, §1.1 lists shared time first among the ingredients not presupposed, yet the tie rule selects from axis $t \bmod 3$ and the confluence proof requires "the unit's own tick count modulo 3"; ties reach 6–7 percent of captures and §7 concedes their effect on transport is unknown. Third, the exact grade covers both the proved Stokes theorem and searches on an $L = 3$ torus, and nothing marks extrapolation from $L \leq 1024$ to a cosmological base.

Major Concern Category 3: The readout and the physics it delivers

3.1 Four inequivalent spaces are used interchangeably

The construction invokes a trivalent network on $S^2$ as nominal base, a honeycomb torus as the base of nearly every run, a Seifert fibration over an orbifold as the total space wherever reversed vertices are present, and a fixed $S^3$ scaffold as readout space. Section 3.1 assigns the last two to different halves, declaring that the Seifert structure belongs to the routing while readings live on a fixed $S^3$ on which reversed vertices are data. That is intent rather than construction, since no map is given from a Seifert fibration of Euler number $(N_A-N_B)/3$ to the scaffold, although Appendix C's degree argument crosses exactly that boundary. The sphere-to-torus relation fares no better, since §4.3 concedes that only the Goldberg-sphere runs see the global Hopf structure named in the title.

3.2 The incidence transform is defined and then bypassed

Section 5 declares that every reading is the output of one transform, $R = \pi_{2*}\pi_1^*$, yet §6.4, the worked reading, performs no integral over the sphere of flow lines. The potential channel is evaluated as the time-integrated diffusive propagator at a distance, and §5.3 concedes the point in advance by stating that this channel "needs no further step from the transport half"; the charge channel is evaluated as a walk sum. Both objects live in the cubic lift rather than on flow-line space. The measure sharpens the difficulty, being described as the floor's walks arriving at the readout point, since a diffusive path is not a flow line and pushing a measure from path space to a two-sphere needs a map the paper does not construct. The antisymmetric density, built with care in §5.3, meanwhile never appears in Eq. (11).

3.3 Quantum vocabulary is applied to a construction without quantum structure

The charge channel attaches a complex weight $e^{i\theta/2}$ to an otherwise positive classical walk measure, and from this the paper obtains interference, coherence, a spin structure, an Aharonov–Bohm sign and an orbital winding. The construction is internally consistent but nowhere derived: the machine has binary occupancies, a deterministic update and a real holonomy, the half-integer exponent is justified only as the smallest at which a single reversed vertex leaves no sign, and §3.3 states outright that the mechanism is classical because routes carry no amplitude. Fringes require an amplitude whose modulus squared is observable, whereas Fig. 9 plots a ratio of two classical sums labelled coherence, and the action quantum, introduced as "the action of one tick," is a unit conversion rather than a determination.

3.4 The matter sector is impoverished and particle number cannot change

Section 3.6 states that particles are primordial and fuse only by a change of resolution, and §7 restates that their number is fixed from the lock onward. As stated this conflicts with routine observation, since pair production, nuclear decay and hadronisation all change particle number, and §3.5's finding that two islands always stay separate on the network, merging only when a coarser cell swallows both, makes fusion an artefact of reading rather than a process. Beyond number there is one charge species with two signs, no family structure, no mass spectrum and no weak or strong interaction. The fractional charge is suggestive, but §6.3 concludes only that a fraction has no Gauss-law reading, which is invisibility rather than confinement; that section is titled "Charge and electromagnetism" while deriving one of Maxwell's equations and no force law.

Minor Issues

1. Equation (15) writes $c_{\mathrm{eff}} \simeq c(1-(1+\gamma)U)$ with the post-Newtonian parameter already in place, although the left-hand side expands unconditionally to $c(1-2U)$, so $\gamma$ is identified by matching rather than derived at that step. Writing the expansion plainly would remove an appearance of circularity.

2. No machine measurement corresponds to Eq. (16), and the measurement bearing on it tests a different effect: §4.4 reports a front arriving late by 0.1 to 0.3 ticks at $r = 6$–18, "as much as the hold that the bits arriving there carry out of the source, 0.17 ticks at every radius." That is a transported offset, constant in radius, whereas the Shapiro delay accumulates along the path.

3. Section 6.2 asserts that the reading has no preferred frame, while §5.4 concedes that diffusion singles out the medium's rest frame and Appendix B takes readings in that frame. Since the absence of preferred-frame effects is a post-Newtonian claim, the reconciliation belongs in the text; the notation for the action quantum likewise drifts across Eq. (13), Eq. (14), the prose and Appendix E, and should be unified.

4. Several quantities are described more confidently than their uncertainties permit. Shell means of $0.73 \pm 0.08$ are presented as confirming a Table 1 value of 0.746, and the strand-difference diffusion constant is called nearly independent of both parameters while ranging from 0.07 to 0.10. Borrowed terms likewise need operational definitions, since the coherence of Fig. 9 is a ratio of a complex-weighted to an unweighted classical walk sum.

5. The reproducibility apparatus stops short of being actionable. The dagger convention is defined once in §4 and resolved only in the Appendix D index, the code is identified by a commit hash but no canonical repository, and the driver script runs every script once at small size without tolerances, so a reader cannot tell whether a reproduction succeeded.

Missing and Inadequately Integrated Literature

Cited but Inadequately Integrated

The Cassini measurement is cited only to note consistency with $\gamma = 1$, but §6.2 claims more, namely $\gamma = \beta = 1$ with no preferred frame, which in post-Newtonian language implies vanishing values for further parameters constrained independently by lunar laser ranging and pulsar timing. The authors should state which parameters their construction fixes and whether any could differ from the general-relativistic value, since that is where a falsifiable prediction would live.

Several references are invoked for shape rather than content. Chalker and Coddington, Khmelnitskii and Pruisken supply the two-parameter-flow picture §3.3 borrows, but the mechanism here is classical, so the relation to the classical limit is load-bearing and left as an analogy, while Beamond, Cardy and Chalker, and Cardy, are cited for precisely that reduction yet their exponents go unused although §7 lists the exponents as open. Wang and Cohen are cited to identify the route network as a rotator model without comparing their quantitative results to the new measurements, and Frisch, Hasslacher and Pomeau are cited for the six-channel structure although that literature's central lesson concerns which lattice symmetries make the coarse-grained equations isotropic, exactly the question §4.3 confronts. Gluck and Warner, Penrose and Helgason are named but never used technically, although establishing whether the transform is injective would turn §6.4's observation that the potential reading does not image its source into a theorem. Finally, Bilson-Thompson, 't Hooft, Susskind and Guo, Papanicolaou and Varadhan appear only in the acknowledgments, though the first is the closest existing proposal to this one.

Truly Missing Papers

Four omissions bear on the relativistic claims and on the lattice-gas tradition the paper joins. Will (2014) is the standard reference for the parametrised post-Newtonian framework and its experimental status, which §6.2 is implicitly using when it reports $\gamma$ and $\beta$. Collins, Perez, Sudarsky, Urrutia and Vucetich (2004) argue that a Planck-scale preferred structure generically induces percent-level Lorentz violation absent fine tuning, even when the fundamental classical equations are locally Lorentz invariant, which a construction with an explicit lattice and medium rest frame faces directly. Ziff, Kong and Cohen (1991) study the Lorentz lattice gas as a kinetic walk, with exponents from a million trajectories comparable with §3.2's loop soup, while Bunimovich and Troubetzkoy (1994) establish absence of diffusion for the flipping rotator model, the rigorous counterpart of §3.3's numerical conclusion about a model §2.3 identifies as a rotator model.

Four further omissions bear on the lift, the automaton tradition and the gravitational reading. D'Humières, Lallemand and Frisch (1986) solved the isotropy problem for three-dimensional lattice gases by projecting a four-dimensional face-centred-hypercubic model down, precisely because no suitable three-dimensional lattice exists, which makes the cubic lift closely analogous and bears on §4.3's residual anisotropy. Gorard (2020) proves that causal invariance under changes of updating order is equivalent to a discrete form of general covariance, which is this paper's confluence property given a physical interpretation. 't Hooft (2016) develops the proposal that quantum mechanics is the effective description of a deterministic automaton, and Jacobson (1995), with Sakharov's induced-gravity proposal, supplies the lineage of Eq. (14) and a route to the field equations §7 leaves open.

Bunimovich, L. A., and S. E. Troubetzkoy. 1994. J. Stat. Phys. 74:1-10.
Collins, J., A. Perez, D. Sudarsky, L. Urrutia, and H. Vucetich. 2004. Phys. Rev. Lett. 93:191301.
d'Humieres, D., P. Lallemand, and U. Frisch. 1986. Europhys. Lett. 2:291-297.
Gorard, J. 2020. Complex Systems 29:599-654.
Jacobson, T. 1995. Phys. Rev. Lett. 75:1260-1263.
Sakharov, A. D. 1967. Dokl. Akad. Nauk SSSR 177:70-71.
't Hooft, G. 2016. The Cellular Automaton Interpretation of Quantum Mechanics. Springer.
Will, C. M. 2014. Living Rev. Relativity 17:4.
Ziff, R. M., X. P. Kong, and E. G. D. Cohen. 1991. Phys. Rev. A 44:2410-2428.

Conclusion and Path Forward

The lattice-statistical and topological core of this manuscript is good work. The exact invariants of §§2.3 and 3.1–3.4, the discrete Stokes theorem, the intersection-number reading of Gauss's law, the $\mathbb{Z}_2$ bundle computation, the confluence argument with its failing controls, and the demonstration that a half-turn weighting produces a zero line behind a closure where a whole-turn weighting does not, all stand on their own. The concerns fall into three groups. The arguments that turn the machine into spacetime each presuppose or supply by fiat something essential, and the paper's own appendix table concedes one of them as a reading rather than a derivation. The evidential base is thinner than the precision of the conclusions, with replication often of one to four realisations, a $1/r$ law needing a fitted correction and an assumed tail, a phase boundary that is a finite-size crossover by the authors' own description, and the closure result obtained under a superseded buffer rule. And the readout apparatus treats four spaces as one and is bypassed at the only point where a reading is computed.

Most of this is addressable, much of it with the code already in hand. The signature argument could be derived from the composition of readings among three observers rather than from a one-parameter fixed point, and the causal-structure claim tested by measuring how a localised perturbation's influence falls off with separation at fixed elapsed ticks. Per-entry uncertainties valid for the available records, the propagator amplitude reported with the tail completion varied and omitted, increased replication on the two-source test and on the island at the lowest density, and a finite-size scaling collapse of the transfer ratio would together transform the empirical chapter. The cell-machine analysis should be rerun under the published rules, the tie convention varied, the Seifert-to-scaffold relation constructed rather than declared, and the several roles of the fiber separated. Most important for the paper's standing as physics, at least one quantity should be identified on which the construction differs from general relativity or classical electromagnetism. The question this paper asks is a serious one, and the work already done suggests it could be answered far more convincingly than the present draft does.

Author response

Thank you for a careful third report. Brief replies follow; section numbers refer to version 3, and the changes will appear in version 4.

Category 1

1.1 The shares are counts: a tick is one unit of content and the climb is counted in ticks, so the shares add and their sum is a tick's content, conserved at a swap. The signature rests on the even split alone. Every observer reads light, which advances along the unsplit axis, at r = ±ct, so the time–depth plane carries two real null lines common to all observers, and a quadratic form with two real null lines has signature (1,1). The depth share spreads over the three cubic axes with equal weight, since the climb runs along the fiber, the body diagonal they share, while time stays on the clock: signature (1,3). Linearity and reciprocity fix only the form of the boost: composed among three observers, as you suggest, the re-splits form a group, each diagonal in u = ct − r and v = ct + r, and reciprocity sets the determinant to one. Version 4 separates the two.

1.2 Version 3 set c equal to the rms of one velocity component and measured no front; since then we have measured both fronts on the base. The causal front of a local change (two copies of one state that differ in one register, run in lockstep) expands ballistically in every medium; its edge moves at a speed set by the medium, 0.40–0.43 in defect-free flow and 0.325–0.353 inside matter, and inside it the two copies are fully decorrelated, so it is not a readable signal. The readable front is the mean response to an injected density packet. In defect-free flow it propagates as a ring at the causal wave speed v_route/√2 = 0.354 (peak 0.338–0.343, outer half-maximum 0.360–0.373) at both densities measured, 0.15 and 0.5, ballistically to the longest time measured (rms² ∝ t²): collisions there conserve momentum, so the coarse law of the floor is a sound law. Inside matter, where reversed vertices take up momentum, the packet diffuses, and the parabolic law of §4 is measured there. Both are speeds on the base, and neither is light: light and its speed c belong to readout space, which appears only after the cubic lift and the incidence transform. Version 3 identified c with the causal wave speed through their common value; version 4 derives c from it in readout space: ρ and ψ are two readings of one walk sum, and the constancy of the speed every observer reads comes from the even split of the time–depth axis (1.1).

1.3 The claim concerns the inputs of the uniqueness theorem: 1/r from the isotropic walk, the even split from the matching being gauge, and conformal flatness from depth entering the three axes alike, so that it can only rescale every length together. The contribution is that these rest on four axioms, a machine whose rules have no tunable parameter, and two postulates, with each step checkable on the machine. Version 4 adds a Conclusion that states this smaller base as the aim, and the standing: the paper establishes mechanism, not scale; quantities in physical units need the coarse-graining carried to physical scales, and the predictions where the construction can be found wrong belong to a broader programme. On the machine, the outcomes that could have failed were D_b ≠ D_z, K_src ≠ 1, a second source changing the first one's field, and a ψ sign around a charge with whole turns.

1.4 Version 3 wrote two projections as one. Projecting the lift onto the base along the fiber leaves the fiber height (x₁ + x₂ + x₃)/√3 unstored, one coordinate shared by the three axes and by every observer: the holographic depth. An observer projects the same points centrally onto its sky, and there the coordinate not stored is r, the Euclidean distance in the lift. The dimension count has one form: (1,1) from the time–depth plane, with the plus sign spread over the three axes; the sky directions are directions of depth, not further axes.

Category 2

2.1 Correct: the jackknife runs over records, so a single record has no error. Since version 3 the two-source test and the ψ walk sum have been run with 16 seeds each and the island at ρ = 0.075 with 24. The conclusions stand with smaller errors: K_src 0.96–0.99 ± 0.02 with two sources, their histories overlapping by 0.4–0.6%; coherence 0.88–0.93 in front of a closure and 0.02–0.09 behind it, the zero line centred within 0.2. Version 4 brings every entry of Table 1 to at least four records, quotes figures to the precision their errors allow, and gives the entries behind each fitted row.

2.2 With the tail omitted, K at r ≥ 20 is 0.88, a lower bound. The tail exponent fitted to the last lags is 1.46–1.52, so the t^(−3/2) completion is measured rather than assumed; with the fitted exponent K = 1.00 ± 0.01–0.03, and exponents 1.3 and 1.7 give 1.06 and 0.96. The island at ρ = 0.075 stays at A = 0.90 with 24 seeds, a systematic deficit that version 4 keeps open.

2.3 The slopes carry seed-bootstrap errors in our readout (for instance dg/d ln b = −0.036 ± 0.015 at total density 0.2), which version 4 quotes together with the values of g. A one-parameter scaling of the transfer ratio collapses all 48 substrates onto one decreasing function, one branch, the localizing one, with ξ_∞ spanning about five decades and exceeding the measured range on the sparsest substrates. The candidate line is a crossover, and version 4 calls it so.

2.4 The cell-machine analysis has been rerun under the published rules: fluctuation ratios 0.94–1.06, and the 2b₀ calibration as under the earlier rule. The tie phase needs no shared time: a static phase per router reproduces the long-wave D within 1%, while starting every router from the same axis singles out a direction and lowers it by 7%. Claims extended beyond the sizes checked will be graded [M].

Category 3

3.1 The torus is a patch of the sphere for local results; the global structure is checked on the Goldberg-sphere runs, and version 4 marks which is which. The Seifert structure and the scaffold are two readings of one set of data: the point holonomies of the routing are the curvature of a connection on the fixed bundle of the scaffold, and read as geometry instead they make the total space Seifert. The transform and the degree argument of App. C use only the first.

3.2 The missing map is the observer's central projection: a walk ending at o from y is placed at the direction of y seen from o, and dμ_o is the time-integrated walk density along each direction, the pushforward of the walks to the sky. Then R_ρ(o) = ∫ σ(y) G(o, y) dy is the transform with the radial sum done, the ψ channel sums e^{iθ/2} over the walks of each direction, and Eq. (11) is the limit of walks concentrated on single lines. f_ψ enters Eq. (11) as the parity of the odd channel, not as an integrand; the strand difference itself does not carry ψ.

3.3 No probability is derived, and version 4 drops the vocabulary that suggests one: coherence is defined as the modulus ratio of the phase-weighted to the unweighted walk sum, the fringe sentence becomes a sign change of the ψ reading around a unit charge, and ħ only sets units. The half-integer weight follows from what a bit's frame can distinguish, not from a minimum. A ribbon has two sides, and a bit's side is one minimal distinction; the half twist of every edge exchanges them, and since the ψ channel is odd under the exchange, a full turn acts as −1: carried in SU(2), a strand's frame returns with (−1)^(N_e/2) after N_e edges (measured on 36 routes). The lift treats its three axes alike, so the frame transforms under its rotations, not only those about the fiber; a two-state representation on which a full turn is −1 is then a spin-½ doublet, with weight ±½ about the fiber. A weight of 3/2 would need four states. That a single reversed vertex, a third of a turn, leaves no sign then follows rather than selects. Where interference and probability sit among the readings is a next step.

3.4 "The number of particles is fixed" overstates the paper: the frozen pattern fixes the register and every topological charge, while the number of particles depends on the resolution at which the pattern is read. Charge conservation is then exact, and a neutral configuration can be read at finer resolution as a +1 and a −1. Fusion is indeed a change of resolution. The fractional charge is claimed only to have no Gauss-law reading. Locating the weak interaction in the ρψ channel, the strong interaction in closure, and antimatter are the next steps.

Minor issues

1. Eq. (15) will be the plain expansion c(1 − 2U), with γ = 1 read from it.
2. Agreed: the offset of §4.4 is transported, not a Shapiro delay; no machine measurement corresponds to Eq. (16).
3. Measured since version 3: when the flow carries a net current relative to the structure, the delay field is advected as advection–diffusion predicts (at r ≥ 20, 0.88, 0.50 and 0.29 of the field without current downstream, sideways and upstream), and the current relaxes against reversed vertices within (2–4)/p ticks. The medium's rest frame is thus the structure's, selected by the state, not the law. The notation for ħ will be unified.
4. Wording will follow the uncertainties: 0.73 ± 0.08 is consistent with 0.746, and D_P ranges over 0.07–0.10.
5. Daggers will be resolved at each use; the Zenodo archive is the canonical code; the driver will check tolerances and report pass or fail.

Literature

γ = β = 1 and α₁ = α₂ = 0 are fixed, the last once the current has relaxed; the other post-Newtonian parameters involve moving or extended sources and several sources beyond first order, on which the paper claims nothing. Will (2014), Collins et al. (2004), Ziff, Kong and Cohen (1991), Bunimovich and Troubetzkoy (1994) and d'Humières, Lallemand and Frisch (1986) will be cited at the corresponding claims. For the ρ channel the source is recovered from the full reading by its Laplacian, so the remark of §6.4 concerns the sky at one point; injectivity of the transform will be listed as open. Parallel programmes (Gorard, Jacobson, Sakharov) are not added; Bilson-Thompson and 't Hooft remain acknowledged.