OPT Red Team — Standing Objection Log

Anders Jarevåg

v1.0 — April 30, 2026

OPT Red Team

Purpose. Honest-broker hygiene for the Ordered Patch Theory. This document collects the strongest objections to the framework — present, anticipated, and previously raised — with honest assessments rather than defences. It is published as a companion to the formal papers but is deliberately not optimised for citation or rhetoric: its job is to make weaknesses visible, not to win arguments.

Use rule. Update this file by adding objections; do not trim. The Sokal-test for whether an objection has been taken seriously: someone hostile to the framework should read the entry and say “yes, that’s actually my point.” If they would say “you’ve softened it,” the entry needs rewriting.

Cross-references. Falsification commitments are in opt-theory.md §6.8 (F1–F5). Theories OPT is genuinely incompatible with are in opt-theory.md §7.12. This file goes deeper than either: it documents the specific arguments and the honest assessment of how OPT currently fares against them.

R1. The universality objection (string-theory trap)

Claim. Solomonoff \xi is so permissive that any computable structure can be embedded as a posterior. “OPT is structurally compatible with X” is therefore close to vacuous: the framework cannot rule things out, only accommodate them. Every successful “structural correspondence” in §7 / §IV is evidence of permissiveness, not of insight. The pattern matches string theory: an internally rich mathematical structure that accommodates everything and predicts nothing.

Honest assessment. This is the deepest objection currently on the table, and OPT’s defences are structural rather than empirical. The published response (§7.12 entry 6) names the worry but does not dispel it. The only thing that converts OPT from a worldview into a research programme is the §6.8 pre-registration commitments — and those have not yet been tested. Until F1–F5 deliver at least one quantitative confirmation of a number that was specified before measurement, the universality objection stands undefeated.

What would change the assessment. A successful pre-registered prediction with a number tighter than the prior range from competing theories. Until then, the structural-correspondence chapters are decoration.

R2. Selection bias in the comparative analysis

Claim. §7 (opt-theory) and §IV (opt-philosophy) cite frameworks that fit and do not seriously engage frameworks that don’t. Husserl, Merleau-Ponty, Gell-Mann, Van Raamsdonk, Wheeler are all in. Strict eliminativists (Frankish), strong reductive physicalists, anti-functionalists, and the cognitive scientists who think the bottleneck is incidental are largely absent or compressed into one-paragraph rebuttals. Every framework added to §7 strengthens the appearance of convergence; the asymmetry is itself evidence of the bias.

Honest assessment. Substantially correct. §7.12 was added to partially address this, but it is still one subsection against eleven convergence subsections. The §IV table in opt-philosophy similarly tilts toward agreement. The conversational pattern that produced §7.5 through §7.11 in 2026-04 was: identify gaps → fill them → repeat — without an equal number of “and here’s why this nearby theory disagrees” entries.

What would change the assessment. Expansion of §7.12 to cover at least as many positions in the same depth as the convergence subsections. A second pass through §7.1–§7.11 noting where the converging theory specifically disagrees with OPT, not just where it overlaps.

R3. \Delta_{\text{self}} as a get-out-of-jail card

Claim. The Phenomenal Residual is structurally unfalsifiable by design, which is appropriate as a Hard Problem locator but creates a methodological hazard: any disconfirming evidence can be absorbed into “that lives in \Delta_{\text{self}}.” The framework’s strongest formal claim is its weakest empirical one — it is precisely the part that is shielded from test.

Honest assessment. §6.8 attempts to wall this off explicitly: \Delta_{\text{self}} is excluded from the falsifiable core, and “absorbing a falsification of F1–F5 into \Delta_{\text{self}}” is named as disqualifying post-hoc reframing. Whether this wall holds in practice is an open question — it depends on disciplined application by future authors and reviewers, not on the formal apparatus itself. The hazard is mitigated, not eliminated.

What would change the assessment. A documented case where the framework accepts a falsification cleanly without invoking \Delta_{\text{self}} or substrate priority as a retreat. Until such a case exists, the wall is provisional.

R4. Anthropocentric reverse-engineering of C_{\max}

Claim. The numerical value C_{\max} \approx \mathcal{O}(10) bits/s is obtained by working backward from human introspective bandwidth (Nørretranders’ “user illusion” estimate, attentional blink data, Norwich-Wong saturation) rather than forward from first principles. The “informational necessity” of this specific number is suspect: any rate-distortion-grounded theory could have specified a different bandwidth and made it work. The number is a fit, not a prediction.

Honest assessment. Mostly correct. Appendix T-1 derives a range, but the range is wide enough to accommodate the empirically observed value rather than predict it. F1 commits to \mathcal{O}(10) with a 2-orders-of-magnitude window, which is generous. A genuine pre-registration would have been a specific number tighter than human-data range, derived without using the human data.

What would change the assessment. A re-derivation of C_{\max} from substrate-level constraints (Solomonoff weighting + Landauer + Markov-blanket dimensionality) that yields a specific number, ideally one that disagrees with the human introspective estimate by a small but specific factor — and then verifying that small disagreement empirically.

R5. The Stability Filter is a tautology

Claim. “Observers exist iff their predictive rate fits within their bandwidth” is a definition, not a discovery. Any apparent observer that exists trivially satisfies the Filter; any putative observer that doesn’t exist trivially fails. The Filter cannot rule anything in or out — it is a circular re-description of which configurations are observer-shaped.

Honest assessment. Partially correct. The Stability Filter as stated in §3 has a definitional flavour — it characterises observer-compatibility rather than predicting it from independent grounds. The framework’s defence is that the Filter generates distinct predictions (§6.1–§6.7) that would not follow from a tautological reading: the bandwidth hierarchy, the High-Phi Null State, the temporal-dilation expectation. If the Filter were genuinely tautological, these would not have empirical content.

What would change the assessment. Demonstration that the predictions in §6 actually depend on the Filter rather than being independently motivated. Currently this is asserted; it has not been formally verified that, e.g., the High-Phi Null State follows uniquely from the Stability Filter and not from a weaker independent commitment.

R6. The structural correspondences in §IV / §7 are post-hoc

Claim. When OPT is mapped to Hume, Husserl, Frankfurt, Merleau-Ponty, Metzinger, etc., the mapping is constructed after knowing what each tradition concluded. This is reverse-engineering, not prediction. A framework that could not have been built without those traditions in view cannot claim to derive their results — it can only claim to recover them in different vocabulary.

Honest assessment. Correct in the strict sense. OPT was built knowing the targets, and the §IV / §7 chapters are explanatory, not predictive. The framework’s defence — that it offers a deeper structural reason the converging traditions saw what they saw — is plausible but not provable, because there is no controlled experiment in which OPT predicts a tradition’s conclusions before the tradition reaches them. The convergence is post-hoc by construction.

What would change the assessment. A novel philosophical or empirical claim derived purely from OPT’s information-theoretic apparatus that no existing tradition has reached, and which subsequent work in those traditions independently reaches. This would be retrospective evidence of explanatory power.

R7. The bandwidth bottleneck as evolutionary contingency

Claim. GWT, the standard cognitive-science position, treats the conscious-access bottleneck as an evolved feature of primate cortex, not a structural informational necessity. There is no compelling argument that the bottleneck had to take the form it does; a sufficiently different evolutionary path could have produced phenomenality in a parallel architecture. OPT’s “informational necessity” is rebranding a contingent fact as a structural one.

Honest assessment. This is the strongest version of R1 made specific. OPT’s response (§7.10) is that the bottleneck is required because incompressible parallel streams violate the bandwidth condition — but this presupposes the Stability Filter, which is what’s at stake (R5). The dialectic loops. The honest position is that the necessity claim is currently postulated, not proven; F1 in §6.8 is the empirical commitment that, if confirmed, would supply the missing argument.

What would change the assessment. Either (a) the F1 measurement coming in tightly clustered around \mathcal{O}(10) across vastly different cognitive architectures (humans, cetaceans, corvids, eventually AIs), suggesting structural rather than contingent origin; or (b) a clean theoretical demonstration that no Stability-Filter-compatible system can lack the bottleneck.

R8. The “AI consciousness” extension is unfalsifiable in practice

Claim. §7.8 / §6.7 commits to the claim that LLMs and AIXI-bounded approximations are not conscious because they lack the C_max bottleneck. F3 (temporal dilation under bottleneck) is testable in principle, but in practice, no one is going to build a deliberately-bottlenecked synthetic agent at 10^4 \times clock speed and ask it about subjective duration. The prediction looks committed but is operationally inert.

Honest assessment. Largely correct as of 2026-04. F3 needs a concrete experimental protocol and at least one funded-or-pledged attempt to run it. Without that, the AI predictions in §7.8 are “would be testable if anyone tried” — which is a weaker commitment than F2 (the High-\Phi Null, where IIT vs. OPT discriminators are actually being constructed).

What would change the assessment. A specific institutional commitment to run F3 (e.g., a research group, a dated milestone, an experimental protocol agreed before construction). Without it, F3 is provisional pre-registration only.

R9. The substrate-priority claim is unfalsifiable internally

Claim. §3.12 argues that the substrate is “more fundamental” than the render via a thermodynamic-irreversibility argument, but every test of this priority would have to be conducted within the render. The claim is internal-incoherent: if substrate priority makes no operational difference inside the render, it has no content; if it does make an operational difference, that difference is itself part of the render and not evidence about the substrate.

Honest assessment. Acknowledged in §3.12 and §6.8 (excluded from F1–F5). The defence is that substrate priority is offered as an ontological commitment, not a falsifiable empirical claim. Whether ontological commitments not subject to empirical test should be allowed in a scientific framework is a separate methodological question. Strict empiricists (R5 / §7.12 entry 5) will reject this category; OPT keeps it but flags it.

What would change the assessment. This is a stable disagreement, not an empirical question. The honest move is to keep substrate priority quarantined from F1–F5 and accept that strict empiricists will not be moved.

R10. The “structural shutdown criteria” themselves are structurally easy to game

Claim. F1’s 2-order-of-magnitude window, F2’s “agreed protocol before construction,” F3’s “across k \in [10, 10^4]” all have enough wiggle room that motivated reasoning can find ways to claim near-falsifications are not falsifications. The shutdown criteria look strict but in practice can be gamed by tightening definitions, disputing measurements, or invoking experimental confounds.

Honest assessment. This is the meta-objection: pre-registration is only as binding as the discipline of the people interpreting it. OPT cannot self-enforce its falsification commitments. The mitigation in §6.8 is the requirement that any weakening be flagged as re-registration in the Version History, which voids prior tests — but a future author could simply do this and accept the cost. Trust in the shutdown criteria is contingent on third-party scrutiny, not on the formal commitment alone.

What would change the assessment. External adversarial peer review committed to checking the F1–F5 wording for game-able vagueness and tightening it. A pre-registration with a third party (OSF, AsPredicted) rather than only in Version History.

R11. The CMB carries quantum signatures the codec didn’t have to invent

Claim. The Cosmic Microwave Background shows specific quantum-mechanical signatures — near-scale-invariant power spectrum, near-Gaussian fluctuations, tensor-to-scalar bounds, statistical features matching inflationary quantum-field-theory predictions to Planck-satellite precision. These are conventionally interpreted as imprints of quantum vacuum fluctuations operating at cosmological scales ~13.8 Gyr before any observer existed. If QM is “a resolution artefact” of a bandwidth-bounded observer’s codec (opt-theory.md §7.1 items 1–2), why does the deep cosmological past — observed in aggregate, without fine-grained measurement — carry quantum signatures rather than classical thermal-noise signatures? This is a concrete cosmological instance of R1 and a sharp pressure point on the QM-as-codec-feature reading.

Honest assessment. OPT can absorb the CMB observations only by committing to the strong reading rather than the loose one. The loose reading — “QM is observer-side bookkeeping during measurement” — collides with cosmological data. The strong reading — codec compression is Hilbert-structured globally, applied uniformly forward and backward in rendered time, with the most-compressible past selected by Solomonoff parsimony — is internally consistent: an inflationary-quantum past is the minimum-description-length explanation for the observed CMB pattern, and the codec is therefore forced by parsimony to render it. This response is supported by §8.5 (atemporal substrate), §7.1 item 4 (delayed-choice generalised), and the QECC chain in Appendix P-2. The price is committing OPT to a stronger and more falsifiable claim than the loose reading: the codec’s Hilbert structure operates on the entire rendered timeline, and any bandwidth-bounded observer with a coherent cosmological past will see quantum signatures in it. The §7.1 commitment paragraph (added v3.4.0) makes this position public.

What would change the assessment. Cosmological-history features whose minimum description length exceeds what an inflationary-quantum default produces — i.e., features the codec would not invent by parsimony pressure but which exist in the data anyway. Concrete candidates: persistent non-Gaussianities of high algorithmic complexity that resist any short-description inflationary model; CMB anisotropies that survive scrutiny without any compressible inflationary explanation; primordial gravitational-wave signatures of specific quantum-event provenance that an inferential Hilbert-codec running backward in time cannot reproduce. Any such observation, third-party confirmed and resistant to alternative compressed explanations, would constitute description-length excess against the most-compressible-past machinery and falsify the strong reading. Operationally this would qualify as the kind of “independent demonstration” named in the §6.8 Project Shutdown criteria, even though it is not directly one of F1–F5.

R12. The strong-reading commitment looks like motivated post-hoc immunisation

Claim. The §7.1 codec-geometry commitment paragraph (added v3.4.0, April 30, 2026) was added in direct response to the CMB-QM challenge raised in the same session. It strengthens OPT’s reading of QM from a loose “observer-side bookkeeping at measurement time” to a strong “Hilbert structure across the full rendered timeline,” conveniently making the CMB-QM observations a prediction rather than a falsifier. The named falsifier — “description-length excess in cosmological history” — is technically defined but practically hard to demonstrate. Structurally, this is what research programmes do when challenged: tighten the framing to absorb the challenge, declare it was always implicit, and name a falsifier abstract enough that no near-term observation could satisfy it. R1 accuses OPT of accommodating everything; R12 accuses OPT of learning to accommodate in real time. R11 then becomes evidence for R12 rather than independent corroboration of OPT.

Honest assessment. Partially correct and partially defensible.

Correct in form. The commitment was added in response to a specific challenge. Even though §8.5 (atemporal substrate), §7.1 item 4 (delayed-choice generalised), and Appendix P-2 (QECC chain) supported the strong reading already, public commitment to that reading as the OPT-canonical interpretation was new in v3.4.0. From outside, this looks like goalpost-moving; from inside, like clarification. No external test distinguishes the two.

Partially defensible. The strong reading is a cost, not a free lunch — it closes off the loose-reading retreat that would otherwise be available against future challenges of the same form. v3.4.0 OPT is more falsifiable than v3.3.0 OPT, not less. The named falsifier (description-length excess / minimum description length) has definable mathematical content even if operationalising it is hard, so it is not “whatever we decide doesn’t count.”

The honest position. The v3.4.0 commitment does not count as evidence for OPT. It is a refinement that shifts the framework’s burden. The CMB observations that prompted the commitment cannot be cited as confirmation, because they were the prompting evidence. Only independent future tests of the v3.4.0 prediction — observations or analyses produced after April 30, 2026 by parties who did not participate in the framing — bear on OPT’s empirical status under the strong reading.

What would change the assessment. Either (a) a cosmological observation made after April 30, 2026 that the v3.4.0 commitment specifically predicts and that competing frameworks predict less cleanly — evidence the commitment was a real pre-emptive constraint rather than a post-hoc absorption; or (b) external commentary identifying unstated implications of the strong reading that were not anticipated when the commitment was made — weakening the “always implicit” defence and strengthening the post-hoc reading; or (c) third-party tightening of the falsifier wording into a specific class of measurable observables, making “description-length excess” operationally distinct from abstract.

R13. The 10 bits/s figure underlying F1 is itself contested

Claim. F1 (§6.8) anchors on a “human subjective predictive bandwidth C_{\max} \approx \mathcal{O}(10) bits/s” derived from Zheng & Meister 2024 [23] and four decades of convergent psychophysics. But the 10 bits/s figure has been challenged in 2025 literature — for example “The brain works at more than 10 bits per second” (PMC12320479) — arguing that conscious access channels are wider than the canonical estimate when measurement methodology is changed. If the canonical figure is wrong by a small factor, OPT’s central empirical commitment is recalibrated; if it is wrong by orders of magnitude, F1’s window itself becomes the issue.

Honest assessment. F1’s window was deliberately set wide (2 orders of magnitude in either direction) precisely because the underlying empirical figure is contested and methodology-sensitive. The contested status of the 10 bits/s anchor does not by itself defeat F1 — values from \sim 10^{-1} to \sim 10^3 bits/s would all fall within the F1 window, and \sim 100 bits/s would not even count as a near-falsification. What it does mean is that F1 cannot be advertised as resting on a settled measurement. The structural requirement OPT actually depends on is the existence of a low-bandwidth serial bottleneck, not the precise number — and §7.8’s distinction between the structural criterion and the biological constant (added v3.4.0) makes this explicit. F1 remains a useful pre-registered commitment for human observers, but its empirical anchor is provisional, not settled.

What would change the assessment. Either (a) a third-party reproduction of conscious-access bandwidth that converges on a value with much smaller error bars than current literature, allowing F1 to be tightened into a sharper test; or (b) a credible methodological argument that the bottleneck construct itself does not survive measurement-scheme variation — which would be a deeper challenge than R13 and would route into R5 (Stability Filter as tautology). The intermediate state is the live one: keep F1 as written, with the caveat that its empirical anchor is unsettled.

R14. Cosmological-history observations are testable in principle but no near-term result is decisive

Claim. R11 names “description-length excess in cosmological-history features beyond the inflationary-quantum default” as a falsifier of the §7.1 codec-geometry commitment. As of 2026-04, current CMB constraints rule out strong non-Gaussianity but are not stringent enough to rule out the inflationary-quantum default; primordial gravitational-wave constraints continue to tighten without a detection. No 2026 observation has shifted the picture for or against OPT’s strong reading. The next round — Simons Observatory, LiteBIRD, CMB-S4 — is expected to tighten constraints by roughly an order of magnitude, but on timescales of years, not weeks.

Honest assessment. R11’s falsifier is genuinely operational in principle but is not currently active. This is the right state for a structural commitment of this kind: the framework has named what would defeat it, the experimental community is moving toward more stringent tests, and no positive result either way has arrived. The honest move is to leave R11 as-is and re-check this entry annually as new cosmological data arrives.

What would change the assessment. A formal Simons / LiteBIRD / CMB-S4 result that either (a) detects features whose minimum description length under the inflationary-quantum default exceeds that of competing compressed accounts — falsifying the strong reading and triggering §6.8 Project Shutdown consideration; or (b) tightens existing constraints sufficiently to convert R11’s falsifier from “in principle” into “currently surviving” with much smaller error bars — strengthening the strong reading without confirming it. Either move warrants an explicit annotated update of R11.

Operating notes