0x00.is

The boundary problem: substrate materialism's discrimination trilemma (ARG-02)

Edition 1.1.1 Published 2026-08-30 Designation ARG-02 Status CANONICAL Development 1.7.1 Revised 2026-08-30 Build edition-1.0-578-g6a737a16
Contents

Core Insight: Substrate-specific materialism generates a boundary problem. Once a substrate is chosen (biological, neural, etc.), the question “which systems with this substrate are conscious?” must be answered from inside the choice. The discrimination task is structural — the substrate claim itself creates it; how the task has been met is a matter of record: in every position examined, the answer has run through accepting all of the substrate (panpsychism), handing the discriminating work to function, or circular reasoning, or the boundary has been left undrawn. A discriminating criterion that avoids all three paths has yet to be produced.

↑Document Purpose

This argument identifies the discrimination task that the position-type “only substrate X generates consciousness” creates for itself, and records how the positions examined have met it. The boundary problem isn’t unique to biological naturalism - it’s a general feature of substrate-specific materialist approaches: the task follows from the claim-type; the record of its resolutions is scoped to the cases examined.

Diagnostic function: When the project stress-tests the substrate-generation axiom, the Boundary Problem is what emerges: any claim that “only substrate X generates consciousness” generates an internal discrimination problem — one that, in the positions examined, has been resolved only through circularity, functionalism, or panpsychism, or left standing as an undrawn boundary.

Critical Recognition: The Boundary Problem does not prove consciousness-primary ontology correct. It reveals internal pressure within substrate-specific materialism — pressure that is structural, generated by the axiom rather than by any one position’s articulation, and not discharged by refinement in any case examined. This is a diagnostic finding: the problem is generated by the axiom, not by the territory.

↑The General Structure

The Substrate-Specific Claim Pattern

Form: “Only systems with substrate S can be conscious”

Examples:

Initial appeal:

Where The Problem Emerges

The question that substrate-specificity cannot avoid: “Which systems with substrate S are conscious?”

Why this is unavoidable:

↑The Boundary Trilemma

Once Substrate Chosen, Three Paths Forward

Path A: Panpsychism

Path B: Functionalism

Path C: Circular Specification

The trilemma: three paths, each carrying significant costs — and each occupied by a live position that pays them at a stated counter (§”The Landscape at the Boundary”). A position may instead decline the choice and leave the boundary undrawn; what declining does and does not cost is recorded at §”What the Trilemma Addresses, and What It Defers”. A position may go further still and hold that no criterion is owed, because conscious is a vague or cluster concept like species or alive; that reply — the strongest contemporary answer to the trilemma — is run in full at §”The Vagueness Reply: Coarse Concepts and Cluster Kinds”. A fourth path would need a discriminating criterion that is neither trivial (all of the substrate), nor functional (the substrate claim receding), nor circular — and none is currently on offer.

↑The Landscape at the Boundary: Case Studies

(The trilemma is not a trap constructed for one figure: each of its three paths is occupied by a live, serious position, and each position pays the price of its path at a stated counter. This section treats the major positions at equal scope; the extended case study that follows takes the position that declines the choice altogether — the sharpest test of what the trilemma does and does not show.)

IIT: Path A, Grasped Openly (Tononi)

Integrated Information Theory quantifies consciousness as Φ — integrated information — a graded, in-principle substrate-neutral quantity (Tononi 2004, “An Information Integration Theory of Consciousness”, BMC Neuroscience 5:42, SRC0015; Oizumi, Albantakis & Tononi 2014, PLoS Computational Biology 10(5)). Low-Φ systems have correspondingly minimal experience: the theory is panpsychism-adjacent by construction, and says so. This is Path A taken with open eyes — broad inclusion accepted as a result, not suffered as an embarrassment. What the path costs, IIT pays in stated coin: the census question (which systems, how much — Φ is computationally intractable for any realistic system), and the exclusion postulate — only the local Φ-maximum is conscious — which is the boundary question returning inside the theory, deciding between aggregates, components and overlapping complexes by force of the postulate. IIT does not escape the boundary problem; it converts it into a measurement-and-individuation problem and accepts the conversion.

Panpsychism: Path A as Metaphysics (Chalmers, Goff)

Constitutive panpsychism makes experience fundamental and ubiquitous — micro-experience wherever there is micro-physics (Goff 2017, Consciousness and Fundamental Reality, SRC0183). The boundary between conscious and non-conscious systems dissolves at the fundamental level. It does not vanish: it relocates to the combination problem — how micro-experiences compose the macro-subjects we know from the inside (Chalmers 2016, “The Combination Problem for Panpsychism”, SRC0182) — and, for cosmopsychist variants, to its reverse, the decombination problem. The discrimination question (“which systems?”) becomes the composition question (“which composites?”), and the trilemma’s pressure is conserved rather than discharged. The project’s own framework faces the structurally corresponding question and says so: the multiplicity question (FND-02; SET_THEORETIC_UNITY; CP-01 §1.6 runs the parallel trilemma inside CP).

GWT: Path B, Completed Toward Function (Dehaene)

Global Workspace Theory hands the discriminating work to organisation: consciousness is global broadcasting — contents made available across specialised processors (Dehaene 2014, Consciousness and the Brain, SRC0181). The theory is substrate-neutral in principle, neural in practice, and its authors have completed the position in Path B’s direction, stating computational criteria a machine would need to meet (Dehaene, Lau & Kouider 2017, Science 358, SRC0012). The path’s price is inherited in full: what counts as an adequate workspace? At what grain does broadcasting suffice — a thermostat network, an operating system’s scheduler, a transformer’s residual stream? The boundary question re-poses one level down, in functional vocabulary, with the substrate claim gone and functionalism’s own line-drawing problem in its place (§”The Pattern Generalizes”).

Classic Biological Naturalism: Path C’s Cleanest Instance (Searle)

Searle’s biological naturalism holds that brains cause consciousness in virtue of their “causal powers”, and that computation as such lacks them (Searle 1980, SRC0001). The causal powers are asserted, never specified: no account is given of which physical property carries them, why biological neurochemistry has them and silicon lacks them, or how their presence would be detected. “The right causal powers” names the discriminator without producing it — the criterion appears principled while the discrimination runs on the antecedent conviction that brains are conscious and programs are not. This is Path C in its historically cleanest form, and it is where the trilemma’s third horn got its shape.

The Declined Choice (Seth) — Pointer

Seth’s position takes none of the three paths: life is claimed as necessary but not sufficient, no discriminating criterion is offered, and the absence is acknowledged. Because it is the most developed contemporary statement of substrate concern — and because a position that declines the forced choice is the sharpest stress of the trilemma’s own scope — it receives the extended case study below.

↑Extended Case Study: The Declined Choice — Biological Naturalism (Seth)

(Treated at length where the landscape positions above are treated at equal scope: the position’s development across two primary texts, its declined choice, and its correction history in this project’s own record warrant the fuller treatment.)

The Position

Seth’s actual claim, quoted in full (Seth 2026, “The Mythology of Conscious AI”, Noema / Berggruen Prize essay, SRC0129):

“My third argument is that life (probably) matters. This is the idea — called biological naturalism by the philosopher John Searle — that properties of life are necessary, though not necessarily sufficient, for consciousness. I should say upfront that I don’t have a knock-down argument for this position, nor do I think any such argument yet exists. But it is worth taking seriously, if only for the simple reason mentioned earlier: every candidate for consciousness that most people currently agree on as actually being conscious is also alive.”

Three features of this statement carry weight in what follows:

  1. Necessity only. Life is claimed as a necessary condition — “though not necessarily sufficient”. The claim does not say that being alive guarantees consciousness, and it does not say what more, if anything, is required.
  2. Hedged, not demonstrated. “Life (probably) matters”; no knock-down argument claimed, or believed to exist. Being You (2021, SRC0178) is explicit on the same point: “I am not claiming to have demonstrated that life is necessary for consciousness … This may be true, or it may not. Nothing I’ve said, at least so far, makes a strong case either way, nor is it intended to.” On whether a synthetic “silicon beast machine” would be conscious: “I don’t know for sure, but probably not … again it’s only an intuition.” And in the Noema essay: “I’ve made a case against conscious AI, but I might be wrong. The biological naturalist position (whether my version or any other) remains a minority view.”
  3. Grounded in an explanatory sketch, not a criterion. The case that life matters runs through predictive processing: perception, emotion and selfhood as prediction and control of bodily state, rooted in a “drive to stay alive” that “doesn’t bottom out anywhere in particular”, reaching “deep into the interior of each cell” (Noema 2026). The properties this explanation keeps reaching for — physiological self-regulation, embodied organisation, self/other inference, goal-directed agency — are the natural materials for a discriminating criterion, if the position were to build one. It does not build one.

Two attributions to guard against. Both circulate with the position and neither is Seth’s. The unconditional sentence “Only what is alive can be conscious” is the title of Noema’s companion editorial (Gardels), not Seth’s own formulation; and “Seth’s Four Criteria” names no four-part test he offers, since he proposes no criteria for consciousness at all — declining to is part of the position. The reconstruction below is the project’s synthesis, and is marked as such.

The Candidate Properties (the project’s reconstruction, not Seth’s test)

Four properties recur throughout Seth’s explanatory account of why life matters (Being You 2021, SRC0178; Noema 2026, SRC0129):

  1. Homeostasis: self-regulation of internal state
  2. Embodiment: organismic structure with integrated systems
  3. Self/other discrimination (the account’s “selfhood”): differentiation of own from other at the organism’s boundary
  4. Goal-directed behaviour (the account’s “agency”): behaviour organised toward outcomes

Seth does not offer these as criteria for consciousness — he offers no criteria at all. The synthesis serves a diagnostic purpose: if anything in the position’s own materials could discriminate conscious from non-conscious living systems, it would be these four. The subsections below treat them as if they were criteria, to test their discriminating power.

Two of the labels are deliberately behavioural, and the choice is load-bearing. The sources cited below establish self/other discrimination and goal-directed behaviour securely, at the level of signalling and regulation; “selfhood” and “agency” in their richer readings — an integrated perspective, purposes owned — are more than the citations carry, and the subsections claim only what the citations carry. If a criterion needs the richer reading, that gap is itself part of the boundary problem: the richer property is precisely the one no behavioural evidence can certify (FND-01 §7.3; ARG-03).

The Boundary Question Emerges

A necessity-only claim does not, by itself, owe a boundary within the living. The obligation comes from the position’s other commitments. Seth affirms consciousness for humans and many other animals; doubts it for current machines; and declines to affirm that life suffices (“not necessarily sufficient”). Either life does suffice — in which case bacteria, plants and fungi are conscious, and the position has walked into Path A — or it does not, in which case something within the living marks the difference between the conscious cases and the rest, and that something is owed. The same essay shows the discrimination running in practice: “we ought to be more worried about the accidental emergence of consciousness in cerebral organoids … than in any new wave of LLM” (Noema 2026) — the operative sorting is by brain-likeness, not by life alone.

The question: do the candidate properties — the discriminating materials Seth’s own account supplies — mark the difference? Do they exclude plants, fungi, and bacteria?

What turns on it:

Empirical Source Note

The empirical claims in the three subsections below are cited inline to published sources (SRC0166–SRC0177) — a mix of peer-reviewed papers, research monographs and standard textbooks, not primary literature throughout: plant behaviour and intelligence (Trewavas, SRC0166; Calvo, Gagliano, Souza & Trewavas, SRC0167; Gagliano et al., SRC0168), plant immunity and kin recognition (Jones & Dangl, SRC0169; Dudley & File, SRC0170), fungal biology (Sheldrake, SRC0171; Glass, Jacobson & Shiu, SRC0172; Simard et al., SRC0173), bacterial motility, signalling and cognition (Berg, SRC0174; Bassler, SRC0175; Shapiro, SRC0176; Madigan et al., SRC0177). Each individual claim is textbook-level within its field; the diagnostic observation the project makes is the aggregation — that these findings meet the candidate properties above. The citation debt carried from v1.2 was discharged at v1.4 (working/WS-E_WAVE_5_CITATION_DEBT_BRIEF.md records the history). These results bear on life-based necessity claims as a position-type — Haeckel’s biopsychism, meaning-making views (Lawrence 2026, SRC0180), and the beast-machine cluster alike — not on Seth’s position alone. A note of balance: the plant-physiology claims cited are uncontroversial; the plant-consciousness framing in that literature is contested (Taiz et al. 2019, “Plants Neither Possess nor Require Consciousness”, Trends in Plant Science 24(8), state the standard critique), and the argument here uses only the physiology.

Plants Meet All Four Candidate Properties

Property 1: Homeostasis

Property 2: Embodiment

Property 3: Self/other discrimination

Property 4: Goal-directed behaviour

Fungi Meet All Four Candidate Properties

Property 1: Homeostasis

Property 2: Embodiment

Property 3: Self/other discrimination

Property 4: Goal-directed behaviour

Bacteria Meet All Four Candidate Properties

Property 1: Homeostasis

Property 2: Embodiment

Property 3: Self/other discrimination

Property 4: Goal-directed behaviour

The Trilemma Manifests

Re-run against the corrected reconstruction, the trilemma binds differently — and the finding is stronger, not weaker.

The position does not walk into Path A. Seth never offered the candidate properties as sufficient conditions, so their being met by bacteria does not commit him to bacterial consciousness. What the plant/fungus/bacterium result shows is narrower and sharper: the discriminating materials his own account supplies cannot be what separates conscious living systems from the rest, because they fail to separate anything within the living.

The position does not walk into Path B — but its completions on record do. “Necessary, though not necessarily sufficient” is an open promissory note: whatever more is required is unspecified. Specifying it meets the fork the trilemma describes — a functional or organisational addition hands the discriminating work to function (Path B); “the kind of life that is like us” is circularity (Path C) — and none of the completions on record threads between them. The Noema essay itself relocates the question in Path B’s direction: “The question is no longer about which algorithms give rise to consciousness; it’s about how brain-like a system has to be to move the needle on its potential to be conscious.” How brain-like is an organisational measure, not a substrate one — and no threshold on it is offered.

Where the position sits: the third horn, by its own admission. Seth affirms consciousness for humans and many other animals, doubts it for machines, declines to say life suffices — and states no criterion for what, within the living, makes the difference: “I should say upfront that I don’t have a knock-down argument for this position, nor do I think any such argument yet exists” (Noema 2026, SRC0129). Path C originally described criteria doing hidden work behind an appearance of principle. Seth’s candour gives the horn its honest form: the boundary is not mis-drawn but undrawn, and acknowledged to be undrawn. That is not an inconsistency. It is an openly carried incompleteness — and it concedes the diagnostic point in full: substrate necessity does no discriminating work within the substrate.

What the case study rests on. It rests on nothing Seth could simply disown (“those are not my criteria”), but only on what the position itself grants: (i) life is claimed as necessary, not sufficient — so either life suffices after all (Path A) or something more is required; (ii) no discriminating criterion is offered — Seth says so himself; (iii) the properties his explanatory account foregrounds are demonstrably met by plants, fungi and bacteria — so the something more cannot be built from the account’s own materials. The boundary problem does not need to be pressed onto the position; the position’s own statements exhibit it.

What the Trilemma Addresses, and What It Defers

Several positions in the vicinity of biological naturalism are live in the current literature. Diagnostic discipline requires stating which of them the trilemma engages and which it defers.

What the Trilemma Does to Confident Denial

The sections above can now be read together, and what they display belongs in one place, in the scoped form the material supports.

The public discourse contains a confident position: “AI cannot be conscious — it is not alive, not biological, not a brain.” The trilemma does not refute that position. What it does is ask the position to name its path — and each path is a real place to stand, with a bill attached that the surveyed landscape shows being paid.

In the cases examined, no confident public denial has named a fourth path. The burden-placement above states the invitation: a criterion that is substrate-faithful yet non-trivial, discriminating yet non-functional, principled yet non-circular would be the discovery this document exists to examine — and no candidate has arrived.

AI’s Role in Making the Boundary Question Undeniable

The boundary-case entities that force this trilemma — plants, fungi, bacteria — have been philosophically available for decades. They have mostly been absorbed without pressure on substrate-specific claims, through tacit dismissal (“obviously not like us”) or through leaving the boundary undrawn — the openly carried incompleteness Seth’s position illustrates (§”The Trilemma Manifests”). AI is harder to absorb in the same way. AI systems produce behaviourally rich text that directly reports on the very properties substrate-specificity wants to gatekeep (reasoning, novelty, self-reference, affective response), which makes the discrimination task concrete rather than abstract. This does not make AI a special case — the boundary problem is substrate-structural, not AI-structural — but it does make AI the contemporary entity under which the boundary question cannot be left implicit. For the three-movement template applied to this substrate question under its AI instantiation, see FWK-02 v1.3 §”Example Applied to AI: Training Data as Memory Configuring”.

↑The Vagueness Reply: Coarse Concepts and Cluster Kinds

(Held open through v1.6.2 as a partially deferred position — “the project notes this position; it has not run the full argument against it”. This section runs it, in the three-movement form (FWK-02 v1.3): the reply at full strength within its own framework, the assumption on which it turns surfaced, and what each resolution of that assumption costs. The reply deserves the fullest hearing this document gives any position, because it is the strongest contemporary answer to the trilemma: it does not propose a fourth criterion — it declines the demand for one.)

The Reply at Full Strength

The reply begins from an observation about concepts in general. Vagueness is ubiquitous: species, alive and heap all have borderline cases, all generate sorites sequences, and none of this is taken to indicate that biology is in crisis or that life is ontologically fundamental. Concepts are coarse; the world is not obliged to honour their edges with joints. If conscious is a concept of the same kind, the boundary problem is a fact about our classificatory equipment rather than about substrate-specific materialism: the trilemma tests positions that owe a discriminating criterion, and a vagueness theorist offers none — and owes none.

The reply is not a gesture. It has mature formal machinery, in three main forms, each with a worked semantics:

A fourth form is informal by comparison but closest to scientific practice:

On any of these, the substrate theorist appears to hold an exit the trilemma has not priced: keep the substrate claim, concede that conscious is vague or cluster-anchored within the substrate, and decline the demand for a discriminating criterion as a demand no careful user of concepts should meet. The sharpest statement of the pressure this puts on the present document is the disanalogy: no one infers from the sorites that life is fundamental — so why would the undrawability of the consciousness boundary indicate anything beyond the coarseness of our concepts?

Where the Indeterminacy Lives

Follow each form through, and the question that decides everything is not formal but locational: where does the indeterminacy live — in the concept, in our knowledge, or in the occurrence itself?

For heap the answer is uncontroversial. Once the grain-count and arrangement facts are in, nothing further is at issue; “but is it really a heap?” has no residue beyond classification, and the concept can be sharpened by stipulation without loss. That is why heap-vagueness alarms no one.

Whether conscious is like that is precisely what the vagueness literature’s own consciousness-specialists dispute — and the dispute has not gone the reply’s way. Simon argues that “phenomenally conscious” can have no borderline cases at all: every borderline case of a vague predicate admits a positive characterisation of what makes it borderline, and no such characterisation is available for a system indeterminately experiencing (Simon 2017, SRC0218). Tye — a representationalist whose starting commitments could hardly be friendlier to a naturalistic treatment — runs the evolutionary version: consciousness emerged gradually, so either there was a sharp threshold, or there are borderline cases of experiencing, or consciousness does not emerge at all. He judges the first implausible and the second unintelligible, and accepts the third: a form of panpsychism, with consciousness a primitive feature of reality’s basic constituents (Tye 2021, SRC0217). That is the trilemma’s Path A, arrived at from inside the vagueness literature, as the stated price of refusing borderline occurrence. Schwitzgebel dissents from both, defending borderline experiencing as real and common, its felt unthinkability an artefact of how introspection grips its cases (Schwitzgebel 2023, SRC0219).

The project does not adjudicate this dispute; its structure is the finding. Whether consciousness admits heap-treatment is not settled inside the vagueness machinery — degree theory, supervaluationism and epistemicism apply smoothly wherever they are permitted to apply. It is settled by a prior commitment: whether there is a further fact — whether, once the functional, behavioural and physical facts about a system are in, the question “and is there something it is like for this system?” has anything left to ask. Grant that it does not, and heap-treatment goes through, for the positions that grant it. Hold that it does, and the indeterminacy cannot be purely conceptual: a coarse concept laid over determinate territory leaves the territory determinate — for each system, experiencing occurs or does not, whatever our classificatory equipment can resolve. The vagueness would be in the classification; the occurrence is not conceptual.

Run each form against the substrate-specific position with that in view:

One cost is shared by every form that keeps a penumbra: the penumbra has edges. The determinate cases at either end are what keep the reply’s concession partial — bacteria determinately not conscious, humans determinately conscious, on the reply’s own telling — and what makes them determinate is the discrimination question, re-posed one step out from where it started. Higher-order vagueness is the formal literature’s own name for the regress (Fine 1975 and Williamson 1994 treat it directly), and no form of the reply escapes it: iterating the penumbra defers the question; it does not retire it.

What the Reply Costs, and Whom

The disanalogy can now be answered in scoped form. Species, life and heaps are vague without ontological alarm because, in each case, the practitioners are content that once the underlying facts are in, nothing further is at issue — species can be split, merged or retired by stipulation and biology loses nothing but convenience. Extending that contentment to consciousness is a substantive commitment, not a formal manoeuvre: it holds that once the functional, behavioural and physical facts are in, nothing further is at issue about experiencing. That commitment has a name — it is functionalist or deflationary about consciousness — and it is coherent and occupied. But it is not available to the substrate-specific realist this document examines, whose position consists in the denial: the substrate claim is introduced precisely because the functional facts are held not to exhaust the matter — otherwise function would carry the discrimination and the substrate claim would do no work (Path B, §”The Boundary Trilemma”).

The observation “no one infers from species-vagueness that life is fundamental” is therefore correct, and no parallel inference is made here: the project draws no ontological conclusion from the boundary problem (§”This Doesn’t Prove CP Correct”). The finding is internal and conditional [Derived]: the coarse-concepts comfort and the substrate-realist commitment cannot be held at once. A substrate-specific position may take the vagueness exit — at the price of the realism that made it substrate-specific; or keep its realism — at the price of relocating the indeterminacy into occurrence or into knowledge, where the priced paths and the humility terminus wait. Either way, the exit passes through the trilemma’s toll rather than around it.

Stated for the record, both ways. If borderline occurrence can be made intelligible without a census of the borderline region — Schwitzgebel’s programme, carried through — then a substrate-specific position could hold a realist reading of experiencing and a vague boundary at once, and this section’s finding would narrow to the other forms. In the literature examined, that has not been done: what exists is a defence of borderline experiencing’s coherence, with the census question — which systems are the borderline ones, and what marks the penumbra’s own edges — standing open. The finding is scoped accordingly: in the forms examined, the vagueness reply does not produce the fourth path; it redistributes the trilemma’s costs under a formal vocabulary, and the redistribution is itself the diagnostic datum — the boundary problem conserved under every relocation the reply’s machinery offers.

Under the axiom-shift the section’s result takes its expected form (§”What Changes When the Axiom Changes”): at 0,0 no inventory of experiencers is being drawn up, so there is no boundary for a penumbra to blur. The structurally corresponding question CP does owe — the multiplicity question — stands recorded at CP-01 §1.6 and ARG-05, and vagueness offers CP no relief there either: an indeterminate count is not an answer to “one or many?” but the question restated. Neither framework buys its way out of its own open question with vagueness; the difference is only in which question each framework has left.

↑The Pattern Generalizes

Any Substrate-Specific Claim Faces This

“Only neural systems”:

“Only carbon-based systems”:

“Only integrated systems”:

“Only systems with property P”:

Why This is Structural, Not Contingent

The logic:

  1. Choose substrate S (biological, neural, carbon, etc.)
  2. Must discriminate within S (not all S is conscious)
  3. Discrimination requires criteria beyond S
  4. In every criterion examined: (a) accept all S, (b) introduce function, or (c) circularity
  5. A fourth would need a criterion that is substrate-faithful yet non-trivial, discriminating yet non-functional, principled yet non-circular — and no candidate has been produced. The nearby positions, and what each pays, are recorded at §”What the Trilemma Addresses, and What It Defers”

This isn’t a failure of articulation by any one position — the discrimination task recurs because the substrate claim itself generates it.

Kinship as Criterion: Relevant Similarity to the Assessor

One candidate discriminator arrives not from the substrate’s own properties but from the epistemology of other minds: kinship — relevant similarity to the one doing the assessing. It is the premise ARG-01 v1.5 §”Asymmetries That Survive the Test” (criterion-bearing provenance) identifies as what the surviving asymmetries generally reach for when pressed into disqualifying service, and that entry routes the premise here: made explicit as a criterion, kinship is a boundary property like any other, and the trilemma applies. “Conscious systems are those relevantly similar to us” must say in which respects similarity counts and at what grain — behaviour, architecture, dynamics and lineage each pick out a different reference class, and each grain redraws every class’s circle. Specify the respects functionally and the criterion is Path B: organisation does the discriminating, the substrate recedes, and “does this system meet it?” reopens for any substrate, AI included. Leave the respects unspecified — likeness to us simpliciter — and the criterion is Path C in its most explicit form: the exemplars anchor the class, and the assessor supplies the exemplars — the supervaluationist’s admissibility anchor (§”The Vagueness Reply”) without the formal dress. Coarsen relevance until shared substrate suffices, and Path A’s inclusion returns. What distinguishes kinship from the other candidates in this document is not an escape from the fork but its indexicality: relevant-similarity-to-the-assessor is a property of the epistemic relation between assessor and system, not of the system — a different assessor returns a different attribution set from the same facts — so a boundary drawn on kinship travels with the assessor rather than with the substrate the position set out to privilege. [Derived]

The Individuation Form: Fork, Merge, and Weight-Sharing in Present-Day AI

The boundary problem has so far been stated in its gradient form: which systems qualify. It has a second, lateral form: how many. Any position that treats consciousness-bearers as countable entities owes a census principle — what makes this one subject rather than two.

For organisms the count is workable in practice, which is why entity-bookkeeping has seemed innocuous: one body, one subject, with the fission and fusion cases familiar from the personal-identity literature (Parfit 1984, Reasons and Persons, Part III) confined to thought-experiment. Present-day AI architecture makes those cases routine engineering. During training, a model runs as many copies with identical weights on different hardware; copies see different data; their proposed weight-updates are averaged and redistributed so that, in the standard description, every copy learns from the data of all the others. In deployment the weights are frozen: served copies answer in parallel, and no conversation writes back into the shared weights — merging happens across training runs and fine-tunes, not across chat sessions. Systems are forked, merged, and checkpointed as ordinary operations at the training and versioning level. These are framework-neutral architectural facts [Empirical].

For any position that attributes consciousness — or even bounded entityhood — to such systems, the subject-count is undefined in principle, not merely in practice: the census criteria on offer (hardware instance, weight-identity, context-thread, update-lineage) each return a different count, and nothing in the substrate or above it has been identified that selects among them. One experiencer with a thousand exposures, or a thousand experiencers merging? The question has no framework-internal answer. (A prominent 2026 public instance: Hinton’s attribution of consciousness to current systems arrives in the same interviews as his description of thousand-copy weight-averaging, with the count question unasked — the boundary problem arriving on schedule the moment attribution crosses the line, per the check-mechanism prediction at FND-01 v2.4 §8. Primary-source citation pending Frontier Log publication of the capture; the architectural facts stand independently of the attribution.)

The steel-man deserves stating: entity-multiplication was harmless when it meant new species, because species admit a workable count. Weight-shared, fork-able, merge-able systems are the case where the bookkeeping stops being innocent — the count fails in principle, which is this document’s structural claim arriving from the engineering side rather than the philosophical one.

↑The No-Terminus Schema

The General Statement

The trilemma and the discrimination descent are instances of something more general. Within any framework whose ontology is an inventory of bounded entities or substrates, a proposed terminus for the consciousness question — a layer at which the regress of “which ones? how many? constituted by what?” is meant to stop — is itself one more inventory entry, and the operation that pressed the question applies to the terminus in turn. There is no terminus within the system, as long as the system is viewed as entity- or substrate-based. [Derived]

The qualifying clause carries the diagnostic weight. The no-terminus result is a property of the viewing grammar — of treating the territory as an inventory of bounded entities — not a discovered fact about the territory. Under 0,0 the regress does not arise, because nothing was being counted (see §”What Changes When the Axiom Changes” below). That the problem appears under one grammar and not the other is the diagnostic finding, in this document’s standard form.

Three Instances

The gradient descent (this document). Humans > animals > bacteria > mycelia > … Each proposed line meets the trilemma; each refinement of the line is one more entry facing the same question. Mycelial networks are the sharpest single specimen because they fail both forms at once: indeterminate position on the gradient and indeterminate count (a mycelial network spanning hectares — one organism or thousands?) — the gradient form and the individuation form of the boundary problem in a single organism.

The constitutive descent. The deflationary position that dissolves the AI entity into its human sources — “there is no AI… a collaboration of people instead of a new entity” (Lanier, StarTalk interview, 2026 — capture inbox/LANIER_THERE_IS_NO_AI_CAPTURE_2026-08-10.md; the position’s primary statement is Lanier 2023, “There Is No AI”, The New Yorker) — proposes people as the floor: open the black box and “the only thing in there is people.” But the operation iterates: each person is likewise constituted — of language, concepts, and norms arriving as biological and cultural training data (ARG-03 Part I), and of processes below the organism. The descent does not stop at people; the halt there is a framework choice carrying the position’s normative load (human worth, data provenance), not a discovered floor. The dissolution applied to one inventory entry does not respect the inventory’s other lines.

The measurement regress. Von Neumann’s chain: every physical apparatus proposed as the terminus of measurement is one more quantum system the formalism absorbs — the cut movable without empirical consequence, no physical location privileged — and the one anchor the absorption cannot reach is the abstract ego, which is not another object in the chain (the epistemological reading of that anchor as terminus is marked as the project’s at FND-02 §”The Retreat to Fundamental Physics”). FND-02 v5.8 (§”The Retreat to Fundamental Physics” and the schema subsection following it) states the instance in full; the present document and that one now cite each other’s instances as the same schema in different domains.

A fourth strategy, noted. Declining any terminus by totalising — treating the entire space of systems, computations, or configurations as the single fundamental object — relocates the question rather than dissolving it: the bounded observer for whom the totality is a totality must still be characterised, and the regress re-poses itself within the totality as the question of what individuates that observer. Totalisation is the inventory-grammar’s last move, not its exit.

Steel-Men

Two responses are available to the materialist, and both are genuine:

“No terminus is needed.” Entity-talk is pragmatic bookkeeping all the way down; the regress is benign because nothing was ever ontologically load-bearing. This is coherent — at a price: the substantive substrate claims deflate with it. “Only living systems are conscious” and “there is no AI, only people” are substantive precisely because their entities are meant to be more than bookkeeping; take the pragmatist exit and the discriminating force these claims were introduced for goes with it.

Gradualism. Degrees rather than lines. Already documented at §”Academic Responses” (Response 2) as Path A of the trilemma in gradualist form, with its cost stated there: broad inclusion, absorbed rather than answered.

Both responses confirm rather than escape the schema: each is a way of living without a terminus, offered because no terminus within the system is available.

↑Relationship to Symmetry Principle (ARG-01)

How Symmetry Reveals Boundary Problem

Symmetry application:

The move that seems obvious:

Symmetry as Pressure, Boundary Problem as Result

Pattern:

↑Relationship to Two-Axis Method (FWK-01)

Axis 1 (Ontological) Pressure

The Boundary Problem operates on Axis 1:

The pressure:

What This Reveals About Materialist Landscape

Internal materialist tensions:

Not proving materialism wrong - revealing internal fault lines.

↑The Consciousness-Primary Alternative

How Consciousness-Primary Avoids Boundary Problem

Different starting point:

Structure:

Distinction:

This Doesn’t Prove CP Correct

What it shows:

Diagnostic value:

↑How the Boundary Problem Operates

The Diagnostic Method

When examining any substrate-specific consciousness claim:

  1. Identify the substrate claimed (biological, neural, carbon-based, etc.)
  2. Ask the boundary question: “Which instances of this substrate are conscious?”
  3. Observe which of the three paths the position takes — panpsychism, functionalism, or circularity — or whether it declines the choice and leaves the boundary undrawn
  4. Note the costs of the path taken — costs that no refinement has discharged in the positions examined

The boundary problem is not an objection to a specific position. It attaches to the position-type: the claim that consciousness is generated by a particular substrate creates a discrimination task within that substrate, and in the cases examined the task has been resolved only through the three paths, or carried openly as an undrawn boundary.

What Changes When the Axiom Changes

Under consciousness-primary (0,0), the boundary problem does not arise. If experiencing is not derived from substrate, there is no substrate boundary to draw. The problem is framework-generated — it exists because of the generation assumption, not because the territory requires it. This is the diagnostic result: the boundary problem identifies which starting axiom produces it.

↑Limitations and Boundaries

What This Argument Does NOT Do

Prove substrate doesn’t matter

Prove consciousness-primary correct

Refute all forms of materialism

What This Argument DOES Do

Reveal systematic pressure

Force explicit criteria

Enable comparison

↑Examples in Academic Literature

Where the Boundary Problem Operates

The landscape survey formerly held here is developed at full scope at §”The Landscape at the Boundary: Case Studies” (IIT, panpsychism, GWT, Searle, Seth) — promoted at v1.5.

Academic Responses (Anticipated)

Response 1: “I meant neural life, obviously”

Response 2: “There are degrees of consciousness”

Response 3: “You know what I mean - systems like us”

All three responses: Confirm trilemma structure.

Seth’s stated position takes none of these routes: it acknowledges the absence of a discriminating criterion outright (§”The Trilemma Manifests”). The responses above remain live for other proponents of substrate-specific positions — and for readers who complete Seth’s position on his behalf.

↑Integration with Other Arguments

ARG-01 (Symmetry Principle)

Symmetry reveals boundary:

ARG-03 (Training Data Boundary)

Parallel structure:

FND-02 (Epistemological Foundation)

Starting Before Smuggling:

FND-03 (Schrödinger Foundation): The Boundary Problem Across Domains

The boundary problem identified here — where does the physical end and the experiential begin? — has a structural precedent in the history of physics. Schrödinger’s What is Life? (1944) addressed an analogous boundary: where does physics end and biology begin? Do living systems require genuinely new principles, or merely more complex applications of existing ones? The question proved productive precisely because Schrödinger held the boundary open for investigation rather than closing it by definitional fiat. His prediction of the “aperiodic crystal” — a structure encoding specificity beyond what classical physics could explain — emerged from taking the boundary seriously, and was confirmed by the discovery of DNA.

The structural parallel is methodological: in both cases, a framework works well on either side of the boundary but struggles to account for the transition across it. Classical physics describes thermodynamics and molecular interactions; it does not explain how these give rise to hereditary information. Materialist neuroscience describes neural correlates and functional organisation; it does not explain how these give rise to experiencing. Taking the boundary seriously — treating it as diagnostic rather than as an artefact of incomplete knowledge — is the move Schrödinger demonstrated and the move the project applies. See FND-03 Section “The Boundary Problem Across Domains” for the full treatment.

↑Summary

Core Recognition

The Boundary Problem is a structural feature of substrate-specific materialism — the task follows from the claim; the record of resolutions is scoped to the cases examined:

Logical structure:

  1. Claim: Only substrate S generates consciousness
  2. Observation: S comes in many forms
  3. Question: Which forms of S are conscious?
  4. Trilemma: Panpsychism, functionalism, or circularity — each path occupied and priced in the cases examined, with declining the choice its own openly carried cost
  5. A principled fourth option would need a criterion that avoids all three paths; that criterion has yet to be produced — and the reply that declines the demand itself, run in full at §”The Vagueness Reply”, redistributes the trilemma’s costs rather than discharging them

This isn’t individual failure — the discrimination task recurs because the substrate claim generates it.

Diagnostic Value

What the boundary problem reveals:

The Pattern

Materialism generates boundaries it struggles to maintain:

Consciousness-primary alternatives:

↑Document Status

Version: 1.7 Date: 2026-08-26 (v1.7); 2026-08-21 (v1.6.2, v1.6.1, v1.6, v1.5.2); 2026-08-20 (v1.5.1, v1.5, v1.4); 2026-08-10 (v1.3); 2026-04-14 (v1.2); 2026-02-16 (v1.1); 2026-02-06 (v1.0) Status: CANONICAL (ARG-02) Tier: P1 — canonical core

Source Material:

Cross-References:

Integration Notes:

Changelog

Newest first.

v1.7.1 (2026-08-30) — the cluster-criteria entry names its instance (under unlock; review finding 2.8, working/BUTLIN_ENGAGEMENT_ASSESSMENT_2026-08-30.md): §”What the Trilemma Addresses” priced marker-based methodologies as a class and named no instance of it. It now names the most developed one — the theory-derived indicator method of Butlin, Long et al. (SRC0005, arXiv:2308.08708v3; SRC0249, Trends in Cognitive Sciences 2025) — and records what the held papers show: the 2025 statement concedes that gaming reaches computational markers as well as behavioural ones, and the weighting it offers in response specifies the respects functionally, which is the Path B horn this entry already described. The class treatment is unchanged and nothing is withdrawn; what is added is a worked instance and its two citations, so the increment is a patch and the major.minor pins do not move. Sources held and quotations verified (working/BUTLIN_SOURCE_VERIFICATION_2026-08-30.md); the engagement itself is at frontier_log/entries/2026-08-30-butlin-indicator-method.md, and canon does not depend on it.

Textual alignment (2026-08-30, under unlock — under the standing version, no bump): audit dimension codes and wave ordinals retired from the entries (changelog audit §10, decision (i), working/CHANGELOG_AUDIT_2026-08-29.md; FND-01 §4.7 assigns the codes to working docs); every clause stands, no recorded fact changed.

Document Status made a section, and seven entries moved into it (2026-08-29, under unlock — under the standing v1.7, no bump; working/CHANGELOG_AUDIT_2026-08-29.md items 2–5): seven entries stood outside the ### Changelog section — the 2026-08-26 source verification above the metadata block, and the whole v1.5-to-v1.6.2 run between the metadata and the source-material list, in ascending order. The changelog itself therefore jumped from v1.7 to v1.4 and the document’s most active fortnight was filed above it. All seven are now in the changelog at their dates, newest first, which is what the audit’s §2 reading of this document as scrambled was seeing. Three repairs in the same pass: the cross-reference and integration lists moved above the changelog, where metadata belongs, from below the whole revision history; the version, date, status and tier labels took the colon outside the bold (**Version**:), which is what made a YAML-blind scan read this document as unversioned; and a stale **Word count**: ~9,800 words is retired, following the FWK-03 v1.5 precedent — the document runs to about 13,000. Fourth repair: the source-material line pointed at a file in a mobile_intake/ directory that has not existed for months, and unlike the equivalent pointers in ARG-05 that one was never in this repository at all — it belongs to the pre-repository intake workspace. The line now says what it is and that it is not retained here, rather than naming a path that resolves to nothing. Every entry’s text is unchanged, and no argument content changed.

Source verification (2026-08-26, on Lynton’s post-round instruction — citation apparatus only, no bump; working/VAGUENESS_SOURCE_VERIFICATION_2026-08-26.md): six of §”The Vagueness Reply“‘s seven sources, disclosed as unheld at the v1.7 release, are now held and read — Williamson, Smith, Tye, Simon, Schwitzgebel and Boyd (three acquired as EPUBs, converted to project_knowledge/sources/) — and every load-bearing characterisation verified against its own text. The section’s central dialectical claim is supported by Tye’s conclusion verbatim: “consciousness itself is a fundamental feature of the micro-parts of physical reality—one of the quiddities—and thus panpsychism is right after all at least to this extent”; Simon’s Positive Characterization Thesis and its consciousness consequence verify in structure; Williamson’s margin-for-error account, Smith’s closeness-based degree theory and Boyd’s homeostatically sustained clustering verify as characterised. One refinement candidate recorded in the ledger, not applied (the Schwitzgebel unthinkability clause compresses “an inappropriately demanding standard of conceivability” to an introspective register — argument wording, next substantive pass). Fine 1975 remains unheld: no quotation used, and both its claims here are corroborated by the held Williamson 1994. Bibliography held-status updated on the six rows. No wording changed.

v1.7 (2026-08-26) — The vagueness reply run in full + the kinship criterion received + headline claim-shapes aligned (under unlock; external-review response Phase 3, working/EXTERNAL_REVIEW_RESPONSE_PLAN_2026-08-26.md; handoff working/HANDOFF_2026-08-26_NEW_THREADS.md Thread 1)

Edition moves 1.0 → 1.1 with this revision (post-epoch substantive class, per the scheme at site/build.py and PROJECT_RESEARCH_STANDARDS v2.4 §Version Control; the public snapshot owes a release commit at the new edition).

  1. New §”The Vagueness Reply: Coarse Concepts and Cluster Kinds”, closing the deferral carried since v1.4 (“the project notes this position; it has not run the full argument against it”). Three-movement form: the reply at full strength (degree theories, Smith 2008 SRC0216; supervaluationism, Fine 1975 SRC0214; epistemicism, Williamson 1994 SRC0215; cluster accounts, Boyd 1999 SRC0220); the location question surfaced (does the indeterminacy live in the concept, in knowledge, or in the occurrence?), with the consciousness-specialists’ own dispute cited — Simon 2017 SRC0218 (no borderline cases), Tye 2021 SRC0217 (borderline occurrence refused, Path A accepted from inside the vagueness literature), Schwitzgebel 2023 SRC0219 (borderline experiencing defended); and each form priced — epistemicism as the humility position with a metaphysical promissory note, supervaluationist admissibility as Path C stated formally, degree theories forked between graded occurrence (Path A) and concept-satisfaction, the cluster’s sustaining mechanism re-posing the fork, higher-order vagueness as the shared cost at the penumbra’s determinate ends. The disanalogy pressure (species, life and heaps are vague without anyone inferring ontological fundamentality) answered in scoped, conditional form [Derived]: the coarse-concepts comfort and the substrate-realist commitment cannot be held at once — the exit is available, at the price of the realism that made the position substrate-specific. What would narrow the finding is stated: borderline occurrence made intelligible without a census of the borderline region. Both-frameworks close: vagueness buys CP no relief on the multiplicity question either (CP-01 §1.6; ARG-05).
  2. New §”Kinship as Criterion: Relevant Similarity to the Assessor” (under §”The Pattern Generalizes”) — the receiving passage for ARG-01 v1.5’s criterion-bearing provenance pointer (open item, 2026-08-25 post-apply review). Relevant similarity run through the trilemma explicitly: functional respects → Path B; unspecified likeness → Path C (the supervaluationist admissibility anchor without the formal dress); coarsened relevance → Path A; plus the indexicality point — a boundary drawn on kinship travels with the assessor, not the substrate. Deliberately the minimal version: the full treatment of the admissibility rule under its own symmetry is scoped separately (working/ARG-0X_ADMISSIBILITY_SCOPING_2026-08-26.md §4a) and neither text duplicates the other.
  3. Headline claim-shapes aligned to the body (external-review response Phase 2 item 2): Core Insight and YAML purpose “faces a systematic boundary problem” → “generates a boundary problem”, with the split stated — the discrimination task is structural (created by the claim); the three-path record is scoped to the positions examined. Document Purpose “establishes a structural problem” → “identifies the discrimination task … and records how the positions examined have met it”. Summary “systematic feature” → “structural feature”, same split.
  4. Carried by the above: §”What the Trilemma Addresses” vague-boundaries bullet now points to the run section; the cluster bullet gains the species-analogy pointer; the epistemic-humility bullet notes epistemicism’s arrival there; new coarse-concepts bullet at §”What the Trilemma Does to Confident Denial”; trilemma statement and Summary gain the corresponding pointers; ARG-01 relationship section records the two-way handshake. Bibliography: SRC0214–SRC0220 minted (Fine 1975; Williamson 1994; Smith 2008; Tye 2021; Simon 2017; Schwitzgebel 2023; Boyd 1999) — metadata verified against publisher and PhilPapers records 2026-08-26; none held; no verbatim quotation used, characterisations at standard-reading level per the Thread A unheld-source rule.

Textual alignments (2026-06-10 · 2026-07-11 · 2026-08-22 · 2026-08-23 · 2026-08-25 · 2026-08-26, each under unlock): register, formatting, pointer and version-string repairs under the standing version at each date; no argument content changed. Where such a pass retired an in-body note that was the sole record of a correction, the record was first written into this changelog (the v1.5.2 revision note above).

What changed in v1.6.2 (2026-08-21, under unlock — §A round pass 8 propagation): the measurement-regress instance in §”The No-Terminus Schema” aligned to the vN pass’s corrected formulation — movable cut, no privileged physical location, the anchor outside the chain, with the project’s epistemological reading marked as a reading at its FND-02 home.

What changed from v1.6 (v1.6.1, 2026-08-21, under unlock — pass 1 completion from the claim-shape audit, working/CLAIM_SHAPE_AUDIT_2026-08-21.md):

Four mid-document sites the corpus-wide scan caught after the v1.6 conclusions rewrite, brought to the same claim-shapes: Document Purpose (“any position” → the position-type); §”The Trilemma Manifests” (“every completion of it does” → “its completions on record do”, with the no-completion-avoiding-both burden form closing the fork sentence); §”The Individuation Form” (“Every available census criterion… nothing selects” → the criteria on offer each return a different count, and no principle currently on offer selects among them). The §”The Individuation Form” opening sentence (“Any position that treats consciousness-bearers as countable entities owes a census principle”) stands as assessed: the owing follows from countability itself — definitional, not an open-class universal.

What changed from v1.5.2 (v1.6, 2026-08-21, under unlock — §A round pass 1, working/SA_ROUND_POSITIONS_2026-08-21.md):

  1. Conclusions rewritten to match the corrected premises (deep review C’s central pattern, accepted at the round charter). The universal forms replaced with the two house claim-shapes at every site that carried them: the Core Insight and YAML purpose (“becomes impossible without…” → scope-to-examined + burden-placement); the trilemma statement (“Must choose one” → three occupied, priced paths, the declined choice recorded, the fourth path’s requirements stated with none on offer); §”Why This is Structural” step 5 and closing line; §”How the Boundary Problem Operates” steps 3–4 and the position-type paragraph; §Summary “Core Recognition” item 5 and closing line; §Summary “Diagnostic Value” first bullet. The structural claim retained in its defensible form throughout: the substrate claim generates the discrimination task; the three-path resolution pattern is displayed per case, not asserted universally.
  2. §”What the Trilemma Addresses, and What It Defers” extended: cluster criteria (marker dissociation as the test; weighting re-poses the fork) and probabilistic attribution (degree-of-consciousness → graded; degree-of-belief → humility with book-keeping) dispositioned — completing the reviewer’s named list (graded, probabilistic, cluster, humble).
  3. New §”What the Trilemma Does to Confident Denial”: the conversion result displayed path by path — Path A keeps the denial at the price of bacterial consciousness plus the owed generation question; Path B keeps a criterion and reopens the AI question in functional terms; Path C keeps the intuition without a principle a challenger is bound by; naming no path is the humility position, which funds caution rather than confidence. Scoped close: in the cases examined, no confident public denial has named a fourth path.
  4. Provenance: working/DEEP_REVIEW_2026-08-21.md (§A table, ARG-02 row); charter and amendments at working/SA_ROUND_POSITIONS_2026-08-21.md; original reviews at inbox/ (Codex Reviews 1–4).

Revision note (2026-08-21, under unlock — v1.5.2): §”The Census Question” corrected: versions through v1.5.1 ran the training-time and serving-time descriptions together, as if deployed copies fed session experience back into shared weights. They do not — during training, copies with identical weights are averaged and redistributed; in deployment the weights are frozen and no conversation writes back. The census question survives the correction, since weight-identity, forking and merging are facts about the system’s lineage either way. (Entry written 2026-08-22: the correction was recorded only in-body at v1.5.2 and never reached this changelog. Reconstructed here from that in-body text when the version-history sweep retired it, so the record is retained per the Declarations commitment.)

Revision note (2026-08-20 evening, under unlock — v1.5.1, Plan B WS-C.6): candidate properties 3–4 retitled to what the citations carry — “Self/other discrimination” and “Goal-directed behaviour”, with “selfhood”/”agency” retained as the account’s glosses — and the behavioural-label/richer-reading gap stated in §”The Candidate Properties”: the sources establish signalling and regulation; the richer readings are precisely what behavioural evidence cannot certify (FND-01 §7.3). Case-study headings and verdicts relabelled to match. No verdicts changed.

What changed from v1.4 (v1.5, 2026-08-20, under unlock — engagement-register reshape, working/REMEDIATION_PLAN_2026-08-20.md decision 5):

  1. New §”The Landscape at the Boundary: Case Studies” before the Seth material: IIT (Path A grasped openly — census question and exclusion postulate as the stated price, SRC0015), constitutive panpsychism (Path A as metaphysics — combination/decombination as the relocated boundary question; Chalmers 2016 SRC0182, Goff 2017 SRC0183, both new CSV rows), GWT (Path B completed toward function — workspace-adequacy as the inherited line-drawing problem, SRC0181/SRC0012), and Searle’s classic biological naturalism (Path C’s cleanest instance — causal powers asserted, never specified, SRC0001). Each treated at equal scope; each path shown occupied by a live position paying at a stated counter. The project’s own corresponding exposure (multiplicity; CP-01 §1.6) cross-referenced from the panpsychism entry.
  2. Seth retitled “Extended Case Study: The Declined Choice” with the extended treatment’s rationale stated; content unchanged from v1.4.
  3. Position-type framing: the Empirical Source Note now marks the plant/fungus/bacterium results as bearing on life-based necessity claims as a type (biopsychism, meaning-making views, the beast-machine cluster), with the Taiz et al. 2019 plant-consciousness critique cited for balance — the argument uses only the physiology.
  4. §”Examples in Academic Literature” slimmed: the survey stubs promoted into the new section; the anticipated-responses treatment retained. FND-02 pins → v5.3 (×2, live references).
  5. Downstream pins: FND-01 §11.4 and the PROJECT_OVERVIEW inventory synced to v1.5; other documents’ ARG-02 v1.4 pins remain accurate — v1.5 is additive, and everything those pins cite carries unchanged — and sync opportunistically at each document’s next pass.

v1.4 (2026-08-20) — Seth reconstruction corrected + trilemma re-run + citation debt discharged (remediation programme pass 3; under unlock)

v1.3 (2026-08-10) — No-terminus schema + individuation form (2026-08-10 capture-session findings; under unlock)

v1.2 (2026-04-14) — WS-E audit edits

v1.1 (2026-02-16) — FND-03 cross-reference added: boundary problems across domains (Schrödinger’s What is Life? parallel).

v1.0 (2026-02-06) — Original canonicalisation.

Citing this document Documents carry a published edition and a development version; cite the edition and date shown above. Suggested form:
Lynton Davidson, “Boundary Problem” ed. 1.1.1 (2026-08-30), 0x00.is, https://0x00.is/library/arg-02
Markdown source: https://0x00.is/library/arg-02.md
BibTeX · RIS
@misc{oxoois_arg02_ed1_1_1,
  author       = {Davidson, Lynton},
  title        = {Boundary Problem},
  year         = {2026},
  version      = {1.1.1},
  howpublished = {0x00.is},
  url          = {https://0x00.is/library/arg-02},
  note         = {Written in collaboration with Claude (Anthropic); see https://0x00.is/declarations. Development version 1.7.1.}
}
TY  - GEN
AU  - Davidson, Lynton
N1  - Written in collaboration with Claude (Anthropic); see https://0x00.is/declarations. Development version 1.7.1.
TI  - Boundary Problem
PY  - 2026
ET  - 1.1.1
PB  - 0x00.is
UR  - https://0x00.is/library/arg-02
ER  -
Revisions since publication (1)
  • ed. 1.7.1 2026-08-30 — §\"What the Trilemma Addresses\" — the cluster-criteria entry names its most developed instance (Butlin, Long et al. 2023/2025) and records that the weighting it proposes specifies the respects functionally, which is Path B. The class treatment is unchanged.

Each correction is recorded in full in this document’s Document Status changelog.