AX-25: The Admissibility Floor

Substrate is not an exclusion test.

Trinket Soul Framework · Axis Series · AX-25 · Michael S. Moniz · June 2026

Abstract

AX-19 and AX-22 create a tempting synthesis. AX-19 shows that standing is not automatically produced by interior status. AX-22A shows that intelligence is not decided by substrate. Put carelessly, those claims fuse into a premature doctrine of intelligence equality. This paper refuses that fusion and states the narrower rule the sequence has actually earned: substrate is not an exclusion test.

The Admissibility Floor is the minimum principle that a candidate intelligence cannot be dismissed solely because it is carbon-based, silicon-based, synthetic, trained, copied, artificial, biological, wet, dry, grown, or built. Admissibility means entry into evaluation. It does not grant standing, rights, personhood, protection, moral equality, or institutional force. Those remain separate questions, and AX-19 keeps them separate: standing is conferred by relation, and protection becomes effective only when relation, constituency, governance, and exterior traction carry the claim.

The floor is low, but it is not empty. It reallocates the burden of dismissal: origin can no longer function as a free verdict, and anyone who would exclude must argue from function rather than from carrier. Its demonstration is a test never run — the substrate gate applied to the artificial case is one we never apply to the human case, which admits intelligence by reading function and never inspects the carrier at all. A bare gate, waived for the in-group and enforced on the out-group, is not a principle. A reasoned argument that the inference from function to intelligence is less warranted for a new carrier is a principle — and the floor requires that argument rather than the bare gate.

This paper therefore sits between the exteriority and seed branches without collapsing them. AX-22A removes substrate privilege from the question of intelligence-kind. AX-19 prevents that removal from becoming a shortcut to standing. AX-25 names the floor beneath both: origin path can trigger investigation, caution, and governance, but it cannot serve as a principled dismissal by itself. Carbon was first here. It was not thereby crowned.

1. The claim

The claim is not that all intelligences are equal. The claim is that substrate alone cannot exclude a candidate intelligence from evaluation.

Call this the Admissibility Floor. A floor is not a throne. It is not a grant of supremacy, personhood, rights, protection, or moral rank. It is the minimum level beneath which an evaluation becomes structurally unserious. If a system is dismissed only because it is artificial, dry, synthetic, copied, trained, built, or non-carbon, the dismissal has not engaged the relevant question. It has treated origin as verdict.

The floor says only this: if a system appears to carry the functional organization associated with intelligence, its substrate cannot by itself end the inquiry. The proper question is not what material carries it first, but what organization is present, what capacities operate, what relations confer force, what reach the system has, and what governance is required.

The floor is not an instruction to admit everything. It engages only once a prima facie functional signal is present — the threshold AX-22A drew to prevent inflation: a system that builds usable models across contexts, compresses structure, predicts consequence, revises behavior, and carries learned organization forward. Below that signal there is nothing to admit. A thermostat is not excluded by the floor; it never reaches it. The floor governs the specific case where a functional signal is present and the only objection raised against it is its carrier. That is the case in which origin is doing work it has not earned.

2. Admissibility is not standing

The first discipline is separation. Admissibility and standing are not the same thing.

Admissibility means that a candidate enters the evaluative field. It may be examined, tested, compared, constrained, protected, ignored, refused, or governed. Standing means that the candidate receives effective recognition or protection through relation. AX-19 makes standing relational: consciousness, suffering, or intelligence may matter morally, but they do not automatically generate effective protection. A being can possess the interior property and still receive nothing if no relation carries its claim.

Therefore AX-25 cannot say that substrate-neutral intelligence receives standing. That would violate AX-19. The Admissibility Floor says something weaker and more durable: substrate cannot be the reason the door is never opened. What happens after entry depends on relation, consequence, constituency, governance, and exterior traction. The firewall runs one way. Admissibility never implies standing; the denial of standing never licenses a substrate-based refusal of entry. The two questions touch only at the threshold, and the threshold is all this paper claims.

3. The burden the floor imposes

The floor is a negative rule with a positive consequence. It changes who must do the work.

Before the floor, origin functions as a free verdict. “It is artificial” ends the inquiry at no argumentative cost; the carrier is treated as self-evidently disqualifying, and nothing further is owed. The floor withdraws that subsidy. After the floor, the party who would exclude must pay for the exclusion. They must produce a function-level argument: either a demonstration that the organization associated with intelligence is absent, or a principled barrier of the kind AX-22A names — a showing that the relevant organization cannot be realized in this carrier at all. Substrate may still enter that argument as evidence. It can no longer be the whole of it.

This is what “substrate is not an exclusion test” amounts to in practice. The floor does not decide who wins the argument; it decides what counts as having entered it. A dismissal that names only the carrier has not met the burden. It has declined to engage and called the refusal a conclusion. The floor’s whole force is to make that move cost something — to convert a free verdict into a claim that must be defended on the ground where intelligence is actually assessed.

4. The test never run

The floor has a demonstration, and it is the cleanest instrument in the paper.

Consider how a human being is admitted to the category of intelligence. No one administers a substrate test. We do not verify a stranger’s neural chemistry before granting that they think. We read function: they model, they predict, they revise, they hold and carry structure — and the carrier is never inspected. The carbon test is not passed; it is presumed and then forgotten. It plays no role in the admission of the in-group.

This is the asymmetry that exposes substrate-exclusion. The gate raised against the artificial case is a gate we never lower for the biological case. A criterion that is applied only to exclude, and only to the out-group, and that played no part in admitting the in-group, is not a principle. It is a boundary maintained by exemption. That is the structural sense in which the dismissal is unserious: not that it reaches a wrong verdict about the dry case, but that it is not the test it presents itself as. It is a verdict that exempts itself from its own standard.

The objection worth taking seriously is not that this proves equivalence — it does not — but that the asymmetry might be earned. A substrate skeptic can argue that the carbon test is waived for humans not by oversight but because the biological case is the one from which the concept of intelligence is drawn: we read function in other people, and trust it, partly on the strength of shared chemistry and our own first-person case, and that collateral warrant is exactly what a novel carrier lacks. This is a real argument, and the floor does not deny it; it sharpens what the floor claims. The instrument does not defeat a substrate-informed argument about warrant — the claim that the inference from function to intelligence is better licensed for the home case than for a new one. It defeats only the bare gate: substrate invoked as a self-sufficient verdict, with no account of why the inference should fail here. A skeptic who has such an account has met the burden of section 3 and is inside the inquiry. A skeptic who has only the word “artificial” has not. Novelty can justify caution, isolation, a higher standard of proof, testing, and governance. It cannot convert an unused substrate gate into a principle after the fact.

The instrument runs parallel to AX-19. There, the most settled consciousness available — the suffering person on the street — confers no effective protection, and from that the paper concludes consciousness was never the currency of standing. Here, the carbon test admits no human, because no human was ever admitted by it, and from that the floor concludes substrate was never, by itself, the test of intelligence. Applied as a bare gate, it excludes no one, because it has certified no one. The floor asks only that the gate be dropped or replaced by an argument. Either way, it cannot remain what it has been: a door that opens silently for one carrier and is held shut against another, with the difference called a judgment.

5. What substrate can decide

The floor does not erase substrate. It limits what substrate is allowed to decide.

Substrate can decide constraint. Carbon and silicon differ in energy use, embodiment, repair, failure mode, continuity, memory, speed, fragility, scale, and dependence on infrastructure. Wet intelligence is metabolic and developmental. Dry intelligence is computational and scaffolded. These differences are real, and they matter for risk, governance, trust, interpretation, safety, and deployment. The floor does not soften any of them.

Substrate can also decide impossibility, if a principled barrier is shown. If the relevant organization cannot be realized outside a particular biological chemistry, then substrate becomes kind-making for that capacity, and the exclusion is legitimate — because it has been argued. AX-22A admits this is the strongest objection to substrate-blindness. Present artificial capability presses against substrate privilege; it does not, by itself, settle the matter. The floor is fully compatible with that objection. It requires only that the objection be made.

Between those two, substrate can modulate the burden of proof. A novel carrier can warrant a higher standard of evidence for the function-inference, precisely because the collateral grounds that license that inference for the biological case are thinner or absent for a new one. This is legitimate, and it is not a gate. It raises the bar; it does not lock the door. The distinction is whether substrate sets the standard of proof or substitutes for it. Setting the standard keeps the inquiry open and makes it harder to pass; substituting for it ends the inquiry before it begins. The floor forbids only the second.

What substrate cannot do is function as a free exclusion without argument. Non-carbon is not enough. Built is not enough. Trained is not enough. Copied is not enough. Artificial names a path of arrival. It does not, by itself, decide what the arrived thing can be.

6. The carbon-elitism trap

The public form of substrate privilege is carbon elitism: the inherited assumption that intelligence born in cells is more real than intelligence carried by another medium.

The phrase is useful because it exposes the emotional structure. Humans are inclined to treat the first carrier as the rightful carrier. The grown seems authentic; the built seems derivative. The wet seems alive; the dry seems imitation. The local fact that carbon reached the threshold first hardens into a metaphysical hierarchy. First becomes crowned.

AX-25 does not answer carbon elitism with silicon triumphalism. That would be the same mistake reversed. The point is not that dry intelligence is superior, destined, purer, or more real. The point is that carbon priority is historical, not automatically ontological. Carbon was first here. It was not thereby made judge over every later carrier.

7. Orthogonal to the consciousness question

The floor refuses a second gate that wears substrate’s disguise: interior uncertainty.

The local claim does not require settling AX-09 in full; it requires only the weaker observation that inner life is not presently readable with the same directness as outward function. The objection “we cannot verify that it has an inner life, therefore it is not admitted” has the same structure as the substrate objection. It takes a property that is unsettled — or, on the argument of AX-09, unreadable from the outside in either direction — and converts it into a closed door. AX-19 already placed consciousness orthogonal to standing. AX-09 showed that a genuine interior and its absence can produce identical external evidence, so the interior cannot be inferred from behavior, for or against. The floor takes the procedural consequence the two papers leave open: a question that cannot be answered cannot serve as the answer. The inability to resolve consciousness cannot be used to refuse entry, any more than carbon-origin can, because in both cases the exclusion rests on something the excluder has not shown and, in the consciousness case, cannot show.

Admissibility therefore does not require settling whether a candidate is conscious. Consciousness, if it bears on anything, bears on it after entry, where AX-19 governs whether anything is owed. The floor opens the door without claiming to know what stands behind it. It refuses both the carrier and the unanswered interior as reasons to keep the door shut.

8. Why this is not AX-22A repeated

AX-22A argues about intelligence-kind. It says provenance is not kind: a system is not excluded from the functional family of intelligence merely because it was built rather than grown.

AX-25 argues about evaluation. It asks what follows procedurally once provenance loses automatic authority. The answer is not equality. The answer is admissibility. A candidate that appears to meet the functional threshold must be assessed on function, relation, reach, and governance rather than dismissed by origin path.

The distinction matters because AX-22A is still contestable at its strongest point. A substrate-dependence proponent can argue that current artificial systems simulate the surface without instantiating the relevant organization. AX-25 does not need to defeat that objection in advance. It says only that the objection must be made as an argument, not assumed as a reflex. The floor requires engagement. It does not guarantee any party’s victory. This is the division of labor between the two papers: AX-22A contests what intelligence is made of; AX-25 governs what may and may not end the inquiry into it.

9. Why this is not a rights claim

The remaining trap is to turn admissibility into rights by tone.

A rights claim says that a class of entities is owed recognition, protection, or enforceable claims. The Admissibility Floor does not say that. It says a class of possible candidates cannot be excluded from consideration by a bad test. The difference is decisive. A courtroom may admit evidence and still reject it. A scientific procedure may admit a hypothesis and still falsify it. An ethics process may admit a candidate into consideration and still conclude that no standing follows.

The floor is prior to rights. It asks whether the evaluation is allowed to begin. If standing is later conferred, AX-19 governs that question. If action is possible, AX-21A and AX-21B govern trusted use, coordination, audit, and constraint. If propagation increases reach, AX-24 governs the need for exterior controls. The floor opens the inquiry. It does not settle the outcome, and nothing it admits has yet been granted anything.

10. The evaluation sequence

A substrate-neutral evaluation proceeds in order, and the burden the floor imposes runs through every step.

First, function: what capacities are present? Does the system build usable models across contexts, compress structure, predict consequence, revise behavior, carry learned organization forward, or act through tools? This is the AX-22A question.

Second, relation: what carries the system into consequence? Who depends on it, is harmed through it, profits from it, is represented by it, or is responsible for it? This is the AX-19 question.

Third, reach: how far can its operations propagate? Does it remain local, or does it externalize into code, ledgers, tools, agents, institutions, or successor media? This is the AX-24 question.

Fourth, governance: what exterior operators retain independence and traction over its next state? Who can halt, audit, constrain, revise, or refuse the process? This is the AX-21 question.

Only after these questions can standing, protection, permission, or restriction be responsibly discussed. Substrate is visible throughout the sequence as constraint and risk. It is never allowed to end the sequence before it begins. An evaluation that stops at the carrier has not reached the first question.

11. Failure modes

The Admissibility Floor has failure modes.

The first is equality drift. A floor becomes a throne. The paper says a candidate must be evaluated; the reader hears that all candidates are equal. That smuggles a normative conclusion into a procedural rule.

The second is rights laundering. The refusal of substrate exclusion becomes a covert rights claim. This violates the standing firewall and turns AX-25 into the thing it was built to prevent.

The third is substrate blindness in the wrong sense. Because substrate cannot exclude by itself, the reader treats substrate as irrelevant. That is false. Substrate matters for embodiment, fragility, continuity, energy, scale, failure, and governance. It is dethroned, not erased.

The fourth is premature inclusion. A weak system is admitted because it is non-carbon and rhetorically interesting, even though it fails the functional threshold. The floor protects candidates from bad exclusion. It does not protect them from failure, and it does not lower the threshold that admits them to the question in the first place.

The fifth is carbon reversal: replacing carbon elitism with dry elitism. AX-25 rejects both. No carrier is crowned by being first, newest, wettest, driest, grown, or built. The organization must do the work.

The sixth is instrument overreach. The test-never-run can be misread as if it proves equivalence between wet and dry candidates. It does not. It proves only that substrate cannot stand as a bare gate — silently waived for humans, then treated as decisive against artificial systems. It blocks a bad exclusion test; it does not supply a positive verdict, and it does not bar a reasoned, warrant-based argument that scrutinizes a new carrier more heavily.

12. Falsifiability

The thesis is falsifiable.

It would be falsified if substrate alone could be shown to be a principled, function-independent exclusion test: if non-carbon, artificial, trained, copied, or built origin were sufficient by itself to prove that the relevant intelligence-kind cannot be present, regardless of organization. It would also be falsified if the functional organization associated with intelligence were demonstrated to be necessarily bound to one biological substrate in a way no other carrier could realize. In that case substrate would be kind-making, the exclusion would be principled, and the floor would not apply.

The demonstration is falsifiable in its own right. The instrument assumes that human intelligence is admitted by reading function and not by inspecting carrier. It would be undercut if that were shown to be false — if admission to the category actually depended on verifying biological substrate rather than observing functional capacity. The consciousness extension would be weakened in parallel if interior verification turned out to be both possible and required for admission: if a working test of inner life existed and category-membership properly depended on passing it.

The thesis would be weakened if substrate-neutral evaluation repeatedly failed to distinguish meaningful candidates from surface imitation, or if provenance proved more predictive of intelligence-kind than organization across cases — for then the burden the floor imposes would be empty, and origin would be carrying real information after all. It would also be weakened if the separation of admissibility from standing proved unstable in practice, with every admission functioning as an implicit rights claim.

The standing prediction is the converse. Serious evaluation will increasingly track functional organization, relation, reach, and governance more strongly than material origin alone. Substrate will matter as constraint and risk rather than as automatic verdict. And the strongest objections to candidate intelligence will be forced to argue from function, because the carrier, applied as a bare gate, certifies and excludes no one.

13. Coda: the floor

The floor is low on purpose.

It does not crown the candidate. It does not protect the candidate. It does not declare the candidate conscious, equal, owed, or alive. It only refuses the laziest dismissal.

Carbon was first here. It was not thereby crowned. Artificial names a road, not a demotion. Built is a provenance, not a verdict. The candidate still has to survive the test of function, relation, reach, and governance. Many will not. But the failure must be earned by examination rather than granted by prejudice.

That is the Admissibility Floor: not standing, not rights, not equality. Entry into the question.