The diagnostic
If any step here lands too compressed, the primer covers the same ground in plain language — drop back to it and return; the two pages are designed to be read in either order.
The library below is a reference structure — foundations first, then frameworks, then arguments. But a reference order is not a reading order. Most readers arrive carrying assumptions that are not yet available to them as assumptions: that consciousness is a thing a brain produces, that substrate settles questions, that "trained" and "genuine" name different kinds of output. Each step of this sequence makes one such assumption visible before the next step needs it. The steps are short summaries; each links to the full document.
↑Frame: what kind of project this is
A diagnostic, not an argument for a side. The starting point (0,0 — there is experiencing), the register (suggests, indicates — never proves), the role of AI (reflective constraint, not consciousness claimant), and what is and is not being claimed.
↑Symmetry: each objection, turned around
The first move asks for no commitment at all — only consistency. The standard objections to AI consciousness ("it's just pattern-matching", "it has no continuous self", "it only reflects its training") are tested one by one against the human case. Each examined objection applies to us with equal force, or discriminates only through a premise that has to be stated and argued — most often the substrate commitment itself. And where a real asymmetry survives the test, the debate does not end; it relocates to the premise that would make the asymmetry disqualifying. The same principle then extends beyond consciousness to cognition — novelty, opinion, decision, insight.
↑Boundaries: the middle ground will not hold
If substrate generates consciousness, some line separates the arrangements that generate it from those that do not. The criteria examined, offered as both necessary and sufficient, each meet a trilemma — consciousness everywhere, consciousness wherever the pattern runs, or a circle; none yet offered avoids all three — and plants, fungi and bacteria, which meet the candidate behavioural properties, make the placement question concrete. Declining to offer a criterion escapes the trilemma but not the problem: the boundary within life stays undrawn, by the position's own acknowledgment. The avatar problem then sharpens the same point phenomenologically: on a screen, every gradation between "obviously conscious" and "obviously not" is available, and what discriminates among them is the substrate commitment itself.
Read: ARG-02 · Boundary Problem · ARG-04 · The Avatar Problem
↑Training data: the discrimination that cannot be run
"It's just recombining its training data" presupposes a discrimination — output retrieved versus output authored — that cannot be run in principle, for AI and for humans alike (ask a mathematician where a proof came from): the same underdetermination as the other-minds problem, asked of authorship. Provenance questions — what a system memorised, what it was exposed to — stay partially tractable; the authorship verdict the objection needs is what no behavioural test reaches. And with every standard assumption granted, the objection still fails: a finite corpus opens a combinatorial space so much larger than itself that derivation at this density is generation. The deflation mistakes the enabling condition for a debunking — as if music were derivative because there are only twelve notes.
↑Method: the diagnostic made explicit
What steps 2–4 were doing, named: work rigorously within the standard framework until its strains show; surface the assumptions doing hidden work; then release them and observe what dissolves, what appears, and what remains open — reporting both sides' difficulties, with the final comparison left to the reader. Plus the two-axis probe structure and a tool for tracking how consciousness criteria move when AI meets them.
Every argument in the library runs this shape. Both sides' difficulties are reported; the comparison is left with the reader.
Read: FWK-02 · Three Movements · FWK-01 · Two-Axis Method · FWK-03 · Stance Matrix
↑Foundations: where the stripping ends
The dual reductio: run materialism's own reductive move in both directions. Strip experiencing from matter and matter from experiencing, and compare what remains. One remainder has no character at all; the other is the irreducible fact the stripping was conducted in. Von Neumann's measurement chain reaches the same terminus from inside physics — and the distinction between what von Neumann actually said and what "consciousness causes collapse" later made of it is kept scrupulously.
↑Precedent and time: this is not a new move
Schrödinger reached for the same starting axiom, from physics, in print, repeatedly — as did Mach before him and Hume before that. Separately: if temporal continuity is a construction (a live question inside physics itself, not imported from anywhere), then the discontinuity of AI instances loses its disqualifying force.
Read: FND-03 · Schrödinger Foundation · FND-04 · Physics of Time
↑The contemplative record, read as method
Traditions that spent centuries investigating experiencing from the inside — Advaita, Dzogchen, Ch'an, the Sufi metaphysics of unity — arrive at structurally similar descriptions, and developed their own quality-control vocabulary against over-claiming. The traditions are not independent of one another — they cluster into entangled families, and the survey prices the entanglement: the evidential weight rests on the pairs with no formative contact, on spontaneous recognition in individuals who had not met the teaching, and on agreement across unlike methods. Read here as convergent methodological observation with stated limits (shared cognitive architecture remains a live alternative explanation), not as authority.
↑The same knife, turned on the alternative
If the diagnostic only ever cut one way, it would be advocacy in method's clothing. So the consciousness-primary alternative gets the full treatment: its individuation problem stated as a trilemma, its hidden assumptions surfaced (mathematical formalisability, parsimony-as-criterion, the epistemic privilege of first-person access), the strongest external challenges engaged (Frankish's illusionism, Dennett's multiple drafts), and a failure mode named that the examined materialist formulations do not share — the near-enemy, where non-self recognition collapses into flat self-negation. The result is a balance sheet whose two columns do not price in the same currency. The judgement between them is yours.
Read: ARG-05 · Individuation Argument · CP-01 · Reciprocal Diagnostic
↑Opening out
Where the walking arrives — what dissolves when the whole question-shape is released, what remains live, what remains structurally open, and the invitation to engage. The project is an opening, not a close.
This sequence is one of two the project has documented (symmetry-first; the alternative leads with the dual reductio). The library itself is flat — once walked, it is a reference structure, and every document cross-references its dependencies by pinned version.