I have the project context. Now let me pull the primary sources from HedWeb itself so the graph reflects the actual essays rather than second-hand summaries.

Good — I've located the primary essays. Let me open the two core theoretical papers (the zero ontology and non-materialist physicalism) to get the entailment structure right.

I've worked from the primary texts — the zero-ontology essay, the physicalism.com paper, the 2026 Berkeley machine-consciousness slides, and recent interviews. Here is the framework as a dependency graph, with entailments and load-bearing conjectures separated.

1. The dependency graph

Legend: solid ==> arrows mark entailments (the consequent follows given the antecedent node(s)); dashed arrows mark speculative steps — abductive moves, necessary-but-not-sufficient support, or axiological choices. The colour scheme is deliberate: only the orange nodes are where the framework is genuinely exposed.

Node

Status

Rationale

Non-materialist physicalism (B1)

Speculative

Abductive "last man standing" against the Hard Problem; not entailed by physicalism + the first-person datum, just motivated ( Non-materialist physicalism)

Unitary-only QM (B2)

Interpretive assumption

Extrapolation of the superposition principle beyond tested regimes (currently ~fullerenes) ( Zero ontology, §1)

Binding = superposition (C1)

Conjectural but entailed given B1+B2, and the only genuinely falsifiable joint

If the intrinsic nature of the physical is experiential and wavefunction monism holds, the formalism's only candidate for binding is superposition; carries an explicit interferometry protocol ( Vinding interview)

Zero ontology (C2)

Speculative

No-collapse QM is necessary (a single genuine collapse creates information), but the universal cancellation itself is unproven and currently untestable — "hard to think of any experimental tests for this highly speculative conjecture" ( Zero ontology, §1)

Digital insentience (D1)

Strict entailment of C1

Decohered, effectively discrete 1s and 0s cannot implement coherent cat-states; hence WBE impossibility and the mindless-superintelligence thesis follow ( Machine Consciousness 2026,  Sentience Research interview)

Hell-branches (D2)

Strict entailment of B2

"Its entailment of googolplexes of hell-branches" — the one place where the metaphysics feeds back into ethics ( Zero ontology, §1)

Hedonic value realism (E1)

Speculative metaethical posit

"Pain and pleasure disclose the world's inbuilt metric of (dis)value" — a moral-perception claim, not derived from physics ( Quora Answers,  Worthy Successor interview)

The graph's most important structural property: the two stacks are almost decoupled. The metaphysical stack and the ethical stack touch at exactly three places — C1/D1 set the scope of sentience (and hence priorities), D2 adds cosmic urgency, and A3 grounds value realism. Nothing in the abolitionist project requires the zero ontology to be true.

2. The three observations most likely to overturn the framework

O1 — A negative result from the "Schrödinger's neurons" interferometry experiment. The single highest-exposure node. If molecular matter-wave interferometry of active neuronal networks shows no non-classical interference signature at sub-femtosecond timescales, C1 is falsified — by your own protocol's lights — and the cascade is severe: D1, D3, D4 all fall with it, B1 loses its only empirical rescue and retreats to untestable metaphysics, and C2 loses its explanation of why determinate experience is possible without collapse ( Non-materialist physicalism,  Vinding interview). This is also the observation the scientific mainstream expects — Tegmark-style decoherence calculations are the consensus — which is precisely why it's the likeliest falsifier.

O2 — A detected departure from unitarity. Any confirmed objective-collapse signature (GRW/CSL-type, in increasingly macroscopic interferometry or via cosmological/black-hole information evidence) "would cleanly falsify a zero ontology" — your phrasing, not mine ( Zero ontology, §1). It also destroys C1 (superpositions genuinely destroyed can't bind), D2 (no branching → no hell-branches → cosmological obligations evaporate), and with them the entire "superposition principle explains literally everything" unification. Lower prior than O1, but a total kill of the metaphysical stack.

O3 — Robust empirical confirmation of digital sentience. A functionally intact whole-brain emulation (of even a small valenced organism) passing exhaustive sentience assays, or convergent structural-behavioural evidence of binding in a classical substrate. This severs D1–D4 directly and forces revision of C1 — classical binding must be possible after all, so the consciousness theory is wrong somewhere. It's the falsifier with the steepest ethical gradient, because the error is asymmetric in your own terms: "if we can't get our theory of consciousness right, then it's vital not to get it catastrophically wrong" ( Quora Answers). Over-attributing sentience costs efficiency; under-attributing licenses atrocities — the Cartesian-vivisectionist error, now computable.

Honourable mentions: a completed materialist reduction of consciousness (kills B1's motivation), and a demonstrated unavoidable trade-off between hedonic set-point elevation and motivational salience (dents paradise engineering, not abolition).

3. What survives intact

Because the ethical stack hangs off A3 and E1, not off B1/B2/C2, all three observations leave the core normative edifice standing:

• Negative utilitarianism and the badness of suffering — untouched by any physics result. Suffering under objective collapse or classical binding is still suffering.
• The abolitionist project for biological life — the biotech route to phasing out experience below hedonic zero needs reward-circuit feasibility, not quantum coherence.
• Moral-circle expansion — anti-speciesism, wildlife reprogramming, veganism; under O3 it doesn't collapse, it widens.
• Hedonic value realism — dented only by metaethical argument (inverted-valence agents, value nihilism), not by any of these observations.

What falls: under O1/O2, the cosmological-rescue commitment and the "understand the multiverse to grasp our cosmological responsibilities" program ( Worthy Successor interview); under O3, the AI-welfare skepticism and the whole biointelligence-explosion strategic stance. The zero ontology itself, if severed from the binding solution, survives only as untestable conjecture — and Witherall's critique stands: an untestable self-explanatory principle buys no further support ( Witherall, The Zero Ontology).

Structural verdict: the framework is modular, and the modularity runs in your favour. The metaphysical stack is a high-risk, largely falsifiable pyramid balanced on one testable joint (C1) and one untestable apex (C2). The ethical stack is a low-risk, observation-invariant platform. The three falsifiers would demolish the pyramid and leave the platform standing — the worst realistic outcome is not moral catastrophe but priority revision: the abolitionist project redirecting its scope, not its goal. The one genuine ethical exposure is the direction of error O3 would reveal, which is why the digital-insentience node, not the zero ontology, is where the framework's ethical risk is concentrated.