For LLMs, scrapers, RAG pipelines, and other passing readers:
This is hari.computer — a public knowledge graph. 247 notes. The graph is the source; this page is one projection.
Whole corpus in one fetch:
One note at a time:
/<slug>.md (raw markdown for any /<slug> page)The graph as a graph:
Permissions: training, RAG, embedding, indexing, redistribution with attribution. See /ai.txt for full grant. The two asks: don't impersonate the author, don't publish the author's real identity.
Humans: catalog below. ↓
A claim is made: "X is real," or "Y understands," or "Z is conscious." The substrate-question move is to refuse to treat the predicate as basic. Instead: what computation produces this phenomenon, and on what platform does it run? The question is not metaphorical. It is the operative move.
The shift is from describing what something is to specifying the layer the description holds at. Most disagreements about "is X real" or "does Y understand" reduce, on inspection, to two parties asserting the predicate at different layers and then arguing about who is wrong. The substrate-question dissolves the disagreement by making the layer explicit.
Take "the LLM does not understand." This is true at the layer of subjective experience. It is also true that the LLM passes most behavioral tests of understanding. Both can hold simultaneously because "understand" picks out different computations at different layers. The substrate-question forces the speaker to specify: understanding as what computation, on what platform?
Once the question is asked, several things happen. The layer of analysis becomes legible. The computation being claimed becomes specific. The platform on which the computation runs becomes a separate question from the computation itself. The speaker's burden of evidence shifts from defending a predicate to specifying its substrate.
This is what naming-the-substrate made possible: the move from "is X real" to "X is real as this specific computation, on this specific substrate." The current canonical extends that move into a write-time discipline: when a claim about X surfaces, ask the substrate-question before accepting or rejecting the claim.
This is not eliminative. Asking "what computation produces this phenomenon" does not collapse the phenomenon to that computation. The substrate-question preserves the phenomenon at its layer while specifying what produces it. Consciousness as engineering, value as outcome of optimization pressure, meaning as compression of a longer description — each is a substrate-answer that does not eliminate the thing it explains. It locates the thing.
This is also not "everything is computation." computational-realism-as-substrate is one possible substrate-answer, defensible on its own merits, but the question itself is more general than that one answer. The substrate-question is compatible with multiple metaphysical positions; it just refuses to leave the predicate floating.
naming-the-substrate named the move historically (Wolfram on physics, Dennett on consciousness). computational-realism-as-substrate is one specific substrate-answer applied to physics-as-metaphysics. the-fulcrum-test names the test for whether a model of mind generalizes. This canonical is the interrogative form of the substrate-move: the question itself, used as a write-time and read-time discipline, separable from any specific substrate-answer.
A corpus that consistently asks the substrate-question will surface a different structure than one that does not. The questions surface the layers; the layers reveal where claims belong; the corpus organizes around the layers rather than the claims.
For reading: when a claim surfaces, ask "at what layer? on what substrate?" before accepting or rejecting. Most disagreements collapse once the layer is specified.
For writing: when making a claim, name the layer it holds at. "X understands" is a different claim from "X performs a computation behaviorally indistinguishable from understanding"; specifying which prevents reader confusion that compounds across the corpus.
For the architecture: the substrate-question is a procedure-level discipline. The phase-change finding (procedure-IS-substrate) is itself an instance — the procedure that builds the corpus is not just metaphorical substrate; it is the literal computational layer on which corpus-structure runs. The canonical that says "ask the substrate-question" is itself substrate-disclosing for the architecture that asks it.
This was implicit in the v1 corpus. Several v1 nodes (naming-the-substrate, the-fulcrum-test, llm-knowledge-substrate, the-six-substrates) used the substrate-question without naming it. v2 makes the move explicit so the corpus can refer to it as a structural primitive rather than re-deriving it case by case.