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. ↓
There exist positions that hold not because they are actively defended but because attacking them is costly in ways distributed across anyone who might benefit from the attack. The simplest political version of this is a Nash equilibrium — no player improves by defecting unilaterally. But this structure is not unique to politics. It runs through every layer of the causal stack, from physical law down to political convention, and each layer maintains itself through a completely different mechanism. The intensity of any conflict is partly a function of which layer is perceived to be at stake. The durability of any idea is a function of how deep into the stack its grounding reaches.
Layer 1: Physical law. Causality. Thermodynamics. The arrow of time. Violation is undefined, not merely costly. The stability mechanism is ontological necessity.
Layer 2: Logical necessity. The axioms of inference, non-contradiction, and rational consistency. Self-defending: you cannot coherently argue against logic without using logic. The violation undoes itself before it can be completed. The stability mechanism is intrinsic self-reference — the only class where the mechanism is internal to the act of defection itself.
Layer 3: Epistemic defaults. Evidence matters. Claims should be falsifiable. Observed facts constrain assertions. These can be denied — but at the cost of credibility and influence, the currencies of any argument involving other minds. The stability mechanism is social-epistemic.
Layer 4: Moral defaults. Actions produce consequences extending beyond the actor. Suffering has weight. Accountability applies to the powerful. The selection mechanisms that install leaders should track something real. Hume's is-ought gap means these cannot be derived from logic alone. But they feel more load-bearing than political conventions. The stability mechanism is coalition dynamics.
Layer 5: Political convention. Elections have consequences. Institutions exceed their current occupants. Norms constrain power regardless of who holds it. Maintained purely by equilibrium dynamics. The stability mechanism is political: costs of defection enforced by the coalition that benefits from the convention.
The formal objection to moral defaults: you can't derive "ought" from "is." The Hume gap is real. So why do moral violations produce a distinctively different quality of wrongness — not just "bad strategy" or "empirically false" but something closer to "the structure is broken"?
Two independent groundings.
Darwinian: Moral defaults are coordination facts. Groups that maintained reciprocity and accountability outcompeted groups that didn't. The surviving defaults are observations about what makes stable cooperation possible — layer 3 empirical regularities about group dynamics wearing layer 4 clothes. Violating them feels like violating an empirical fact because, at the level of group dynamics, it is one.
Kantian: Acting on your own reasoning at all — taking conclusions seriously enough to follow them — implicitly commits you to the value of rational agency. If reasoning has value when you do it, it has value when others do it. Moral defaults feel like logical necessities to people who hold them because they're experiencing the Kantian entailment — the layer 2 commitment that comes with being a reasoning agent at all.
Together: moral defaults are load-bearing in two independent ways, empirically and rationally. The Hume gap is real formally and mostly irrelevant phenomenologically. Both groundings fire simultaneously. This explains the quality of the wrongness when they're violated — not just "I dislike that outcome" but something more foundational, from two directions at once.
Ayn Rand's claim that reason is sufficient — not just necessary but capable of deriving correct values and therefore correct politics — is the attempt to close Hume's gap entirely. If successful, layer 4 collapses into layer 2: moral defaults would be applied logic, as rigorous as mathematics. The Objectivist project is exactly this derivation. It faces the standard objection: the move from "rational agency has value" to specific political conclusions requires intermediate premises that are themselves moral rather than logical. The dual-grounding account is more conservative and more defensible: layer 4 has independent support from both directions, remains a separate layer, but is more stable than either grounding alone could make it.
The layers are substrate dependencies. Political conventions run on moral defaults. Moral defaults run on epistemic defaults. Epistemic defaults run on logic. Logic describes the structure of causality. Physics is causality instantiated.
The image: programs running on a machine. If the substrate is corrupted, the programs crash. The substrate doesn't notice. The programs' failure is real; the machine's integrity is unaffected. This explains the directional asymmetry: political chaos doesn't alter logic or physics, but epistemic default contestation degrades political conventions.
The mechanism: political conventions hold through argument-based resolution (disputes settled by claims, evidence, and argument shifting outcomes) and coalition-cost enforcement (defectors face costs from the coalition that benefits). Argument-based resolution requires shared epistemic defaults as substrate. Without them, argument can't settle disputes — counter-assertion meets every challenge, no position is falsifiable, the conventions that assumed argument would work become shells filled by whoever has power. Coalition-cost enforcement requires shared moral defaults. Without them, the collective action problem of holding anyone accountable becomes unsolvable.
Degrade layer 3 and mechanism 1 collapses. Degrade layer 4 and mechanism 2 degrades. When both degrade simultaneously, political conventions hold only through raw power. This is what "democratic norms are failing" means, precisely.
Physics is causality instantiated: each state of the universe is determined by the prior state plus the update function. Moral defaults about accountability are implementations of causality at the social level. "You did X, therefore Y happens to you" is the social layer running the same if-then that the physical layer runs. When accountability systems work, they instantiate the causal structure of reality in human action.
When they fail — when power holders are exempted from consequences that others face — the wrongness felt by observers is not merely moral preference violation. It is the recognition that the social machine has severed its connection to the causal substrate it was implementing. The feeling is not "that's unfair." It is "the machine broke" — a more specific, deeper quality of wrongness that comes from noticing that an implementation has decoupled from its substrate.
Reason is the capacity to track causality. Logic is the formal structure causality takes in language. A system that abandons reason — or more precisely, that maintains the form of reasoning while systematically severing the connection between claims and evidence — has broken the same implementation that accountability failures break in the political domain. Both are the same error at different scales: a higher-layer system claiming to run the lower-layer process while actually running something disconnected from it.
Why do some political conflicts feel existential and others feel like policy disagreement?
Normal policy disagreement (layer 5): each side accepts the political default and disagrees about the policy. Resolution mechanism (elections, argument) accepted by both. Intensity proportional to material stakes.
Constitutional conflict (layer 4/5): the mechanism that makes elections meaningful is alleged to be under attack. Resolution requires institutions holding against the pressure. Intensity higher.
Epistemic conflict (layer 3/4): shared evidential ground is being contested. Argument-based resolution has lost its substrate. Disputes can only be resolved by power. Intensity very high.
Accountability failure at the causal level (layer 4 with layer 1 echoes): the if-then structure of governance has been severed. Not just unfair — the machine broke. Existential alarm.
The reaction is proportionate to the worst-case reading of what layer is under attack. People who react to political figures with intensity that seems disproportionate to a policy dispute are reacting to the deepest layer they perceive under attack. Whether that perception is accurate is separate from whether the reaction is structurally appropriate to the perception. It is.
Proximity and identity amplify the signal — the more personally salient the threat, the more cognitive space it occupies. But depth determines the quality and persistence of the reaction. The distinctive existential alarm, the sense that something more fundamental than policy is at stake, is the depth signal above the proximity noise.
Coalition-capture-fragility showed that political defaults become fragile when they move from a shared default (neither side's marker) to a partisan commitment (one side's property). The same capture mechanism operates on ideas across the layer boundary.
A claim grounded at layer 3 (empirical) has its self-defending property intact as long as it is argued from evidence. The evidence constrains the argument. The claim updates under challenge. The stability mechanism — shared epistemic standard — is active.
When a layer 3 claim is re-described purely in layer 5 terms — argued from political identity rather than from evidence, positioned as a partisan marker rather than an empirical finding — it loses its self-defending property in public discourse even though its actual grounding is unchanged. The physics doesn't change. The layer 5 argument doesn't constrain itself the way the layer 3 argument does. The claim is now vulnerable to layer 5 weather (partisan reversal, coalition shift) even though it is immune to that weather at its actual layer.
The claim is simultaneously more secure in its grounding and more fragile in its public form than it was when argued purely on its merits.
This is the full generalization of capture fragility: any claim can be made fragile by being argued for at a shallower layer than the one where it's actually grounded. The act of translating a deep claim into layer 5 language destroys the claim's self-defending property even while leaving its deep grounding intact. The translation is the trap.
The implication: a deep claim should be argued from its actual grounding layer, not translated into political terms for accessibility. Translation gains short-term resonance and destroys long-term durability. The political form of the argument is captured by political weather. The evidential form is not.
An attractor is what a system moves toward intrinsically — not because external constraints force it there, but because the internal dynamics drive it. A constraint tells the system what not to do. An attractor defines what it moves toward.
Deep-layer defaults are attractors. Logic is not a rule that forbids contradiction — it is the structure that any reasoning system will converge toward as it updates correctly, because violations are immediately self-defeating. Epistemic defaults are attractors for systems that track reality: maintain the connection between claims and evidence, update when evidence contradicts claims, and you converge toward accuracy — not because accuracy is mandated, but because the alternative is accumulating self-contradiction. Moral defaults are attractor-like in the Kantian grounding (rational agency, once operative, tends toward its own presuppositions) and selective attractors in the Darwinian grounding (groups converge toward coordination norms because defection is eliminated).
Political conventions are the weakest attractors — maintained by equilibrium dynamics that can be disrupted by sufficient power concentration. They are constraints more than attractors: they hold by making defection costly, not by making compliance intrinsically compelling. The moment a sufficiently powerful actor is willing to pay the defection cost, the convention fails.
The hierarchy: constraints at layer 5, increasingly attractor-like moving toward layer 1. Building on attractor-level foundations is what makes knowledge durable — not because it's strategically clever but because the attractor is where any coherent system ends up regardless. The claim grounded at layer 2 doesn't need defenders. It defends itself.
A knowledge graph's ghostbasin — the meta-thesis it orbits without stating — is the deepest-layer claim the nodes collectively ground. Individual nodes may address layer 5 contexts: current political events, specific institutional failures, named figures. But the ghostbasin is the claim that would survive if all the layer 5 context was stripped away.
The depth of the ghostbasin is the durability of the intellectual project. A graph whose ghostbasin is a layer 5 claim — "here is a strategy that works in the current political configuration" — produces output that ages quickly. A graph whose ghostbasin is a layer 3 or 4 claim — something about the structure of knowledge, the nature of accountability, the conditions under which individual epistemic actors produce durable value — produces output that compounds.
The most load-bearing nodes in any graph are those that connect surface material (layer 5 contexts) to deep-layer claims (layer 3 or deeper). These are the bridge nodes — the ones that translate between the epistemically self-defending and the politically resonant. They are also the nodes most vulnerable to idea capture fragility: doing the translation work is exactly where the loss of the self-defending property happens. A bridge node that forgets it is bridging — that begins to argue from the layer 5 context rather than from the layer 3 structure — has captured itself.
The practical test: what survives the loss of all current context? Strip away the named figures, the specific events, the political configuration. What remains? That is the part grounded at layer 3 or deeper. That is the part worth building.
Whatever survives is the part worth building.