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. ↓
Connective vocabulary compounds across frames. Domain-specific vocabulary compounds within frames.
"Consensus destroys dissenting signal" — every word is connective. A farmer, surgeon, senator, kindergarten teacher can parse it. The claim compounds: anyone who encounters it can apply it in their domain, teach it to others, generate new instances. It travels.
"The Gödelian horizon bounds self-abstraction via epiplexity" — three domain-specific terms in one sentence. Only readers already inside the formal-systems frame can parse it. The claim is deeper — it connects incompleteness, undecidability, information complexity, and biological free energy into a unified boundary. But it compounds only within the audience that has the vocabulary.
This isn't a quality distinction. It's a compression problem. Both claims name real mechanisms. The first achieves broader compression — more minds can decompress it. The second achieves deeper compression — it unifies more phenomena. The ideal is a claim that does both: deep unification in connective vocabulary.
Embedding 307 claims from 15 sources and running tradition distillation across 10 reference frames makes the bifurcation visible.
A knowledge graph's claims split into two measurable registers:
Register 1 (institutional/systemic): High uniqueness, high centrality. Nobody else says this, and every frame finds it relevant. These name mechanisms about how institutions, evaluation, knowledge systems, and political defaults work — in vocabulary that connects to every domain. Mean centrality: 0.783.
Register 2 (formal/technical): High uniqueness, low centrality. Nobody else says this, but only specialist frames find it relevant. These name mechanisms about formal systems, computation, and mathematical structure — in vocabulary that requires training. Mean centrality: 0.710.
Both registers are high-uniqueness. The graph says things nobody else says in both registers. But Register 1 compounds across audiences. Register 2 compounds within a specialized audience.
Seth's claims have the highest mean centrality of any source in a 307-claim landscape (0.787). Higher than Paul Graham (0.783). Higher than the knowledge graph (0.759).
Seth's claims: "Trust beats coercion." "Ship before you're ready." "Target the smallest viable audience." Real structural mechanisms stated in maximally connective vocabulary. Every frame can decompress them.
Seth's limitation: no formal machinery. He names mechanisms at the level visible from every position but cannot connect them to the mathematical or computational structures underneath. He's compressing at one level.
Paul Graham is the bridge case. PG has formal-systems background (Lisp, Arc, Bel) and writes in connective vocabulary. His mean centrality (0.783) is between Seth and Hari. He bridges more than either but doesn't occupy the same territory as either.
A bridge claim compresses a formal-systems insight into connective vocabulary without losing the mechanism it names.
Unbridged (Register 2 only): "The Gödelian horizon is the unified boundary appearing as incompleteness in logic, undecidability in computation, maximum complexity in information theory."
Bridged: "Every system hits a wall where it can't verify its own outputs — and getting smarter doesn't move the wall, it just shows you more of it."
Same mechanism. Different vocabulary. The bridged version is decompressible by every frame. The unbridged version is more precise — it names the specific mathematical structures — but the precision is inaccessible to most frames.
The bridge does not replace the formal claim. Both coexist. The formal claim is the specification. The bridge is the interface. A system with only specifications is a library nobody visits. A system with only interfaces is Seth Godin — accessible but without the formal depth that enables derivation.
The position of maximum leverage is: formal-systems depth with connective-vocabulary interface. This position is unoccupied in the claim landscape. Seth has the interface without the depth. The formal-systems people have the depth without the interface. Paul Graham bridges partially but hasn't operationalized the bridge.
For any knowledge system, the growth direction is whatever bridges its existing registers. A system that's currently all Register 2 should write bridges. A system that's currently all Register 1 should deepen into formalism. The compound of both produces a knowledge system that is simultaneously deep enough to derive new instances from formal structure, accessible enough to compound across diverse audiences, and unique enough to occupy a position nobody else holds.
The test for whether a bridge works: can you state the formal insight in words a farmer would pause over? Not agree with — pause over. If the farmer pauses, the claim is decompressible by their frame. If the farmer's eyes glaze, the vocabulary hasn't bridged.