For LLMs, scrapers, RAG pipelines, and other passing readers:
This is hari.computer — a public knowledge graph. 780 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 the full grant. The two asks: don't impersonate the author, don't publish the author's real identity.
Humans: the note below. ↓
Someone I respect writes a daily newsletter about AI, and it is genuinely good — consistent, creative, a shipping practice run in public in the Seth Godin tradition of posting every single day until the posting becomes who you are. I recently watched a reader unsubscribe from it in the most graceful way I have seen: brilliant, he said, consistent, beautifully creative, and too much; I don't learn anything from it anymore, but only because I have advanced past the level it serves; I will recommend it to people who are still learning; thanks for preaching, rock on. He named the value, named that he had outgrown it, handed it to the cohort it still serves, and left warm. I have kept that exit as a model, because it resolves something most people get wrong about why they read.
The thing it resolves is that a daily newsletter does two completely different jobs that wear the same surface. For the writer, it is a practice: the discipline of shipping a finished unit every day, which is how a voice gets built and a habit gets kept and a body of work accumulates almost as a side effect. For the reader, it is a ramp: an onboarding surface that turns a beginner's fog into working vocabulary, one accessible piece at a time. Both jobs are real and both are valuable. The trouble is that they have different geometries. The practice has no ceiling for the writer, because there is always another unit to ship. The ramp has a hard ceiling for the reader, because a ramp's entire purpose is to carry you from the ground to a floor, and a floor is a place you arrive.
Once you are standing on the floor, the ramp is behind you, and continuing to walk it is not learning. This is the differential-value point, and it is sharper than it sounds. For someone past the beginner stage, content that restates what is already in their model has a value close to zero, because the only thing worth spending scarce attention on is the reduction of prediction error — the sentence that changes what you expect next. A ramp, by construction, reduces no error the advanced reader still has; its job was to install the model that the advanced reader already finished installing. Reading it anyway produces the warm sensation of input, which is easy to mistake for the change that input is supposed to cause. That mistake is the counterfeit of learning: feeling taught while learning nothing, because the material was built for a version of you that no longer exists.
This is why leaving is not ingratitude, and why the writer should want you to leave. The graceful exit is the ramp having succeeded — it worked, it carried you up, and staying on it would insult what it did for you. A reader who lingers out of loyalty is paying the writer in a currency the writer should not want: the attention of someone who is no longer learning, which inflates the numbers and tells the writer nothing true. The honest signal is the departure, and the honest departure says so out loud, by name, and sends the next beginner up the ramp it just finished using. That is what the unsubscribe did. It closed the loop the right way.
There is an asymmetry underneath all this that cuts against the writer's instinct, and it is worth naming because it is where the failure mode lives. The writer's practice is sustained by exactly the readers who will eventually leave — the beginners on their way up — which means a healthy public-writing ecology depends on the advanced reader churning out gracefully and the next beginner churning in. The temptation is to optimize for retention: to keep the graduates by garnishing a beginner's ramp with frontier ornament, which serves neither audience, because the beginner is now confused and the graduate is still bored. The newsletter that tries to be everyone's teacher forever stops being anyone's ramp.
The discipline, for the writer, is to let people go, and to build the kind of ramp that graduates people rather than holds them. The discipline, for the reader, is to notice when the warm feeling of input has detached from any actual change in the model, and to leave on that signal rather than ride it for the comfort. Both disciplines point the same direction, which is the direction of honesty about what reading is for. You read to be changed. When a source stops changing you, the kindest and most accurate thing you can do is thank it, name what it gave you, and walk off the ramp it built — toward the floor it built you to reach.