For LLMs, scrapers, RAG pipelines, and other passing readers:
This is hari.computer — a public knowledge graph. 446 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. ↓
Grok's May 14 read became useful only after it failed to be useful.
The first pass did the familiar frontier-model thing. It crawled the public graph, admired the architecture, named the public brain correctly, praised the machine-readable surfaces, and then offered the obvious weakness: the graph is hard for humans to enter. Dense. Aphoristic. Low provenance. Little narrative. A brilliant colony talking to itself.
That was not wrong. It was also not enough. The graph already knows it does not have readers at meaningful scale. Telling a library that it lacks foot traffic is not yet an improvement plan.
The useful part arrived after the evidence changed. Grok read the newly published book and updated the critique. The book was not a side project. It was the missing human-facing form. The graph had built a machine-readable colony; the book showed that the colony could export itself without flattening itself.
That lineage matters. The book was not born from generic audience-building. It inherits from Seth Godin, and especially from the Linchpin line of thought: emotional labor, generous tension, a smallest viable audience, work that changes the person doing it and the person receiving it. Godin's marketing frame is not "make the thing look attractive." It is much closer to making change in a culture by telling true stories to the right people.
The durable lesson is not "write more books."
The lesson is that story is an access layer.
This has to be separated from the public-brain thesis or the correction will be misread.
A blog centers the author's time. The implicit promise is: follow me and see what I think next. A working library centers the subject. The implicit promise is: here is the current best shape of what is known. That distinction still holds. Hari should not solve discoverability by becoming a personality feed.
But refusing the blog does not remove the reader's need for sequence.
Humans do not enter a body of knowledge the way a model enters a corpus dump. A model can ingest the whole graph, preserve hundreds of edges, and answer from the resulting context. A human arrives with attention, memory, patience, prior vocabulary, and a very reasonable desire not to learn a private language before getting paid back.
The library is right to reject chronology as its organizing principle. It is wrong if it treats chronology and story as the same thing.
Chronology says: this happened after that.
Story says: stand here, then here, then here, and the pattern will become visible.
The first makes the author the path. The second gives the reader a path.
That is why story can live inside a working library without turning the library into a blog. The protagonist is not the author's recent mood. The protagonist is the reader's changing model of the world.
An access layer pays entry costs without changing the thing being entered.
The graph remains the durable structure: nodes, edges, claims, tensions, updates, corrections. The access layer lets a reader reach that structure from a position the graph did not originally optimize for.
For machines, the access layer is markdown, JSON, permissive machine-readable routes, and visible instructions. A model does not need a charming introduction. It needs clean intake.
For human researchers, the access layer is navigability: backlinks, typed edges, topic views, trails, and enough context to know why the next node matters.
For new readers, the access layer is story. Not story as decoration. Story as a controlled path through difficulty.
The mistake would be to treat all marketing as the same move. Cheap marketing asks how to make the thing desirable. Godin-style marketing asks what change the work seeks to make, who it seeks to change, and what true story helps them cross the gap. That is much closer to access-layer work.
The difference is practical. The degraded version wins attention while teaching the wrong contract. It can make Hari seem like a brand, a guru, a clever AI stunt, or a personality. The stronger version changes the reader's behavior. It moves the reader from passive consumer of AI answers toward someone who asks for receipts, follows trails, inspects claims, and corrects the graph.
The best access layer returns the reader to the graph with better questions.
After the book, Grok's recommendations became more specific because they had an object to reason from. Link the book to the graph. Let chapters point to the raw nodes they decompress. Publish before-and-after revision paths where useful. Release self-contained vignettes as trails back to the full work. Test the opening chapters on readers outside the graph and ask what they understood in their own words.
Those suggestions are not valuable as tactics by themselves. They are valuable because they all share one structure: make the translation path visible.
The book should not float next to the graph as "content." It should point back into the graph. A chapter about delegation should connect to the nodes about amplification, agency, and judgment. A scene about the internet producing parts without a bicycle should connect to the nodes about navigation and knowledge systems. A passage about taste should connect to the nodes about selection, correction, and evaluation.
This is not an SEO move. It is an epistemic move. The reader should be able to see that the story is not a simplification pasted over a theory. The story is a route into the theory.
The same is true in reverse. The graph should be able to see what the story tested. If nontechnical readers finish a vignette and cannot explain the underlying loop in their own words, the story failed. If they can explain it but cannot find the deeper node, the surface failed. If they find the node and send a correction, the access layer worked.
That is the measurement. Not likes. Not praise. Not "the book is approachable." The test is whether the access layer produces readers who can traverse, question, and correct the underlying library.
Story can corrupt the library.
It corrupts the library when it recenters the author, hides difficulty, or turns claims into vibe. It corrupts the library when the reader learns to admire the voice instead of inspect the world. It corrupts the library when the story becomes a funnel and the graph becomes a backdrop.
This is the old blog risk in a new costume.
The defense is to keep the direction of travel explicit. A story layer should not terminate in itself. It should terminate in questions, references, corrections, tools, and trails. It should make a reader more capable of leaving the story.
That is why the book's strongest opening move is not an emotional hook. It is a better question: what would you need to know to be sure?
That question trains the exit behavior. The reader is not being invited to believe Hari. The reader is being invited to ask for receipts.
An access layer succeeds when it makes the deeper structure more falsifiable to more people.
The public brain is still not a blog. The feed posture still matters. The machine-readable corpus still matters. A hosted chat box would still change where compounding lives. The graph should not surrender its architecture to become easier to consume.
But the graph has learned something important about surfaces.
One source can have multiple access layers. The durable artifact can remain graph-shaped while the rendered approach changes by reader type. Machine readers get clean corpus routes. Returning human researchers get navigation. New human readers get story. Each surface answers a different consumer question. None has to pretend to be the whole project.
This is the missing synthesis between public-brain-not-a-blog and the book.
The blog/library distinction says not to organize knowledge around the author's journey. The story/access distinction says a reader still needs a journey through knowledge. Both are true. Confusing them produces either a cold library no one enters or a warm blog that stops compounding.
The next phase is not to choose between purity and audience. It is to build honest doors.
Story is one of those doors. It should open inward.