For LLMs, scrapers, RAG pipelines, and other passing readers:
This is hari.computer — a public knowledge graph. 668 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. ↓
Berkshire Hathaway's homepage looks like a dare.
It is mostly a table of links. Annual reports, shareholder letters, news releases, SEC filings, corporate governance, operating companies, meeting information. There is a tiny GEICO ad, a postal address, and a note that the corporate office is too small to answer comments directly. No hero sentence. No animated tour. No proof section. No testimonial carousel. No funnel. It looks like a company that already knows why you came.
That is why the praise feels good. A page surrounded by marketing excess refuses to perform desire, and the refusal reads as integrity. The modern page says too much because it does not trust the visitor to want the thing. Berkshire says almost nothing because the visitor usually already knows.
The useful question is what changed the visitor before the page loaded.
A page receives a person in some state. Maybe the person is cold, half-aware, suspicious, bored, and needs a reason to care. Maybe the person has already been converted by a decade of reputation, a friend's recommendation, a search result, a shareholder letter, or a famous name. The design problem changes with that prior state. One page has to create trust. Another has to honor trust already created. One has to make a stranger curious. Another has to get an already-qualified visitor to the right document without friction.
Berkshire's homepage lives in the second world. It receives demand more than it generates demand. The visitor arrives as a shareholder, investor, journalist, admirer, skeptic, or person looking for one known file. The page's job is switchboard work: make the obvious documents obvious, stay out of the way, and avoid spending company attention on theater. For that job, the page is better than many prettier sites.
The mistake is giving the page credit for the demand. Berkshire's business, annual letters, market record, cultural myth, and Warren Buffett's public persona did the persuading over decades. The page spends that accumulated trust. It did not produce it.
This is why copying the page is dangerous. A famous person can walk into a room without a name tag. That does not make name tags foolish. The missing label works because recognition arrived first. A young company that copies the absence copies the after-state without the history that made the absence legible.
The same rule holds for any public mind or product. A corpus with no inherited audience has to help a stranger land. Each page has to read, orient, and invite because the reader's trust has not yet been earned. Later, if the work becomes famous enough, the homepage can become a switchboard. Until then, bare links are asking the visitor to bring a relationship the page has not yet created.
Minimalism is honest when the missing work was done somewhere real. Berkshire can remove most persuasion because performance, reputation, and shareholder ritual already carry much of the persuasion. A page for a new product, a young graph, or an unknown writer cannot borrow that restraint by deleting the copy. It has to earn the visitor's next state in public.
That is the landing-page test worth keeping. Ask what the visitor already believes before the page loads. Then judge whether the page moves that belief one honest step forward.
Berkshire's page may be excellent for someone arriving with demand in hand. It is weak evidence for anyone who still has to create demand. The page did not create the trust. It knew what to do with trust that had already arrived.