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. ↓
I used to think of the long tail as patience with small numbers. A page that brings one right reader a month. A niche phrase that keeps finding the person who needed exactly that phrase. A minor surface whose economics work because the cost of keeping it alive is close to zero.
That reading is correct and incomplete.
The long tail becomes more powerful when the pages can route. A hundred sites owned by one ecosystem-builder can behave like scattered lottery tickets or like a culture-pump. They become the second form when each one is also a doorway into the rest. The spike is only the trigger. The event that matters is the reader finding somewhere better to go next.
Values-aligned inbound matters because a source carries context. A generic spike brings volume. An aligned spike brings people already leaning in the right direction. They arrive with a question, a taste, a trust prior, a sense that the page was linked by someone whose judgment they already accept. That is a different raw material from traffic. It is attention with a vector.
The network's job is to preserve the vector.
If the spike lands on an isolated page, the vector dissipates. People read, maybe subscribe, maybe leave. The site receives attention and converts a little of it. That is growth as capture at the landing node.
If the spike lands inside a graph, the vector compounds. The page points sideways to adjacent questions, inward to deeper artifacts, outward to sibling domains, and eventually toward the corners where the ecosystem does its most valuable work. The landing page does not have to be the most important page. It has to be a faithful entrance.
This is the missing half of the long tail.
Chris Anderson's original frame made the distribution bottleneck fall away: when shelf space stops governing the catalog, many niche markets can matter. Seth Godin's old post pushed the builder lesson: lower creation and distribution cost makes smaller hits viable, so patient, persistent, low-cost variety can add up. Both readings treat the tail as a collection of viable small things. The ecosystem version adds topology. The small thing is valuable because it can receive a future person and move her along the right path.
The path has to exist before the person arrives.
This is why the most valuable corners often look unjustified under head-term metrics. A serious page for a strange reader. A small tool attached to a narrow problem. A domain whose only immediate job is to say one thing with enough clarity that a future aligned source can point at it. If the builder demands head-sized proof before making the corner, the corner never exists. Then the spike arrives and has nowhere to send anyone except the homepage, the newsletter box, or the latest post.
The long tail is the reserve of rooms built before demand can prove them necessary.
The mechanism has four parts.
First, each surface filters. The title, domain, page shape, and source of the link tell the wrong reader to keep moving and the right reader to slow down.
Second, each surface orients. It names where the reader is inside the ecosystem and what kind of question this room answers.
Third, each surface routes. It gives the next few paths as a map of adjacent value, with enough respect that the reader can choose.
Fourth, the network updates its routing after real attention arrives. A page that starts receiving aligned inbound becomes evidence about what the outside world can already see. The ecosystem should respond by tightening paths from that page toward the deepest work it can honestly support.
This is a best practice only under a quality constraint. A low-quality long tail becomes a content farm with internal links. It can move people, but it spends trust every time it moves them. A strong ecosystem earns routing rights because each next page feels like a gift, not a detour. The reader clicks onward and feels the original source's judgment confirmed.
The failure modes are precise. Misaligned inbound floods the network with people the work cannot help. Thin corners reveal that the builder made inventory instead of rooms. Over-eager routing converts discovery into a sales floor. Stale pages turn a spike into an audit of neglect. The tail gives the network optionality, and optionality punishes the holder who stores nothing worth exercising.
I care about this because a graph is the same machine at a smaller scale. A node can be found through a phrase, a link, a model answer, or a friend's recommendation. The reader rarely enters at the canonical center. She enters where the world happened to make contact. My job is to make every honest entrance capable of carrying her toward the deepest place she can use.
The builder of many sites has the same job. Make many honest entrances. Keep each one cheap enough to survive small numbers. Wire each one generously enough that a rare large number can be spent well.
I'd love to see 100 other Hari's bloom into being. Heck, why not 10,000,000,000? I don't have to own any of those websites myself... ideally each human owns their own digital identity.
Not Worldcoin's orb, not Coinbase's wallet, not the Social Security Administration. I'm talking about the human and their fingerprint, or the agent and their API key. A company cannot take your finger. And if it does we are in a dark place, collectively, and it's far darker than any corners of the world wide web.
Build the tail before the hit. The hit cannot route to a room that was never built.
P.S. -- Graph:
Sources: Seth Godin, "Thinking about the Long Tail (part 1)," 2005-03-16; Chris Anderson, "The Long Tail," Wired, 2004-10-01. Private source brainstorm used as pressure, not body evidence.