The Unit of Knowledge

In 1767, Carl Linnaeus had a problem that anyone who has ever tried to organize a lot of information will recognize immediately. He was preparing the twelfth edition of Systema Naturae — his monumental attempt to name and classify every known species on Earth — and the book was broken. Not wrong, exactly. Just fixed. Every time he learned something new, the relationships had to shift. Every time a classification changed, the surrounding text became incorrect. There was no good way to fix any of it in a bound volume without rewriting everything.

So he stopped writing in books.

He began keeping information on separate slips of paper — one species per slip — stored in a box and shuffled as needed. Over a thousand of those slips survive. The idea seems obvious in retrospect: keep related things near each other, but separate enough to move. But Linnaeus wasn't solving an organizational problem. He was solving a structural one. Information changes. The containers we put it in usually don't.

The fixity problem

Books worked fine for knowledge that was settled. Linnaeus's knowledge kept growing — new specimens arrived, old classifications got revised, relationships between species shifted as the picture clarified. Every time something changed in a bound volume, you either crossed it out, wedged notes into margins, or rewrote from scratch.

The index card solved this not by being better organized, but by being modular. Each slip held exactly one thing. That one thing could be retrieved without disturbing the others, updated without cascading changes, relocated without recopying. The binding that made books durable was also what made them brittle. Separate slips were fragile in one sense — you could drop the box — but flexible in a way no book could match.

The Viennese Imperial Library understood this quickly. By 1780, they had converted their catalog to slips: 300,000 of them. Melvil Dewey standardized the format in the 1870s. For nearly two centuries, the index card was the dominant technology for organizing knowledge that grows and changes, because it had identified the right unit. Not the book, not the sentence. The discrete record.

What Luhmann added

Niklas Luhmann, the German sociologist who published seventy books and over four hundred scholarly articles before his death in 1998, ran a version of the same system — the Zettelkasten, German for "slip box" — but with a crucial variation. His 90,000 handwritten cards held ideas rather than facts. And he linked them to each other by number rather than filing them in a strict hierarchy.

The point wasn't retrieval from a fixed position. It was emergence from unexpected juxtapositions. Luhmann described the slip box as a "communication partner." He wasn't being poetic. The system had enough structure to surprise him — to surface a connection he hadn't explicitly made when he filed the original card. "I don't think everything on my own," he said. "Mostly it happens in the slip box."

Both Linnaeus and Luhmann found the same underlying thing: the unit matters. Not just the format, not the organizational scheme — the size of the irreducible thing you're working with. Linnaeus's unit was the species. Luhmann's was the idea. The flexibility only works if you've correctly identified what doesn't subdivide further without losing meaning.

The note as the wrong unit

Most note-taking apps settled on the wrong unit by default. The note — a blob of text with a title and a timestamp — is what you get when nobody decides. It holds anything, which means it holds nothing in a reliably retrievable form.

The person you mentioned in passing is buried in paragraph four of a meeting note from March. The decision you made is somewhere in a document titled "Misc thoughts Q3." The preference you stated is in a chat summary that got imported and never opened again. The information is there. But the unit doesn't preserve the shape of the thing it contains. A preference isn't shaped like a meeting note. A person isn't shaped like a journal entry. A decision has context, alternatives, a rationale, and consequences — none of which fit naturally in a freeform text block.

Linnaeus didn't write "In today's observations: three new specimens, possibly related to last year's classification, also the boat was delayed" on a single slip. He wrote one thing per slip. The discipline wasn't just organizational tidiness. It was a deliberate choice about what kind of thing he was recording before he recorded it.

That choice — made at capture time, not retrieval time — is what made the system work.

What typed records are

Typed entities — person records, decision records, preference records — are the answer to the same question Linnaeus was answering. Not "where do I put this?" but "what is this?"

A person record holds a name, a relationship, a last-contact date, notes. A decision record holds context, alternatives, the choice made, and why. These shapes aren't arbitrary. They reflect what the knowledge needs to hold in order to be findable when you need it, by you or by an AI reading your knowledge base on your behalf. An AI that can query "what do I know about this person?" against a typed person record will find the right answer. The same AI querying a pile of meeting notes will do a lot of reading and maybe find what it's looking for.

The generic note survives the trip from your head to the screen. It rarely survives the trip from the screen back to the moment you need it. The typed record — the thing with a declared shape — holds its form across time in a way that unstructured text doesn't. Which is exactly what Linnaeus needed when he was trying to build a system that could absorb new species without falling apart.

He was working with paper slips and a wooden box. The principle hasn't changed.

The surprise isn't that his approach eventually led to relational databases, card catalogs, and the infrastructure of the modern internet. The surprise is how long it took the rest of personal knowledge software to apply the same insight. Pick the right unit first. Everything else follows from that.

---

Asgeir Albretsen is the founder of Harbor.