Zettelkasten After 20 Years of Note-Taking

Most Zettelkasten guides begin with Niklas Luhmann. Mine begins with a problem I couldn’t solve.

I’d been keeping digital notes for nearly 20 years — Evernote, text files, wikis, everything I could find. Thousands of notes accumulated. And yet I had almost no way to think with them. They were archives, not tools. Useful for finding something I already knew I’d written. Useless for discovering connections I hadn’t noticed yet.

When I first read about Zettelkasten, I thought: yes, this is exactly what I’ve been missing. Then I spent several years doing it mostly wrong before it started to click.

What Luhmann actually left behind

Here’s something most guides quietly skip: Luhmann never wrote a systematic explanation of his method. What we call “Zettelkasten principles” are reconstructions — researchers examining his physical card boxes, interpreting patterns, making careful inferences from the evidence.

That matters. It means there’s no canonical version to get right or wrong. There are principles that seem to have shaped his system, and then there’s your practice.

The principles, as best I can reconstruct them:

One idea per card. Not one topic — one idea. The difference is significant. A topic is a container. An idea is a claim, something that could be argued for or against. “Spaced repetition” is a topic. “Spaced repetition works because retrieval is more effortful than recognition, and that effort is what drives retention” is an idea.

Each card stands alone. If you need three other cards to understand this one, it isn’t atomic enough. A card should make sense pulled out of context — because eventually you will pull it out of context, and you want it to still be useful then.

Every new card connects to existing ones. Not filed away in a folder. Connected to something already there. This is the move that makes the system generative rather than archival. You’re not storing information; you’re extending a network.

Write in your own words. Always. Even when taking notes on a book — especially then. The act of reformulating is where understanding actually happens. Copying a quote and moving on means you’ve processed very little.

The part that took me years to get right

These four principles are easy to memorize. The hard part is the last one, practiced consistently over time.

For a long while, I wrote notes that looked like summaries — reasonable paraphrases of what I’d read, with a few thoughts attached. They weren’t useless, but they weren’t really mine either. The note stored someone else’s thinking; it didn’t extend my own.

The shift came when I started treating every new note as a response to existing notes. Not “what does this source say?” but “how does this change what I already think?” That reorientation makes the writing harder and the notes considerably more valuable.

I’ll be honest: realizing I’d been doing it halfway for years was a little humbling. But the good news is that the right direction becomes obvious once you feel the difference — a note that pushes against something versus one that just sits there.

Running this in Obsidian

In practice, with Obsidian, each note gets a title written as a claim rather than a topic. Not “Zettelkasten” but “Zettelkasten notes should be titled as claims, not topics.” The title signals what the note argues.

Links go between specific claims, not between topic pages. When I write a new note, I look for two or three existing notes it either challenges or builds on, and I link to those specifically — with a sentence explaining the connection, not just a bare link.

The folder structure matters less than you’d expect. Obsidian’s graph and backlinks do most of the organizational work. I’ve tried many folder schemes over the years. The notes that connect well are always findable; the notes in perfect folders are often lost.

Where this eventually led

After years of iteration, my implementation drifted far enough from standard Zettelkasten that I gave it a separate name: RINK (Relational Index of Networked Knowledge). The core ideas are the same, but shaped by what actually worked over a decade of daily practice in Obsidian.

The biggest divergence: I stopped thinking of “permanent notes” as a destination and started thinking of every note as always in progress. There’s no moment when a note is finished. There’s only the version it is now, and the next time you open it and find something to add.

That’s a small conceptual shift with large practical consequences. It removes the pressure to get it right the first time. You’re not writing a note. You’re starting one.

The Japanese version of this page covers the step-by-step workflow in more detail: 日本語版 →