Tag: reading

I want to start with the problem, because most Zettelkasten guides skip it.

You read something genuinely useful. You highlight it, maybe jot a note in the margin, and move on. Three months later, you’re trying to connect that idea to something you’re writing, and you cannot for the life of you remember where you read it, what exactly it said, or how it fit into whatever you were thinking at the time. The idea is gone. Not because you’re not smart enough to remember it. Because that’s not what human memory is for.

Our brains were built to make connections between things, not to be filing cabinets. The filing cabinet instinct — highlight it, dump it in Evernote, never think about it again — is exactly backward. You’re outsourcing the thinking part and keeping the forgetting part.

The Zettelkasten method fixes this. I’ve now been running my system for several years, first built it seriously when my doctoral reading volume became genuinely overwhelming, and I’m heading into year four of the dissertation — near the finish line — with a system that has become part of how I work and think across every domain of my life, not just academic writing. The dissertation is almost done. The Zettelkasten is permanent.

Here’s what I’ve learned.

What the Zettelkasten Actually Is

The words are German: Zettel means “slip of paper,” and Kasten means “box.” Slip box. That’s the whole thing — Niklas Luhmann, a German sociologist writing in the mid-twentieth century, kept a box of index cards where he recorded his ideas, one per card, linked to other cards through a numbering system he developed himself.

He published 70 books and over 400 scholarly articles. He credited the slip box. Not his intelligence, not his work ethic — the system.

What made Luhmann’s approach different from just keeping notes wasn’t the index cards. It was that the cards talked to each other. Each note referenced other notes. Ideas linked to ideas. Over time, the box developed what Luhmann called a “conversation partner” — a second mind that could surprise him with connections he hadn’t consciously made.

That’s the thing most people miss when they first hear about Zettelkasten. It’s not an organizational system. It’s a thinking system. The goal isn’t to store information — it’s to generate new ideas by forcing your notes into relationships with each other.

How I Got Here

My first encounter with anything like this was reading about how Ryan Holiday writes his books. He uses a notecard system — one idea per card, physically sorted into categories, pulled out when he’s writing. It’s not quite Zettelkasten, but it’s the same instinct: single ideas, physically handled, connected by the writer’s judgment rather than a folder hierarchy.

When I started my doctoral program and the reading volume became genuinely overwhelming — dozens of articles a week, books on top of books, sources I knew I’d need to cite but couldn’t reliably locate again — I needed something more systematic.

The Zettelkasten method, as popularized by Sönke Ahrens in How to Take Smart Notes, is what I landed on. Ahrens’s book is still the best entry point if you want to understand the theory before building the practice.

The Three Types of Notes That Actually Matter

Most Zettelkasten guides give you five or six note types and immediately make the whole thing feel complicated. In practice, I work with three:

Fleeting notes are the raw capture. Something I heard, read, or thought that seems worth keeping. No polish required. I write these in my Field Notes notebook — the one that’s always in my back pocket — with a date stamp and whatever I can get down in thirty seconds. They’re temporary. Their only job is to get the idea out of my head before I lose it.

Literature notes are what I write after sitting with a source. When I finish a book or article that matters, I go through my fleeting notes and highlights and write one note per idea — not a summary of the chapter, not a quote, but what I think this means and why it matters. In my own words. This is where the thinking starts.

Permanent notes are the keepers. These are the ideas that survive the literature note stage and earn a place in the main system. Each one stands alone — a complete thought that makes sense without context. Each one links to other permanent notes where the connection is real, not just topical.

The discipline is: no permanent note without a connection. If a new note can’t be linked to anything you already have, either the note isn’t ready yet, or you’re missing a bridge note that should exist.

My Actual Setup: Cards Plus Notion

I run a hybrid system. The physical component is 4×6 ruled notecards — the sweet spot for a single idea with enough space to actually develop it. I use a date stamp to record when a card entered the system. I write with Blackwing pencils because the erasability matters when you’re still working out what a note should say.

The cards live in a card box on my desk, organized into loose topic clusters that shift as the system grows. I don’t use a strict numbering system — I’ve found that topical clusters with cross-references work better for my brain than pure alphanumeric sequences.

The digital component is Notion. When a permanent note is fully formed, it gets entered into Notion with tags, links to related notes, and a reference to the source. This is where the search capability becomes essential — finding a note about something I read eighteen months ago takes seconds.

The hybrid approach sounds redundant, but it isn’t. Writing by hand forces slower, more deliberate thought. The physical card is where I work out what I actually think. Notion is where I store it and connect it at scale.

What This Looks Like in Practice

Here’s a concrete example from my dissertation work.

I read an article about distributed cognition — the idea that human thinking isn’t just what happens inside our skulls but includes the tools and environments we think with. I write a fleeting note while reading: distributed cognition — thinking happens in the system, not just the thinker. Interesting connection to why PKM matters?

Later, I write a literature note: Hutchins (1995) argues that cognition is distributed across people, artifacts, and the environment. Navigation example: the ship’s navigation system is the unit of cognition, not any individual sailor.

That becomes a permanent note: Tools are not just extensions of thinking — they are part of thinking. A well-designed external system (like a Zettelkasten) is literally part of the cognitive process, not a substitute for it. Linked to: notes on Vygotsky’s zone of proximal development, notes on embodied cognition, notes on why writing clarifies thinking.

Months later, I’m working on a section about student-centered learning environments. I pull the distributed cognition note. It connects, in ways I didn’t plan, to three other notes I’d written about classroom design and technology integration. The Zettelkasten hands me an argument I didn’t know I was building.

That’s the thing. It surprises you.

The Next Frontier: Obsidian + Claude Code

I’m going to be honest: I’m still experimenting with this, so take it as a field report rather than a recommendation. But it’s too interesting not to share.

Andrej Karpathy — co-founder of OpenAI, former Tesla AI director, one of the clearest thinkers working in AI today — recently published a pattern he calls the LLM Wiki. The idea is deceptively simple: instead of keeping your notes in a system that you navigate manually, you keep them as structured plain-text markdown files, and you point an LLM directly at that folder to find connections, synthesize ideas, and build new understanding across everything you’ve written.

Karpathy’s framing is sharp: the shift is from retrieval to compilation. Traditional search asks “which note answers this query?” The LLM wiki asks “build and maintain a persistent, cross-referenced knowledge base that already contains the synthesized answer.” The AI doesn’t just search your notes — over time, it helps write and maintain them, surfacing connections you didn’t consciously make.

What makes this particularly interesting for Zettelkasten practice is that it doesn’t replace the method — it extends it. The atomic note principle, the linking discipline, the permanent note as a self-contained idea: all of that still applies and in fact becomes more powerful when an LLM can read the whole vault and identify non-obvious connections across it.

The workflow I’m exploring: Obsidian as the front end (free, local files, excellent graph visualization of note connections), Claude Code as the intelligence layer pointed at the vault. You give Claude Code a schema file that tells it what the wiki is for and how it’s structured, then feed it sources — articles, book notes, research papers, your own existing notes — and it builds and maintains the wiki, linking ideas across everything you’ve given it.

Karpathy himself manages wikis of over 100 articles this way. The graph view in Obsidian, showing every connection between notes visually, is something you have to see to understand — it’s a map of how your ideas actually relate to each other, not how you filed them.

I’m at the stage of migrating some of my Notion notes into an Obsidian vault and running Claude Code against it to see what connections it surfaces that I haven’t made manually. Early results are genuinely surprising in the way the best Zettelkasten surprises are — the system finding threads you didn’t know you were pulling.

If you want to explore this yourself, Karpathy’s gist is at github.com/karpathy, and there are now several good community implementations. Start small — one topic domain, a handful of sources — and see what happens.

The Biggest Mistakes People Make

Highlighting is not note-taking. A highlight is a bookmark. It says “I thought this was interesting” and nothing else. Unless you return to it and write what you think it means, it’s not knowledge — it’s a marker in a document you’ll probably never reopen.

Too many categories too early. The instinct to organize before you have enough material always produces a structure that fights the content. Let the connections emerge from the notes themselves. Restructure when the natural clusters become clear.

Skipping the rewrite. Writing a literature note in your own words — not copying the quote, not paraphrasing loosely, but actually reconstructing the idea from scratch — is where the learning happens. It’s uncomfortable because it forces you to distinguish between what you actually understood and what you just recognized.

Treating it as a productivity system. The Zettelkasten is slow. A well-formed permanent note might take twenty minutes to write. It will pay for that time a hundred times over when you need it, but if you’re measuring output by notes per hour, you’re measuring the wrong thing.

Abandoning it when life gets busy. The system only has value if it has continuity. Even one card a week keeps it alive. The temptation during high-pressure periods is to stop feeding the system — exactly when you need it most.

The Tools

Physical:

• 4×6 ruled index cards — sized right for a single developed idea
• Field Notes notebooks — for fleeting notes and daily capture
• Date stamp — stamp every card when it enters the system
• A card box — simple and adequate

Digital:

• Obsidian — free, local files, excellent bi-directional linking, increasingly my recommendation for anyone starting fresh. The graph view alone is worth it.
• Notion — what I’ve used for years and still use; better for combining notes with project management
• Claude Code — for the Karpathy LLM Wiki pattern; points directly at your local Obsidian vault

Books:

• How to Take Smart Notes by Sönke Ahrens — the essential text. Read this before you build anything.
• Building a Second Brain by Tiago Forte — broader personal knowledge management; complementary rather than competing
• Digital Zettelkasten by David Kadavy — practical and short; good if you want to get started digitally without reading the full Ahrens

Is It Worth It?

I’m four years into a doctoral program — near the finish line — and the Zettelkasten is the primary reason I’m not drowning. The reading has been relentless, and the connections between sources are what the work lives on. Without a system that forces me to make those connections explicit and retrievable, I’d be starting from zero every time I sat down to write.

But here’s the thing I’ve come to understand about this system: it was never just a dissertation tool. The notes I’ve written about instructional coaching, about technology in education, about how people actually learn — those connect across my classroom work, my doctoral work, my writing, my thinking in every direction. The system doesn’t belong to a project. It belongs to the thinker.

The Karpathy LLM Wiki pattern is the next chapter of that idea. If the Zettelkasten is a conversation partner you build note by note, an LLM pointed at your vault is something like that partner getting a significant intelligence upgrade. I’m genuinely excited to see where it goes.

Start small. Write one permanent note today about something you read this week. Not a summary — what you think it means. Link it to one thing you already know.

That’s the whole thing. Do it again tomorrow.

The tools I use for my Zettelkasten — notecards, date stamp, Blackwing pencils, Field Notes, and more — live on my Favorite Gear page. If you want to see how the Field Notes fit into daily planning, that post goes deeper on the daily capture side of this system.

The Eclectic Educator is a free resource for everyone passionate about education and creativity. If you enjoy the content and want to support the newsletter, consider becoming a paid subscriber. Your support helps keep the insights and inspiration coming!