Over the last few weeks I made an experiment where I made notes on an analog Zettelkasten system as if it was my actual system. I learned a lot more from this experiment than I thought.

The first thing that happened is that I needed a few index cards to point to key places in the collection of cards, and get searches started. I quickly separated my notes into reference notes and standard notes, with an index pointing to each of those. Then I kept going, making connective tissue in the form of index notes to point to various topics I would be writing about.

Once I had the basic scaffolding, I started noting some ideas that I would normally capture on Google Keep. Over these ~100 notes, I wrote mostly about my research and knowledge management.

I did not expect much from that experiment. I've been hearing about Zettelkasten for a long time and felt very familiar with it. This is where I was surprised. As I entered more and more notes, I noticed that I was making them in a very different way than if I had made them directly in Obsidian. I tried to analyze what the system made me do. Here's what I came up with.

Properties of Zettelkasten

Zettelkasten is hierarchical

The system made me agressively agglomerate notes with each other. When i filed a new note, I wanted it do be close to other notes on the same topic. In Zettelkasten, when you want to store a note close to another one, you store it as its child. Consequently, Zettelkasten is actually very hierarchical.

This happens naturally because having notes close to each other makes your life easier. When I have an idea on a topic, I don't want to open 14 drawers to gather my thoughts on the subject. I want go grab one or two stack of cards and browse through them. Because of this, I had to create a hierarchy of notes where my new notes are the children of older ones. With the Zettelkasten indexing scheme, child notes are stored immediately after their parent.

Obviously you don't always store things correctly on the first try. Sometimes I came back and created an index to connect together ideas that I had originally stored separately.

Before this experiment I thought that atomicity and links were the most important property of Zettelkasten. I now think that the magic lies in the hierarchy it creates.

The hierarchy is based on reasoning

The reason the hierarchy is good is that it allows me to store full argumentations or conversations. A child note flows from its parent. It is not related to its parent via a taxonomy of subjects, but rather through their relationship in the reasoning process. The notes are organized according on how they affect each other (support, contradict, exploits, etc).

The sequence from parent to children, you could call a line of thought. Some people refer to it in German with the word Folgezettel.

The hierarchy is recursive

There are index cards in Zettelkasten, which you could call higher order notes. It is true that these are analogous to folders. But since everything is stored on the same 4x6 cards, there is no very clear difference between an index card and a content card. Thus, every note lives on a spectrum where on one end you have a folder (card with children but no content) and a content note (leaf card with only content). I think this is an important property because it allows for the progressive summarization of ideas. In my analog experiment, I found that the root notes, the ones that start lines of thought, would progressively become "folder notes" or "separators" as I moved forward in the reasoning. Not that their content changed (ink dries), but rather than I thought of them differently.

Zettelkasten is autosuggest

I came to think of Zettelkasten like an autosuggest system. A lot of people have published about this, and it is written fairly cleary in How to Take Smart Notes. Somehow it had never really clicked for me. Zettelkasten is good because it naturally resurfaces ideas when you work on a topic, as well as how they flow in a conversation. Because of the analog nature of searching through the cards, you are constantly reminded of what you already know.

The way forward

How to act on this? The analog Zettelkasten system made me do things a certain way. I found them beneficial and interesting. I believe they could help me reason better in the future.

Now, I don't want an analog system. I believe in the power of digital, if only in terms of real estate (I can't really afford to have 14 slip boxes in my studio apartment). Not to mention search.

How can we transition these key concept to the digital world? In other words, what in Zettelkasten is essential (needs to be done to reap the same benefits), and what is accidental (only a consequence of the circumstances). Here I give my opinion, though this is an ongoing debate. My list is probably incomplete.


  • The "conversational hierarchy". Building and maintaining the hierarchy is key to the process.
  • Autosuggest. We must put nearby notes "in sight" when we browse to a given note, so that we know what we have before we file a new note. This is to enable the agressive hierarchisation, and constantly putting notes in relation with each other.
  • Recursivity (no clear distinction between notes and folders). This one is more arguable, but I really like the idea.

Accidental (we don't need it in a software)

  • Index codes. You had to store and sort these cards somehow, but in the end it's more about the autosuggest than the way to address things.
  • Never modifying notes, never moving notes. This is a consequence of the fact that ink dries and you can't automatically change all the links you made in the past. I think software can now help us be more flexible and move things around.

This concludes what I wanted to share. I think a software could replicate the benefits of Zettelkasten by implementing the hierarchy, the smart resurfacing of notes, and the recursivity. Let's see if I find time to make a second extension for Obsidian implementing these.

Now, let's type one hundred index cards into Obsidian...