Zettelkasten Forum

edited June 2020

A Zettelkasten is a personal tool for thinking and writing that creates an interconnected web of thought. Its emphasis is on connection and not mere collection of ideas.

• I maintained the largest collection of dishes served by restaurants for more than 10 years. It generates a few hundred million dollars annually in revenue to is owner. The dishes recipe were captured by different people via Evernote.

70% of the metadata was captured via tags, manually added or automatically extracted from plain text. Our clients notified us that some of the entries were of poor quality: for instance some dishes were tagged with "vegan" and "beef" at the same time.
We organized a few tags into "categories" - aka list of mutually exclusive tags.
We also added "taxonomies", eg "mayo" implies "egg" and "oil".
Introducing some constraints on tags did limit what was possible to express but also improved the experience of our clients (and revenue).
I wanted to share this example to avoid the dogma of a strong position against categories. They are another tool in the toolbox and are, in my opinion, only a more specialized set of tags for specific occasion.

• I'll make the libertarian argument here: Everybody has the right to mess up his or her own life.

If you are dealing with anonymous people you surely want to force quality. But you can have a little more trust in yourself than in strangers.

I am a Zettler

• Hi, i'm pretty new to the zettelkasten method, which i think is one of the powerful method i heard about. Your posts describing the process, and tips are awesome, clear, precise and understandable. Points adopts good structure and are easy to follow. Though there are some links that i think are broken, e.g: in this article, the link under the word "assimilation" early in the page leads to a japanese article about real estate ^^ (which is not really a domain i'm interested in). Thanks for your contributions and the time and efforts you put into those materials.

• Fixed the links! 3/6 were broken, geez. Thanks!

Author at Zettelkasten.de • https://christiantietze.de/

• In my work, back at the beginning of the digital age, when computers were mainly for writing and crunching numbers in spreadsheets, I had a very effective analog filing system.

I had a four drawer lateral filing cabinet. I had two very broad categories of files, “names” and “topics.” I wrote the labels on manila filing folders in pencil so I could change them. Periodically I would reorganize the files, sometimes just a small subset of them, occasionally a more drastic revision.

These were in no way primitive “zettelkasten.” It was simply a way to retrieve paper documents. I was always able to find things pretty quickly. Not infrequently my colleagues would come to me for copies of their files. I usually had them and could retrieve them pretty quickly.

I’m not sure what lessons here are for me in this as I begin my zettelkesten journey. I know that I want to let structure to emerge. Perhaps it is to keep my categories flexible and subject to revision. I don’t know how that fits with zettelkasten’s “rigid ID.”

I’m not sure the broad categories of “names” and “topics” are appropriate in the current context. Perhaps there are others that could function in a similar way. At the moment I have no idea what they might be.

• edited August 4

My Zettelkasten starts with a Table of Contents, which links to ten "top-level-category" notes, which more or less represented my interests at the time. The understanding is that these categories would soon "break" and that I would need additional "category notes." The ID system is also a combination Folgezettel+timestamp. The Folgezettel portion indicates where in the top-level-category the note falls. If a note falls under a more specific category, it links to the corresponding category note as well. Since the TOC and (top-level-) category notes are structure notes numbered so that they are always at the beginning of the file listing in Zettlr, there is no problem remembering what the numbers mean. So we have both structure notes, Folgezettel, and timestamps.

Perhaps I should show you what this looks like in Zettlr.

Try not to laugh at my choice of categories. Sneering at them is acceptable, however.

The top-level categories are shown in the left-hand pane at the top, under TOC. Here's a convention: the last digit of a top-level category is the first digit of an ID that was inserted under that category. In the example shown below, 0000.0000.05 is the ID of the top-level category note for Mathematics. The note at the right has an ID that begins with 5. It has a few links to notes, including the top-level category note for Mathematics, and a category note for Topology. This is an example where I add new categories within the graph. I never extend the initial list of ten categories.

An aside: the top-level category note for Hard Science (Physics, Chemistry, etc) is followed by a top-level category note for "Strategic Interaction." What is that about? The economist Herbert Gintis has said that strategic interaction distinguishes the hard sciences from the behavioral sciences. That sounded reasonable to me. Under strategic interaction I might have notes about game theory, economics, law, sociology, psychology etc.

Those categories are enough for me--maybe I've left out a category that could accommodate travel or geography, for example. Well, I don't travel and like every other American, I have no sense of geography. Never heard of it. But in the Procrustean Bed I made for myself, categories such as "travel" or "geography" would have to reside under the existing categories. In the case of "geography," since this is often associated with geology, I might add a category note for "geology" under "hard science," then add "geography" under "geology." Under "geography," I might include "travel"--assuming I leave my apartment and have something to say about this. That's one way that the system might work--it's not the Dewey Decimal system. After several iterations, I tried to design system of IDs and categories for a digital Zettelkasten that would exhibit the "internal ramification" that our patron saint Niklas Luhmann wrote about. I also wanted the design to be compatible with the note categories that Ahrens writes about but doesn't illustrate. That's where I come in.

The point is that my categories serve to bootstrap my Zettelkasten. They might not be useful to anyone else. Certainly not anyone with any sense.

By now you may have stopped laughing at my choice of top-level categories. It's time to look at a non-top-level category. We'll look at topology, since that occurs in the example.

The right-hand pane of the following shows some of the notes that refer to Topology. (Just the least interesting ones.)

The first thing to notice is that the category note for Topology includes a link back to the top-level category containing it. The next thing to notice is that the category note contains the hashtag #category-note. The presence of a link to the containing category, and a hashtag identifying the note as a category note are required. Apart from the H1 header, with the ID and the note title, only those elements are required. This note includes a few entry points (so does the TOC and some of the top-level category notes).

The note on "The Simplicial Category $(\Delta)$" appears at the bottom of the Related Files pane on the right. The listing isn't complete--the examples shown are elementary (the more interesting notes are elsewhere). There is a category note for "Daniel Quillen" in the Related Files pane for the Topology category note. The category note for "Daniel Quillen" includes a link back to "Topology."

That's how this works, except for a discussion of the IDs. The Folgezettel and Timestamp portion of an ID is separated by a 0. They have to be at least 14 characters long, begin and end with a digit, and they must match the following regular expression enclosed by parentheses: (\d[\w.]{12,}\d). I describe this in more detail elsewhere.

It's considerably more time-consuming to typeset math notes in Markdown and LaTeX in a Zettelkasten than to write with pen and paper. However, on occasion some notes are worth the trouble, if I feel like it. The habit to read with pen and paper is worth cultivating, whether the notes are rewritten for the Zettelkasten. Another possibility is to adopt a somewhat telegraphic style. Scanning notes is a possibility. My inclination though is to do nothing.

Maybe other aspects of the Zettelkasten would be more interesting. I kept notes on the workflow as this evolved. Perhaps I might write about them.

Post edited by ZettelDistraction on

Erdős #2. ZK software components. “If you’re thinking without writing, you only think you’re thinking.” -- Leslie Lamport. Replies sometimes delayed since life is short.

• I’m very early to zettelkasten. I have somewhere between 30 and 40 notes. Most are structural. I’ve defined six top level categories, but they are not nearly as “top” as yours. They reflect just my somewhat diverse personal interests.

For the time being I’ve committed to what I guess I should call the folgezettel numbering convention: manually constructing numbers to reflect the development of my thinking. I add a bit of text to the number to indicate the topics or concepts addressed in notes. An example of one progression is: “3.0 politics, 3.1 abortion, 3.12 personhood, 3.12a disagreement.”

My commitment to folgezettel is tentative, experimental. For the moment I’m convinced that manually constructing the number to reflect the development of your thinking is important in getting the effect on the development of thinking I understand zettelkasten’s primary benefit. I’m sensitive to Christian’s argument for starting “zettelkastening” in an unstructured way, letting structure emerge in the process of making notes. I imagine I’ll be creating a fair number of unnumbered notes, naming them to indicate only topic or concept addressed.

I had been trying to see what I could accomplish zettelkasten-wise in vim using the wiki.vim plugin. Difficulty synchronizing multiple devices has led me to give that up. For the time being I’ll rely on bare markdown, possibly aided by research tools, in creating and structuring my notes.

• @ericweir said:
I’m not sure what lessons here are for me in this as I begin my zettelkesten journey. I know that I want to let structure to emerge. Perhaps it is to keep my categories flexible and subject to revision. I don’t know how that fits with zettelkasten’s “rigid ID.”

Categories and structures have nothing to do with "rigid IDs". Think of the ZK ID as just a random number that is used to find a particular zettel. The magic in your ZK is in a) how your zettels are connected (which allow you to "jump" from one zettel to another), b) what tags you use (to create non-rigid structure), and whether/how you use structure cards, which can impose structure from above.

• edited August 5

@ericweir said:
I’m very early to zettelkasten. I have somewhere between 30 and 40 notes. Most are structural. I’ve defined six top level categories, but they are not nearly as “top” as yours. >They reflect just my somewhat diverse personal interests.

Likewise for my top-level notes, though I am more interested in mathematics these days than anything else. The notes under computing (a category note) include some of the idiosyncratic aspects of software systems--application design isn't an entirely reasonable subject.

I see nothing wrong with bootstrapping a Zettelkasten with categories. Like Luhmann's system, the setup is designed to "recover" from a breakdown of the classification with a form of internal ramification. For some reason this is a controversial and provocative claim. I think the pronouncements from on high about how a Zettelkasten should be organized have driven prospective users away.

I don't think I've developed a single insight from my Zettelkasten. It's a system for organizing notes on topics that I would like to remember and write about. That's all. What distinguishes it from other systems is that it remains manageable over time. The word "system" here should be unpacked. The "system" comes with certain practices, such as reading with pen and paper in hand, keeping track of references in a reference manager, maintaining a standard note format, reformulating notes concisely in your own words if an improvement on the source is possible, and so on. I don't require each note to stand alone as a self-contained introduction to everything contained in it. That to me is a waste of time. Luhmann's notes became more telegraphic as he got older.

I'm hard-nosed about what rises to the level of an insight or an idea. Any worthwhile problem I've solved or insight I might have had comes from thinking a long time about something, and from starting over when an approach doesn't work out. It's not because I followed some links between notes.

For the time being I’ve committed to what I guess I should call the folgezettel numbering convention: manually constructing numbers to reflect the development of my thinking. I add a bit of text to the number to indicate the topics or concepts addressed in notes. An example of one progression is: “3.0 politics, 3.1 abortion, 3.12 personhood, 3.12a disagreement.”

That numbering aligns with my understanding of Niklas Luhmann's Folgezettel numbering. I have an ID format that combines Luhmannesque Folgezettel with Timestamps. (Folgezettel is a useful term Luhmann never used.) The ID assignment of numbers gives me a rough idea what the original sequence of notes was, but I don't agonize over the assignment. The timestamp portion gives the date. The ID format, TOC, top-level category notes, category notes and note format is described on my github. It's possible with this ID format to have two or more notes with the same Folgezettel, but different timestamps. This gives you any number of siblings. Also, since the ordering is lexicographic, 1.0.0 comes before 1.00.0.

One source of awkwardness is that after generating the ID, I have to remember to rename the file to the ID within Zettlr. But that's about it. I have a script to tell me which notes need to be updated. I'm used to the workflow.

I tried working with Timestamp IDs, but this was completely unsatisfying. I couldn't tell at a glance where anything in the Zettelkasten was. Having to search for a note or traipse through structure notes is too much work. The system I have helps me to situate notes. The file listing pane in Zettlr includes the note title (with minimal configuration), and the Related Files pane shows "backlinks."

Post edited by ZettelDistraction on

Erdős #2. ZK software components. “If you’re thinking without writing, you only think you’re thinking.” -- Leslie Lamport. Replies sometimes delayed since life is short.

• @ZettelDistraction said:

I see nothing wrong with bootstrapping a Zettelkasten with categories. Like Luhmann's system, the setup is designed to "recover" from a breakdown of the classification with a form of internal ramification.

I’ve got a few basic categories to work with. My system has “broken down” having just barely begun. I have only a minimal technology: vim and markdown. I’m going to make the best of it I can. I’m going to prioritize reading, thinking, writing. I imagine there will be a lot of unstructured notes. Perhaps I’ll be able to bring them back in later, either with my bare bones technology or with something more fancy should I happen onto it.

I don't think I've developed a single insight from my Zettelkasten. It's a system for organizing notes on topics that I would like to remember and write about. That's all….

I'm hard-nosed about what rises to the level of an insight or an idea. Any worthwhile problem I've solved or insight I might have had comes from thinking a long time about something, and from starting over when an approach doesn't work out. It's not because I followed some links between notes.

My interests are all over the place. My reading and thinking is unfocused, undisciplined. I fancy zettlekasten, primitive or more sophisticated, is going to help it become less so. I think I read somewhere that Luhmann let his reading, thinking, writing be guided by what was already in his zettlekasten. I’m a naive beginner. At this point I sense that zettelkasten is not simply a way of storing and retrieving information, but a productive way of thinking. I imagine thinking about how a current thought relates to an earlier thought or thoughts generates learning, produces still another thought.

…I have an ID format that combines Luhmannesque Folgezettel with Timestamps. (Folgezettel is a useful term Luhmann never used.) The ID assignment of numbers gives me a rough idea what the original sequence of notes was, but I don't agonize over the assignment. The timestamp portion gives the date…. It's possible with this ID format to have two or more notes with the same Folgezettel, but different timestamps…..

I tried working with Timestamp IDs, but this was completely unsatisfying. I couldn't tell at a glance where anything in the Zettelkasten was….

Again, I have only a handful of notes. I started with time stamps. The vim plugin I was using made it easy to assign them and pair them with a bit of text. It solved the problem of wanting to make more than one note on a topic.

But I sensed it was too easy. It short-circuited the thinking: “Why this note?” “To what does it relate?” “Does it extend previous thinking?” “How?” Perhaps it’s a departure. A significant one. So be it. Make a note of it.

But as I say, I’m a naive beginner.

• You are overthinking things. What is meant by letting categories naturally emerge is the following.
1. Take the book or article you are currently reading and make some notes on the ideas that you find interesting.
2. If the book is on Marine Biology, you have your first category.
3. Read a second book and take more notes.
4. If this book is also on Marine Biology, those notes will be in the same category. If the book is "How the Universe Got Its Spots by Janna Levin" (for example), you'd have a second category, Math.

Will Simpson
“Read Poetry, Listen to Good Music, and Get Exercise”
kestrelcreek.com

• edited August 5

@ericweir , there are several approaches. @Will might be right.

Without very explicit examples or the benefit of hindsight, including the bitter experience of repeated failures until a workable system emerges, you could be underthinking as well as overthinking. I did both when I started.

I won't advocate for my approach.

For me, the decision to use Timestamp IDs was underthought. They didn't work for me. I couldn't stand them. Until I wrote From Fleeting Notes to Project Notes, I hadn't thought enough about Ahrens's note categories and the practices that went with them. Some of my note templates and virtually all of my interminable posts about IDs resulted from overthinking.

So much of this is underspecified that it's easy to "overthink" to attempt to fill in the gaps that come with the underthought under-specification. Do you have a standard format for your notes?

My point? Unless you can look into someone else's mind, the accusation of "overthinking" can fairly be called patronizing, especially in an area like this where the ratio of usable specific examples to generalities is so low.

But Timestamps and Structure Notes work for many people. They may even be the path to Zettel Erleuchtung.

Post edited by ZettelDistraction on

Erdős #2. ZK software components. “If you’re thinking without writing, you only think you’re thinking.” -- Leslie Lamport. Replies sometimes delayed since life is short.

• edited August 6

@ericweir One other comment - I think it is counter-productive, when creating a series of zettels say from a book, to always start with a structure note and work out from there. When I am processing books, articles, notes from talks, etc., I write the zettels first and connect them. There may be a number of independent zettels coming out of one book or article - why would I associate them through a structure note just because they ostensibly are related (i.e., are found in the same book)? If it makes sense to create a structure note (say, for example, I have a lot of zettels on a particular topic), then I will. Outside of that, I focus on zettel creation and linking, not on creating structure notes. I believe starting with structure notes puts the cart before the horse.

Others will of course have different opinions on this topic. I just wanted you to know there is (at least) one other way to approach creating a zettelkasten.

• @Will said:

Actually, my own inclinations and frustrations with the technology I’m attempting to use, make me very sympathetic. Cognizant of, anxious about , the fact the while I’ve been fiddling with technology, that is mostly way over my head.

I had reached the point where I was just going to start throwing unnumbered notes into the folder. I do wonder about creating links after the fact, i.e., the technology for it, if there is any. But that is putting the cart before the horse.

The issue is fundamentally one of thinking and the development of thinking. Ideas about linking emerge from making notes, from thinking. They don’t, or shouldn’t, guide note making, at least at the beginning.

So maybe I’ll just start throwing unnumbered notes into the folder. Maybe being cognizant of possible links while not numbering.

• @ZettelDistraction said:
@ericweir , there are several approaches. @Will might be right.

. . . . .

So much of this is underspecified that it's easy to "overthink" to attempt to fill in the gaps that come with the underthought under-specification.

I agree. Beyond the rough idea it’s certainly underspecified in my mind. And the technology is all over the map, from paper to high-fallutin digital. And, of course, about substance my mind bounces around on the surface of a variety of topics and issues. An hope regarding zettelkasten is that it will discipline my thinking. Being undisciplined I’m hardly in a position to impose discipline.

Do you have a standard format for your notes?

No format. A title at the top that reiterates the file name, empty space for the actual note, maybe occasionally a couple rough lines at the bottom about sources.

My point? Unless you can look into someone else's mind, the accusation of "overthinking" can fairly be called patronizing, especially in an area like this where the ratio of usable specific examples to generalities is so low.

At this point I’m not gonna feel patronized. Examples and generalities, even if partisan, are food for thought.

• @GeoEng51 said:
@ericweir . . . . I think it is counter-productive, when creating a series of zettels say from a book, to always start with a structure note and work out from there. When I am processing books, articles, notes from talks, etc., I write the zettels first and connect them. There may be a number of independent zettels coming out of one book or article - why would I associate them through a structure note just because they ostensibly are related (i.e., are found in the same book)? If it makes sense to create a structure note (say, for example, I have a lot of zettels on a particular topic), then I will. Outside of that, I focus on zettel creation and linking, not on creating structure notes. I believe starting with structure notes puts the cart before the horse.

There’s a philosopher I’m interested in for her ideas about causation. She is prolific—books and articles out the wazoo. I collected a bunch of them and started reading. My notes are a few pages long, largely regurgitation of her ideas, often even of her words, paper-by-paper, chapter-by-chapter.

I suppose, on a level that’s difficult to retrieve, my thinking was affected. But these “notes” are largely unusable. They require to be read themselves, and having real notes made on them.

So, going forward, I’ll be writing short notes. Of my own thoughts about what I’m reading, or observing, or whatever. Perhaps that’s not your advice. But perhaps you’re not unsympathetic to it.

• edited August 8

@ericweir said:
At this point I’m not gonna feel patronized. Examples and generalities, even if partisan, are food for thought.

I probably shouldn't have volunteered my opinion on "overthinking." I've never found it particularly helpful to be told that I (notice the pronoun) am overthinking, but you're right to ignore it. It's more helpful to be informed what the procedure is without getting into personalities and inferences about their thinking, but I digress.

I have a Zettel format on my github, for what it's worth. It's simpler that the original version. The ID format of that sample doesn't match the ID format I currently use. And bear in mind that others have their formats. @Will keeps the ID to the right of the title. Mine is derived (sort of) from the formats of @Sascha and @ctietze. Also my setup was partly imposed by software, since I use Zettlr and I need to produce LaTeX documents. The constraints aren't a straight jacket. I was able to work through all of them and develop a workable system.

Erdős #2. ZK software components. “If you’re thinking without writing, you only think you’re thinking.” -- Leslie Lamport. Replies sometimes delayed since life is short.

• @ericweir said:
There’s a philosopher I’m interested in for her ideas about causation. She is prolific—books and articles out the wazoo. I collected a bunch of them and started reading. My notes are a few pages long, largely regurgitation of her ideas, often even of her words, paper-by-paper, chapter-by-chapter.

I suppose, on a level that’s difficult to retrieve, my thinking was affected. But these “notes” are largely unusable. They require to be read themselves, and having real notes made on them.

So, going forward, I’ll be writing short notes. Of my own thoughts about what I’m reading, or observing, or whatever. Perhaps that’s not your advice. But perhaps you’re not unsympathetic to it.

You've stated it well. I have done the same thing - taken copious "notes" that essentially parroted back what the writer said. The trick is to write a very brief statement of an idea and then to state what you think about it. It's hard work to do it well. But that's why each zettel is supposed to be atomic. However, I agree with what you are saying.

• @ericweir said: There’s a philosopher I’m interested in for her ideas about causation. She is prolific—books and articles out the REDACTED.

If I may ask, which philosopher is this? BTW I also find notes with enough to reconstruct an argument more efficient than fully detailed, self-contained notes. I use links for details and background. Ideally each note represents a "chunk" or a "schema." Links to other notes can fill in the details of a "chunk." The aim is for compression.

Erdős #2. ZK software components. “If you’re thinking without writing, you only think you’re thinking.” -- Leslie Lamport. Replies sometimes delayed since life is short.