π

Zettelkasten/Org-roam/Org-brain Is Crap

Show Sidebar

Update 2020-06-16: Comment by Alan Schmitt

First, I'd like to apologize for the rude title. I used this rhetoric trick to get the right kind of people into reading these lines: People who care about the Zettelkasten-method and who are able to express the usefulness of it for themselves.

Now that I have your attention, I'd like to write down my personal concerns about the Zettelkasten method which may not reflect your reality. I then would love to read about your arguments why my prejudice might not be valid in general.

My pile of Org mode data is huge. There are certain costs attached to migrating my data to a Zettelkasten method, even if this would only affect my notes.org file which contains my general knowledge base.

At this moment in time, I doubt that I will move to a Zettelkasten approach. Nevertheless, I'd love to read your comments that try to falsify (or confirm?) my worries on that topic.

Zettel, Zettelkasten, Zettelkästen, Umlaute

If you do not know about Zettelkasten method, you might want to check out this brief intro with a non-Emacs tool or the Wikipedia article.

My mother language is German. Since I've noticed that non-German people are confused about those terms, I would love to add a brief explanation here.

For practical reasons, I'm using an "US international no dead keys" keyboard layout. Therefore, I'm able to tell you how to write German umlauts with it: Combine " with any other character to get something like "ä": " + a → ä

See? You're now able to type those funny looking characters on a non-German keyboard layout.

Multiple Zettelkasten-Domains

I'm not sure how difficult it is to run a setup that consists of:

  1. Normal Org files that contain my TODOs and data that does not relate to knowledge-base
  2. A Zettelkasten for my personal non-research knowledge-base
  3. A Zettelkasten for my personal PIM-research knowledge-base
    • This might be merged with my non-research knowledge-base
  4. A Zettelkasten for my business knowledge-base
    • This definitively needs to be separated from my personal knowledge-base Zettelkasten.
  5. Potentially even more Zettelkästen with different/distinct knowledge-bases for absolute separated domains

Search

As far as I understood, besides navigating via links, searching within Zettelkästen is mostly based on tags and titles of those Zettel. However, I prefer the body text being part of the search space.

Currently, I usually use tags when normal search results return too many candidates to skim. In recent months, I learned to love helm-org-rifle-* for searching the current file or all of my agenda files.

Complexity

With any Zettelkasten implementation, setup and usage complexity seems to be significantly higher. Some even require a non-Emacs-based server or data-base.

Especially when multiple computers are involved (personal desktop, personal notebook, business notebook), I'd love to keep the same, synchronized setup among those devices. Currently, I share a common Emacs configuration which is easily set up on a new machine.

Mobile Usage

Normal Org files may be used on my Android smart phone with Orgzly. I don't think that usage patterns, especially search, is possible with synchronized Zettelkasten files on my phone.

No Radio Links for Glossary

There can not be a common glossary: automatic links of known "concept" to their definition by radio targets. I learned to love this feature for my business files. Whenever I stumble on a new term, I write down its definition and add those radio targets. All occurrences of this term are now automatically linked to its definition. This has tremendous value to me.

With a business Zettelkasten, I'd have to add links to definitions or concepts manually. I'm afraid this is almost impossible for multiple reasons. First of all, I'd have to remember about their existence by heart which I don't, given this large amount of terms. And if this would not be the limiting factor, I guess that would be too much effort for me.

Performance

With having that much data in Org files, I do face various performance issues. With switching from "few but large Org files" to "many different smaller Org files", I do have a gut feeling that several operations might be affected in a very bad way.

Split Brain Issues

Currently, I don't have to care where I'm adding new TODOs since all of my major Org files are part of my agenda. With Zettelkästen, I don't think that this can be continued.

I could add all Zettelkasten files to my agenda but this most likely would have a dramatic effect on performance. Therefore, I would have to change my set of mind. Before adding a new TODO, I'd have to think twice if I'm within a Zettelkasten file or in a normal Org file. If

I'd like to add a TODO when reading/writing a Zettel, I'd have to spot a non-Zettelkasten heading that is suited as a parent node for my TODO.

I'm not certain how much this would affect me since my knowledge-base tends to hold less TODOs than the other files. However it bothers me a bit.

Same holds true for my blog. I wrote the lazyblorg blogging software so that I am able to define blog entries everywhere. This would not be valid any more for blog articles within Zettel.

Surfing Concepts

I'm not sure about the usefulness of browsing the links, which seems to be one of the central advantages of Zettelkästen.

Maybe that'd change with getting personal experienced within lots of connections. So far, I highly doubt it. Maybe this is different for scientific notes such as organizing and annotating papers?

Your Turn

If you do have experience with the Zettelkasten method, please do tell me your thoughts on my concerns. If mine are wrong, partially wrong of even when you find that I do have good points where Zettelkasten does have some disadvantages that might be mitigated or not.

Please do mention briefly what Zettelkasten implementation you are using and probably link any setup reports you might have authored somewhere.

Comment by Alan Schmitt

So I discovered the Zettelkasten (singular) method a few months ago, and I've been trying it. I currently use Org-Roam, with deft for full text search (but I don't need it that much) and org-journal for the TODO integration and ease of input.
What I like about the approach is that I don't have to think about where to store something when I have an idea. I just need to think about something related, and link to it. I cannot say whether this will work out in the long term, but I'm currently finding things faster than with my monolithic brain.org file. I'm also writing down more things, and revisiting them more often.
I have not migrated my old notes. What I do is put links to them, and when I revisit them I migrate them. But I don't do this much.
I only have one Zettelkasten. I don't feel like having to compartmentalize things, otherwise I'll get back to the "where do I put this" question. So I have a huge toto.org file, a huge brain.org file, and many tiny org file in my Zettelkasten. Agenda uses the daily Zettelkasten files (through org-journal) and the big todo.org file.
Searching is easy using deft or rg (for instance through projectile, it's all part of the same git repo).
I don't use multiple computers. If I did, I would run org-roam-db-build-cache after a git pull. I'm not sure it's necessary, I have not tried it. I also don't use org files in mobile (I have a write-only approach where I use orgzly to input ideas, but they get processed on the computer).
Radio target are nice, but with company enabled I have links created automatically to entries in my Zettelkasten (I get to choose them of course).
I have not encountered any performance issue for the moment, but I've only been using the method for a few months.
Regarding split brain issues, one does not need to use the approach for TODO items. It can only be used for reference/diary, and TODOs can still live in a monolithic agenda file (with interlinking when necessary). I'm experiencing with the org-journal TODO approach, but it's too early for me to give feedback.
I have found Zettelkasten very useful to explore complex ideas, such as taking notes learning category theory. I also follow some of [[https://rgoswami.me/posts/org-note-workflow/][this workflow to take notes when reading scientific articles]]. It works particularly well.
To sum up, I'm happy with it, maybe because it fits well with my way of working. I won't be advocating for it until I've used it for a year… so if you want to ping me around January 2021, I'll tell you it stuck.

Thanks Alan for your comments! I've added a few links here and there.

Comment via email or via Disqus comments below: