Zettelkasten Forum

What physical steps do you take when linking existing Zettels to a new one?

It takes me about 20-30 minutes to create 1 Zettel sometimes because my linking process often involves sifting through my Slipbox to find the right note to place my note behind. As my Slipbox grows, it becomes cumbersome to keep this up.

I see plenty of posts share abstract advice for linking Zettels together and implementing Folgezettel.

  • Each link should create knowledge
  • Create a new Zettel when you're starting to branch into a second idea
  • "Find" Zettels that match your topic (but don't you dare use the Search function to do it!)
  • You need Folgezettel / you don't need Folgezettel (I personally like the structure, so I've chosen to keep it)

What I'd like to learn from experienced users is this: What are the PHYSICAL steps you take when you've completed a Zettel and you're linking it to your other Zettels?

By physical steps, I mean what you actually do, GTD-style Next Actions.

For instance, here's my physical action steps.

  1. Search my index for the topic of interest
  2. Look within that cluster to find a Zettel that's either a source of that idea or similar to that idea.
  3. Number the new Zettel after the one I chose from #2.
  4. Use the Obsidian search function for a few keywords within my Zettel to find relevant links, adding context phrases as I do so.

But the problem is that as my Slipbox grows, I'll slowly forget the Zettel clusters and may make less-than-ideal connections. Any guidance here would be greatly appreciated. Thanks!


  • @Mach3Maelstrom

    1. As I "complete" each zettel, I make sure it has a thorough set of tags, including one called "#unlinked".
    2. Depending on time and interest, I may make some initial links to the zettel I've just completed, rather randomly I admit.
    3. Every week I spend time on ZK maintenance. One of those is to search for the tag "#unlinked" and pick one or two zettels for a focused linking session.
    4. I click on each tag on the current zettel to see what other zettels have the same tag, and check to see if any of them should be linked.
    5. I search using other terms that I think are relevant to the current zettel and do the same thing.
    6. Once I have more than 2 or 3 links, I remove the "#unlinked" tag from the zettel. The number of links is a bit arbitrary - sometimes I think there should be more, so I create more before removing that tag.
    7. Another regular maintenance item is to scan my list of tags (that list is created using a Keyboard Maestro macro shared by @Will ). I don't want any tag to appear on more than 5 to 10 zettels. If say the zettel count for a particular tag is more than 10, I will review those zettels and then split the tag into two related but distinct tags and re-tag the related zettels. For example, currently the tag "#Family" shows up on 12 zettels; I could split it into "#Family-Immediate" and "#Family-Extended" (if that made sense). My list of tags then becomes a sort of index for my ZK - but only because I don't let any tag get too bloated.
    8. At present, I have 300 zettels (after 3 years of steady but you might say "slow" work - I've explained why at the following link) and 180 tags. My zettels are typically 100 to 300 words long.

    See https://forum.zettelkasten.de/discussion/2431/creating-a-personal-history-in-your-zettelkasten/

  • Thanks for sharing your process, @GeoEng51 ! There are a few points I'll test out in my own Zettelkasten process:

    1. Using tags to aid in linking. I personally prefer to use a dedicated Index, but seeing how your tags aid in linking to Zettels helps me see how I can leverage tags myself. I'm not using them at all in my Obsidian-based Zettelkasten.
    2. Limiting number of notes per tag. This is a good solution to help keep tags under control.
    3. Dedicated linking session. I'll test separating the linking from the creation to see if I can reduce the amount of time it takes to produce Zettels. If it doesn't work out for me, at least they'll be tagged #unlinked, so it's an interesting experiment.
  • What physical steps?

    You put your Zettel in
    You put your Zettel out
    You put your Zettel in
    And you shake it all about

    You do the Zettelkasten
    And you turn yourself around
    That's what it's all about

    As for procedural steps, I try to link notes as soon as I add them. Otherwise I review my ZK on Sunday. I wrote a script to find notes whose H1 header doesn't match the filename, and I correct those.


    Given $(n)$ nodes, there are $(3^\binom n 2)$ possible directed graphs with at most one directed edge between nodes, which means that you can expect to spend most of your life tending to your Zettelkasten and never finishing if you add notes over time and you are a link perfectionist. I have links in both directions in some cases, so I might have to choose between $(2^{2\binom n 2})$ possible digraphs.

    What about diminishing returns?

    If you're a programmer, you might have heard about "continuous improvement." Some people will never understand opportunity cost. Look at the graphs below and or the formulas above and you might appreciate that the brains of such individuals didn't prune away or otherwise weaken the connections that somehow underwrite (or underlink) this non-economic notion that diminishing returns don't exist and that it always worthwhile to improve anything, no matter what.

    Maybe you don't care about producing anything with your notes, and your aim is to become a world-class Zettelkasten gardener. More power to you, in that case.a


    I've gotten Zettlr and Obsidian to work together somewhat interchangeably. Obsidian can be configured with about three or four plugins to use the same Pandoc format and LaTeX template files as Zettlr, work the same with Pandoc citations as Zettlr, show the H1 header as the filename in the File Explorer instead of only the ID, which Zettlr can be configured to do, and so on.

    I use the graph view in Obsidian to see what links where. Obsidian's graph view has improved. I still prefer to edit in Zettlr. For example, suppose I want to focus on one of the alphabetic index nodes. Previously these were category notes until I decided to switch to an alphabetic index. The indexing is somewhat chaotic because of this, but chaos is supposed to be good in the land of the Zettel.

    You can adjust the link distance, node repel force, central force, turn arrows on, zoom in and out (with Alt-, Alt=) so that the graph display becomes useful. It's also possible to display hashtags instead of titles. These views help with adjustments, I can see where I need to do more work, and so on.

    Here's a note on LaTeX--this is more of a hub node. The graph tells me that the link to the index note, which was "Computing" and now "C-E" should be changed to point to "H-L", which itself has too many letters. Even with the index jumbled after I switched from categories to alphabetic indexing, the alphabetic indexing still useful.

    A parting shot

    I said that I did away with categories and category notes in favor of an alphabetic index. What an alphbetic index gives you is a collection of "soft" category notes, or a collection of structure notes which organize the notes of your Zettelkasten alphabetically.

    Huh, whut??

    A distinction without much difference in practice

    There isn't that much difference between a category note and a structure note, or a hub note. As long as you have a structure note, links from it form a category by definition: the set of notes linked to by the structure note. Typically there is an organizing notion that binds the links of a structure note together, which gives you a category. Conversely, a category note is already a structure note--if there are only backlinks from other notes to the category note, you could convert the category note into a structure note by adding all of those backlinks to the category note. This process is reversible like a change of coordinates, so you might as well call a category note a structure note.

    The argument over structure notes versus category notes is over a distinction without that much of a difference. I will say though that starting with an alphabetic index seems to be preferable to starting with a selection of categories in the long run. But the alphabetic index is another kind of collection of category notes, under which other notes that start with a given letter are in the same category. Or else it's another collection of structure notes, under which other notes that contain words starting with a given letter (in the title or elsewhere) are linked to from the same structure note, or which link back to the same hub note.

    aThis is not medical advice, but it helped me to obtain a transcranial direct current stimulation (tDCS) machine and connect the electrodes to my scalp in the F3-/F4+ montage at less than 2mA. (The 2mA current upper bound is a de facto standard, but do your own research.)
    Perhaps categories are less general than structure notes, which can be entirely random. I leave such fine distinctions to the analytic philosophers.
    It's possible that an alphabetic index doesn't box you in as much as an initial list of categories, relying on internal ramification to find subcategories of these, and so on. A hub note which serves only as a target for backlinks can save some typing (in the body of the hub note). The notes such a hub note classifies are visible in a "what links here" pane in some ZK editors. We are past the point of diminishing returns.

    Post edited by ZettelDistraction at

    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.

Sign In or Register to comment.