Another Discussion on The Value of UIDs

I’m guessing I’m not alone on this…

I feel friction between Zettelkasten principles and the powerful technology provided by apps like Obsidian. One place for me is in the UID Debate.

I began my Zettelkasten with UID files names (e.g. 202009120833) and then when I found Obsidian and its magical updating of backlines, I evolved to UID + Titles (e.g. 202009120833 - An Argument for Intelligent Design). Then, after the helpful influence of @nickmilo and his (IMF v3 (LYT Kit v4 coming in mid-Sept)), my File names became just titles (e.g. An Argument for Intelligent Design). (I really try to make my file names a summary statement of the argument of the note.)

But I am feeling friction when linking because sometimes, the title will make perfect sense in a note, other times it actually confuses me and makes it hard understand the content. In other words, the title can be distracting. One of the struggles I have with the Principle of Atomic Notes is that it fragments my arguments so much I get lost. My field of research is theology, so I am often involved in complex, abstract arguments that only make sense with a lot of context. So when I begin to make atomic notes, I get lost while linking.

Here, I can see the advantage of a UID-only file name because I can then link with a UID and then add a description of the note based on the needs of that instance. If I’ve written a good note, this description may already be contained in the H1.

==The problem then, is how do FIND my content?!== Obsidian’s [[]] search seems to like words in the file name and I’m not a heavy user of tags. So if I use UIDs I’m afraid my content, in a folder-less and title-less structure, will get lost.

As far as I know, there is no search operator that allows me just to search my Headers #'s. Is that correct? I know I can search the Headers in a specific file, but I have to choose the file first.

Is anyone using a UID based system and have a solution for searching? Or is adding a search operator for searching Headers a feature request?

(Side note: I’ve found it is hard to articulate my PKM questions. Its a learned skill for sure.)

Your headers normally are formatted # Someheaderwith one or more # that show the level.

Why wouldn’t you be able to search on that.

Type /#/ in the find/search window.

This wil give you a list of all Header 1 headers in all the notes of the vault.
You don’t need the / (they indicate a regular expression)

Hope I didn’t misunderstood your question

@RikD Thanks for the reply.

Type /#/ in the find/search window.

That’s helpful. However I was actually referring to the back link search dialogue box [[]]. That search only includes file names.

Yes, it is hard talking in PKM language, partly because the language doesn’t really exist yet, so we make it emerge as the necessity arises…

My initial thought is "have you tried the “pipe link” aka “aliasing” your filename titles? So even if the filename isn’t written in a way that would flow with your current sentence, you could make it flow.

For example,

I'm consciously thinking about how to increase the number of [[Shot on Goal|Shots on Goal]] I take.

This solution allows you to keep Word IDs but adjust the Display to best fit the context of the current note.

Pipe Links were one of the things that gave me confidence that my PKM would be okay even without UIDs. YMMV

@nickmilo Thanks for the reply. I have used the pipe links some. Perhaps I need to keep trying them. For some reason it doesn’t feel very intuitive. Plus it looks really messy in the Edit view. Either way, I might need to keep at it. Thanks again.

I’ve also experienced minor difficulties with note titles. To illustrate, I had a note on ego depletion; the note’s file name at first was “Willpower is limited,” which I then linked to a bunch of notes. I then decided to change the name to “Ego depletion.”

That of course disrupted how the note was used in the different contexts of other notes, i.e. …keep in mind that [[Willpower is limited]]. became …keep in mind that [[Ego depletion]]. And so whenever I would need to rename notes, I would have to go back to its backlinks to see if the name change has disturbed its usage in other notes, which is a hassle when you have a bunch of backlinks.

Sometimes I do feel that it makes sense to link to notes the way that Andy Matuschak does and the way that I did in my example above, and sometimes I end up feeling that UIDs are much better as I don’t feel overwhelmed to decide on a concrete title on the get-go.

By using UIDs, I would need to use pipes every time I link to notes; though it’s certainly less hassle when changing the note’s title as custom text on links do not change, it’s also annoying to see an extraordinarily long link when I’m in edit mode (a bit of a nitpick-y complaint but it’s what’s appealing about linking with note titles instead of pipes).

I think my problem above would be mitigated once WYSIWYG mode becomes a thing. And as been mentioned in previous posts about UIDs in Obsidian, the difficulty of finding notes with UIDs as file names would be somewhat mitigated if we are able to use H1s instead of file names (this feature request: Use H1 or front-matter title instead of or in addition to filename as display name). That said, I don’t think the debate on UIDs in Obsidian would be resolved if the devs decide to implement these as people will continue to decide for themselves which would be the right route ¯\_(ツ)_/¯.


Good reply. I’m glad to hear I’m not alone in this.

This is the exact problem I’m having. I agree, my friction with piped links is how confusing it looks in edit view.

I think the feature request Use H1 or front-matter title instead of or in addition to filename as display name would solve it for me.


@Squatpushpull my current thoughts on it below. I currently do Time ID + Title (e.g. "202006011558 Metacognition).

So this is why Luhmann used an index combined with structural linking (notes 1a, 1b, 1c are structurally/thematically linked together) and direct links (e.g. writing “see note 1b” in the middle of a note). So he would use the index to get as close as he can to the topic he is thinking about, then manually go through the notes until he found what he was looking for. This is what you’d have to do unless you give every note a unique tag, which is essentially a secondary title.