Retaining the benefits of Luhmann's numbering system without damaging key Obsidian functionality?

It seems to me that one of the greatest benefits of Luhmann’s Zettelkasten approach was a byproduct of his numbering system, and it’s something I have not been able to entirely reproduce in Obsidian or its plugins without limiting the autocomplete links and/or file explorer functionalities. The benefit in question is that Luhmann was able to generate topics without ever once having to identify the topic.

  1. Let’s assume a numbering system of 1, 2, 2a, 2b, 3a, 3b, 3b1, 3b2, 3b2a, etc).
  2. Let’s also assume that every time we write a new note, in addition to linking it to related notes, we file it directly after the note to which it feels most pertinent. So if it’s most pertinent to note 4c2a, we number the note 4c2b.
  3. After several weeks of work, it becomes apparent that clusters are forming. Let’s say there are 30 notes between 4c2b and 4c2c.
  4. We can be reasonably certain that there is a fertile area of inquiry starting at card 4c2b, and we have discovered it without our thinking being once distorted, disrupted, or limited by the question of taxonomy.

Graph view offers a partial solution to this problem, but it makes no distinction between the weak links of mere association versus the extraordinarily strong links implied by direct serial proximity. Our choice of which single note we filed a new note behind contains important information that is lost in graph view.

One obvious solution is to just name the files 1.md, 1a.md, etc. But that destroys autocomplete and file explorer functionalities.

The next idea is to name files e.g. “1a mirroring and free will.md”. This is the method I’m using, but it’s less than ideal. It retains most of the functionality of autocomplete but limits file explorer functionality. It slows down the process of creating links and makes markdown significantly uglier and less legible. Nobody wants to see a bunch of links that look like this in their header: [[1a1a1a1a1a testimonial knowledge|testimonial knowledge]]. Moreover, this approach limits the functionality of the file explorer, in which every filename will be preceded by note numbers of varying character length, so that in order to see the more informative part of the file name (‘testimonial knowlege’), we’ll have to enlarge the file explorer pane significantly, reducing the size of our workspace. It also clutters and makes hideous the graph view.

I am quite confident that there is already an elegant solution to this, and that I just haven’t found it yet in the forums/general PKM discourse. Can anyone point me in the right direction?

Thank you!

13 Likes

I also am interested in naming conventions. My habit recently is to collected literature notes with a standard “DATE Article Title” format. In a scratch note I write bullet points of those articles under a link to the original, forming a long chain of links and bullets. I slowly rearrange those until I have enough that are forming a coherent topic to cut the group into their own, named note. This pulls the links with them, thus updating the graph. If a set of bullets have multiple use in other Evergreen notes, I will copy them (with the link) multiple times.

The name of that “Evergreen” note itself follows the “DATE Statement/Topic” format combining the literature note and Evergreen note formats. It works so far, but I would also like to find something more elegant.

1 Like

It seems to me that Use H1 or front-matter title instead of or in addition to filename as display name would solve this provided that I could keep the files sorted by actual filename while displaying the title from the frontmatter.

3 Likes

I’m just reading How to Take Smart Notes, and that is my only knowledge of Zettlekasten, beyond trying Zettlr so far. It definitely seems like every software implementation of Zettlekasten that I’ve seen so far completely misses the idea of his linking convention, by only including a date-based UUID, which doesn’t link conceptually at all.

(Zettlr uses UUIDs, but you can also include names in the links, and the links recognize and use the UUID and ignore the text.)

(I haven’t tried http://zettelkasten.danielluedecke.de/en/ because it requires Java to install. But Smart Notes recommends it. I’m wondering if it has a better implementation that does have those kinds of strand links. Do you know?)

This doesn’t address the numbering you are talking about, but it could be a good supplement. Have you read up on the concept of Maps of Content by nickmilo? On the process of making MOCs

5 Likes

Thank you for these ideas! Glad I’m not the only one puzzled by the date-based IDs.

If I remember correctly from Smart Notes, Zkn3 does indeed have note sequences, though I’m not sure if the feature is recursive. I haven’t tried it yet for the same reason you cited.

I actually am using MOCs right now, but in my opinion the ‘Mental Squeeze Point’ nickmilo refers to in the documentation…

A ‘Mental Squeeze Point’ is when your unsorted knowledge becomes so messy it overwhelms and discourages you.

…which according to nickmilo is the triggering occasion for starting an MOC, occurs long before a truly rich cluster has the time to accrete.

Not to mention the fact that in order to have an MOC, you need an MOC filename, and in order to usefully name the file, you have to think taxonomically or in terms of a project, which is what I’m desperately trying to avoid doing too early in the process.

Although it’s very possible that I just haven’t wrapped my head entirely around MOCs and my grievances are unfair. It usually takes me a little while to digest PKM concepts fully. If there are any MOC die-hards reading, please feel free to school me.

3 Likes

I don’t see any danger in prematurely naming a MOC file. Since you can rename files and have any links automatically be updated. Are you more worried about biasing your own early thoughts?

But I definitely understand what you are saying about how these squeeze points can occur, and halt progress.

This is completely tangential, but I love the Random Notes plugin. I do random sessions fairly often, and it feels a bit more like wandering and gardening organically, rather than engineering, and attempting to architect a correct structure ahead of time. It removes a bit of the burden of needing to remember or organize upfront. It’s a review process that removes a bit of the biased knots of these squeeze points.

(But again, that’s a tangent, and doesn’t address sequences.)

3 Likes

This is exactly my concern, yeah. I’m imagining working with a note extensively in the context of a specific MOC to such an extent that I start to think sort of rigidly and lazily about it.

1 Like

Luhmann’s system gave him both a UID and a sequence. As you say, software has picked up the first aspect but not the second. Many argue that there’s no need for the sequence, but that ignores the gain from conscious development of network links.

Linking is the obvious way to go, but it would require the ability to specify types of link. I can think of a number of uses for being able to do that. Having them in different colours in the graph would be cool.
Ideally it could be done as part of making a link, but I think that would require another syntax extension.
Maybe a plugin could cobble something together using a Parent link in the YAML.

3 Likes

You could have two types of links. The first using the normal [[ ]] syntax and another type of link where you force a connection not to a node but to a group of nodes? The question becomes how do you define groups of nodes. This could be done by an algorithm or manually if you decide that some nodes should be grouped in a certain way.

1 Like

Yeah, I think it would be wonderful if in the future we had some kind of a semantic or logical link structure.

In the way that you can process and monitor signal flow through a modular audio system, or connect things via specific datatypes in programming functions, or in node-based media apps. You can do powerful things when you can control and shape the structure of the connections and flow.

You could conceivably have better filtering, analysis and navigation if links had purpose and meaning baked in, without requiring a specific naming convention in the filenames or text of the link.

@lexane mentioned “logical links” here: The state of backlinks

And in this thread, there is discussion of semantic links, and the way Tinderbox works. Making mindmaps from notes a graph with a defined semantic relation

In Tinderbox, (I’ve briefly tried it, but gave up due to complexity) the shape, placement of the network is part of the structure, more like a purposeful mind map.

5 Likes

So it turns out there has been a debate raging for years over the value of note proximity in digital contexts-- the term among Zettelkasten die-hards is folgezettel. For anyone interested, you can read cases for and against at the links below. I remain convinced that digital folgezettel is valuable, but I’d be interested to hear other people’s thoughts.

  1. https://zettelkasten.de/posts/luhmann-folgezettel-truth/
  2. https://strengejacke.wordpress.com/2015/11/01/you-underestimate-the-power-of-the-dark-folgezettel/
  3. https://zettelkasten.de/posts/understanding-hierarchy-translating-folgezettel/
  4. https://improveism.com/digital-folgezettel/

EDIT: lizardmenfromspace has a much better roundup of the folgezettel debate here: Obsidian Zettelkasten

8 Likes

Actually, it seems to me this is a simple issue, or, more accurately, a non-issue.

What counts at the end of the day is what works for each person individually. If someone finds value in having folgezettel, then why not? If someone prefers to do without them because they don’t see value, then that is a valid choice too.

One can debate about it ad infinitum, but it will remain a moot point, but there cannot and should be a conclusion as to one or the other being right or wrong.

1 Like

If you aren’t using tags for anything else, you could re-purpose tags for identifying folgezettel sequences. You could just add tags like #1, #1A, #1A2, etc.

3 Likes

Just dropping a note here since I came across this discussion, to say I am experimenting with Luhmann-style IDs (after initially thinking they were rubbish) and have found some very interesting things emerging in my note taking. When I was first setting up my note system using the LYT-ish approach I was building a variety of notes, some short and some long, and trying to use the Andy Matuschak style of note taking but not having much luck. Much of what I was doing started to feel like replicating wikipedia entries – it was a “code smell” - an indicator of an anti pattern to be avoided.

I’ve since revisited Luhmann’s approach and done some reading including translating and reviewing article about his process from German. I’ve come to think the hierarchical branches of his ID system – which was probably a hack he developed to deal with the constraint of his box filing system – has a profound shaping effect on the notes that are taken. At least for myself, so far, the notes I’m writing are far more tightly focused, and a note that elaborates on a previous note is right there next to the original idea.

As a side effect it encourages me to write more in the Andy Matuschak style, which I was already sold on as it has had a powerful effect on my thinking even in the short time I’ve been using this tool. This method makes that style of writing easier to me. Contrast that with the other methods which to me were too free form. Constraints shape our behavior and encourage creativity, and this seems to be an example.

And to make it easier to see the hierarchical branching of concepts I added this custom CSS snippet which converts the filenames (not folder names) into monospace and underlines the folder names. So now there is a clear visual distinction between folders and files at a glance and the file names with the Luhmann ID prefixes are in monospaced format so the hierarchy is actually easily observed when scanning the file explorer.

.nav-folder-title {
  text-decoration: underline;
  color: var(--text-faint);
}

.nav-file-title-content {
  color: var(--text-muted);
}

.nav-files-container {
  font-family: 'droid sans mono for powerline';
  font-weight: light;
}

The above styles look good to me on dark mode with the Atom theme. Others’ mileage may vary.

I disregarded this hierarchical ID at first, but after reading about how his system actually worked I believe he was onto something and that attempts to replace his numbering system with timestamps or to just use links misses something subtle but powerful.

One other benefit: since one of the main benefits of his system is that you can build sequences/subtrees of thought throughout the ZK filing system, and since the “root” of every one of these sequences anywhere in the system are also notes, then any note can become a “root note” or “hub note” and the distinction between “is this a note or a collection of links to other notes” starts to blur. Which creates a surprising effect that the idea of a “MOC” or “hub note” basically becomes any note you want anywhere in the hierarchy, and by creating it you are effectively declaring a new hub that has additional context and semantic meaning due to its proximity to other (presumably related, at least roughly) notes e.g. siblings. It’s not perfect (he eschewed formal taxonomy, which is something many miss and get lost trying to create up front, which is flawed) but it worked extremely well for him. I’m still early in my exploration of this but it feels shockingly powerful and flexible.

Maybe that helps someone else considering this approach. Take care.

9 Likes

@davecan: an interesting analysis and approach, which makes a lot more sense to employ than any of the reasonings I have seen so far. Having said that, I still believe a Luhmann system of note sequences is obsolete in a digital set-up.

Nevertheless, you seem to be making a useful transition from the physical to the digital, pushing boundaries, blurring boundaries, and that, in my view, is new. Congrats, and if it works positively for you, well done!!

Discussion about syntax for link types/attributes: Add support for link types

Link types can already be visualized in Neo4j Graph View Plugin.
It works with any naming style (Luhman, timestamp, Obsidian title, …) because structure is represented by link type (for now specified before link in markdown).

Maybe Luhman sequences can be visualized by mentioned ‟Hierarchical layout” for “continuation”-type links, reducing significance of other link-types
Linking to group of nodes requested by @rui in post #10 ↑ can be done by creating file for desired group. That file contains list of “member”-type links to each node/file which belongs to the group.

1 Like

Hi @davecan ! Thanks for your info, it’s extremely helpful.

I’m interested in what you’re doing, would you mind sharing some pictures of how you have set this up? It would be super useful.

Thanks

Hi @Sebastian. Actually, since I posted this I discovered the above list was encouraging me into an anti pattern of allowing the hierarchy to establish the forward-linking implicitly, instead of explicitly linking from each note to the next note in sequence. My belief is the forward linking is powerful and that was also one of Luhmann’s primary rules in his system, that every note must be linked from a note it builds on somehow. So the heuristic is every note must be linked from somewhere. What the above hierarchy ended up doing was creating that relationship via the numbering system rather than links, so it was actively fighting against the built-in hyperlinks in Obsidian. It was also ossifying my thinking as I started thinking about where a note should go in a hierarchy of notes, so in effect any note with “child” notes was just a folder in a sense, which is an anti pattern.

I’ve since abandoned this approach and switched to a modified timestamp format as a Zettelkasten UID prefix. I use YYDDDDHHSSS which is the Julian date (year + day of the year) followed by the hour and then the current millisecond as a pseudo-random number. Since the previously discussed numbering system was forcing me into too much of a hierarchy I opted to use this which minimizes collisions while still retaining some element of randomness.

I’m doing this to consciously and proactively fight against centralization and hierarchification (not a word but should be) which has always been a natural tendency I’ve had in previous note taking systems I’ve adopted.

For elaborations that make sense in a sequence I can link to their notes sequentially in another standalone note, rather than trying to force sequentialization through file naming conventions. This gives me more freedom since its just another note and I can mark it up as I see fit. (in other words, I can create or expand on one or more MOCs with the set or sequence of links if needed)

7 Likes

@davecan: so, if I understand your last comment correctly, you have abandoned Luhmann’s sequential numbering system?

1 Like

Thanks for your insight! I guess this is an evolving process, and I’m glad that you’re making progress.

I’d like to know the same as @Klaas , also I have two questions now:

  1. How does this new UID with a date format and pseudo-random numbers works with this statement you made previously:

At least for myself, so far, the notes I’m writing are far more tightly focused, and a note that elaborates on a previous note is right there next to the original idea.

This seemed like a good method, but I’m not sure you’re achieving that with your new system.

Also, would you elaborate or direct me with what you mean with anti-pattern?

thanks again!

1 Like