# Is the three-layer structure of evidence a generally useful principle?

In a previous post, @sfast wrote that evidence captured in ZK notes can be on three layers, but I am wondering whether this is really a useful general principle (as it's presented) or is something that is actually more specific to a more limited subset of certain types of knowledge work. The layers are:

1. Data description and patterns.
2. Interpretation of descriptions and patterns.
3. Synthesis of patterns, descriptions and interpretation.

Although this division makes sense in the abstract, I have had trouble seeing how these layers could be implemented with my workflow, which is primarily writing about academic theory in the humanities. @ctietze Christian's comment on Sasha's post partly highlights the issue:

Sascha’s post and separation of layers reminds me of my struggle and I think that I could’ve used this approach for non-empirical reading of Kant as well: (1) What does Kant say, quoted? (2) What do I think does this mean? Why is it interesting? How does secondary literature interpret the part? (3) … well, there’s not much practical application of Pure Reason, besides beginning to think differently about the universe and everything of course.

Christian's 2nd layer actually includes a number of different things - his interpretation, others' interpretations, and maybe his analysis of their positions. And the third layer he cites need not be practical application, but could be a synthesis of one's interpretation of Kant's arguments, or of others' interpretations, etc.

My concern is that the three layers proposed are appropriate for communicating summaries and first-level interpretations of objective data/facts. A very typical case would be the type of writing that it seems to me that Sasha is doing, which is summarizing scientific findings to support a particular recommendation of a process or set of behaviors. (Of course, I'm making some inferences here, so feel free to correct me.)

The problem is that a lot of research is not based on summarizing sets of empirical facts and describing one's interpretation and synthesis of it. Rather, it can involve entering years-long conversations in which many people have made lots of different arguments, commented on each other's arguments, and responded to objections about those arguments. Recently I was taking notes on a source where author A was arguing against author B about a topic (specifically, interactive media), and I thought that author A's criticisms were wrong based on both logic and on other empirical data not mentioned in the source.

One problem in creating notes here is the notion of atomization - what counts as one idea? Is author B's idea one note, and then author's A's opinion is a 2nd linked to the 1st, and my objection to author A's opinion a 3nd note linked to the 2nd and 1st, and the empirical facts supporting my objection a 4th note linked to the 3rd and 2nd and 1st, etc.?

But also, what are the three layers here? Initially, you might think that author A's criticisms are data and I am interpreting the data. But those criticisms are themselves interpretations of author B's words, which are themselves interpretations of more objective facts about the world. And synthesizing B's specific arguments in a way that highlights the general, unstated conceptual problem underlying the individual problems with B's specific arguments - which would be layer 3 – is actually only a first step towards my explanation of how to reconceive the topic problem in a wholly new way. And that itself is just one step in a large argument about why we should think of this topic in this new way, which also involves providing examples and responding to potential counter arguments.

I suspect that @sfast will say something like "it doesn't really matter how many layers you use as long as it helps you produce articles", and I agree. But it's very hard to tell at the beginning of the process how to set up the ZK so that it will eventually lead to more a productive workflow than my current process (which is far from ideal, hence my looking into this).

I may be one of the few people in this community dealing with this type of writing and research, so if so, I apologize if it is not relevant to others. But since I have seen this three layer post frequently mentioned elsewhere in the forum, I'm interested in others' views on the topic. Are others using their ZK to take notes and write about ongoing interpretations and conversations? If so, how are you structuring your work?

«1

• Hi @cobblepot these are very good questions and provide me with an opportunity to work through my own thinking on the method by @Sascha

I am in somewhat the same situation as you, mainly humanities studies, being a philosopher. However, for some of the things I do, I rely a lot on empirical studies from cognitive linguistics, evolutionary biology, social psychology and neuroscience.
For the purpose of my reply I will take the three-layer structure approach is unproblematic for the latter part of my work.

As for the philosophical part it is quite the way you have described it above with many authors and voices adding to a subject’s interpretation, leading to some more canon interpretations of subjects (like the philosophy of mind’s treatment of cognition in the traditions of representationalism and computationalism, which in the last 30 years is challenged more and more by newer traditions, or if you will Kuhnian research paradigms).

In my setup I have tried aiming at broadly two ways of using the three-structure approach as an intellectual tool for sense-making of these debates.

The first of these ways is the one I consider more or less staple philosophy (in a ’history of philosophy’ way):

• in layer A is whatever a major (often canonical, say the Aristotle, Kant and Wittgenstein boys) author said on a subject
• in layer B is some part of historical debate or interpretation regarding the subject
• In layer C I put my own synthesis, or thoughts, or argumentation to contribute to the debate

I take it this is mainly unproblematic too, but can easily lead to a less cleanly implemented methodology depending on the subject matter (maybe it is something good old Plato never wrote on, but is a point only recently introduced in debates) or subfield you work in. I have as of yet not run into problems here because I more or less treat layer A to consist of quotes from primary sources with some added pointers by myself, often only when I link those notes up to layer B and C notes (providing context).

The second way I use the three-structure approach is as a heuristic in deconstruct both the writing of contemporary texts from other authors as well as, or more importantly for me at the moment, to reverse engineer my own writings like academic papers but mainly my thesis.

As for contemporary authors' texts I go with layer C being this book or paper, then B being the (current) debates this engages, and A being the historical or primary literature on the subject.
For my own writings the same sort of approach goes, but in the case of my thesis and one or two papers this is complicated slightly because I also use empirical studies to make philosophical arguments with. But in this I follow the above.

I hope this helps a bit! Feel free to ask of course.

I am a Zettler, ie 'one who zettles'
research: pragmatism, 4e cognitive science, metaphor | you can't be neutral on a moving train

• @cobblepot said:
In a previous post, @Sascha wrote that evidence captured in ZK notes can be on three layers, but I am wondering whether this is really a useful general principle (as it's presented) or is something that is actually more specific to a more limited subset of certain types of knowledge work. The layers are:

1. Data description and patterns.
2. Interpretation of descriptions and patterns.
3. Synthesis of patterns, descriptions and interpretation.

Although this division makes sense in the abstract, I have had trouble seeing how these layers could be implemented with my workflow, which is primarily writing about academic theory in the humanities. @ctietze Christian's comment on Sasha's post partly highlights the issue:

Sascha’s post and separation of layers reminds me of my struggle and I think that I could’ve used this approach for non-empirical reading of Kant as well: (1) What does Kant say, quoted? (2) What do I think does this mean? Why is it interesting? How does secondary literature interpret the part? (3) … well, there’s not much practical application of Pure Reason, besides beginning to think differently about the universe and everything of course.

Christian's 2nd layer actually includes a number of different things - his interpretation, others' interpretations, and maybe his analysis of their positions. And the third layer he cites need not be practical application, but could be a synthesis of one's interpretation of Kant's arguments, or of others' interpretations, etc.

My concern is that the three layers proposed are appropriate for communicating summaries and first-level interpretations of objective data/facts. A very typical case would be the type of writing that it seems to me that Sasha is doing, which is summarizing scientific findings to support a particular recommendation of a process or set of behaviors. (Of course, I'm making some inferences here, so feel free to correct me.)

The problem is that a lot of research is not based on summarizing sets of empirical facts and describing one's interpretation and synthesis of it. Rather, it can involve entering years-long conversations in which many people have made lots of different arguments, commented on each other's arguments, and responded to objections about those arguments. Recently I was taking notes on a source where author A was arguing against author B about a topic (specifically, interactive media), and I thought that author A's criticisms were wrong based on both logic and on other empirical data not mentioned in the source.

One problem in creating notes here is the notion of atomization - what counts as one idea? Is author B's idea one note, and then author's A's opinion is a 2nd linked to the 1st, and my objection to author A's opinion a 3nd note linked to the 2nd and 1st, and the empirical facts supporting my objection a 4th note linked to the 3rd and 2nd and 1st, etc.?

But also, what are the three layers here? Initially, you might think that author A's criticisms are data and I am interpreting the data. But those criticisms are themselves interpretations of author B's words, which are themselves interpretations of more objective facts about the world. And synthesizing B's specific arguments in a way that highlights the general, unstated conceptual problem underlying the individual problems with B's specific arguments - which would be layer 3 – is actually only a first step towards my explanation of how to reconceive the topic problem in a wholly new way. And that itself is just one step in a large argument about why we should think of this topic in this new way, which also involves providing examples and responding to potential counter arguments.

I suspect that @Sascha will say something like "it doesn't really matter how many layers you use as long as it helps you produce articles", and I agree. But it's very hard to tell at the beginning of the process how to set up the ZK so that it will eventually lead to more a productive workflow than my current process (which is far from ideal, hence my looking into this).

I may be one of the few people in this community dealing with this type of writing and research, so if so, I apologize if it is not relevant to others. But since I have seen this three layer post frequently mentioned elsewhere in the forum, I'm interested in others' views on the topic. Are others using their ZK to take notes and write about ongoing interpretations and conversations? If so, how are you structuring your work?

I divide the three layers regarding philosophy as such:

The first layer is for statements and logical structures. Or: very careful analysis on the primary sources.

For Nietzsche, I have one Zettel for each aphorism he wrote. Each Zettel of this layer only deals with description like comparisons to other aphorisms on the basis of same words used or grammatical traits and similar things. I often make a statement on the exact meaning which is not so clean as one might want it in the first layer. But most of the time is stick to writing quite unchangable things like the overarching idea the aphorism is connected to.

I use a method Luhmann described as generalisation and re-application as opposed to mere relating. For example: One of Nietzsche models individual development. One, quite primitive, first layer comment on an aphorism would be: "This is a characterisation of the general model of development."

The second layer would have a Zettel like "Nietzsche on Individual Development". There I'd develop a position refering to the Zettel of the first layer, only occasionally to the original work.

The third layer would be something like a general department of my Zettelkasten for individual development using Nietzsches Position as a tool.

I am a Zettler

• I’ll add that to my list too Sascha, thanks.
It’s a third approach I hadn’t thought of.
I reckon Nietzsche’s aphorisms especially lend themselves to that.
However, one could apply the same method to for example Wittgenstein’s Philosophical Investigations.

I am a Zettler, ie 'one who zettles'
research: pragmatism, 4e cognitive science, metaphor | you can't be neutral on a moving train

• edited March 2020

@John , thanks for your detailed response. Since we may have some similar obstacles, let me ask you this: does the three-layer framework affect the way you actually write or organize individual notes? Or is it primarily a cognitive tool that provides one way to analyze something you're reading?

@Sascha , I suppose one note per aphorism works for a philosopher whose writing is primarily studied in aphorisms, like Nietzsche or Wittgenstein, but what about the case I mentioned: Writer B has a theory of interactivity based on analysis of several works. Non-aphoristic modern writer A has a developed criticism of writer B. A's argument has several steps and assumptions essential to the criticism. I think that A misunderstands B, and thus the criticism is wrong. I also think that B's original argument is wrong for different reasons. How to structure the Zettels? It doesn't have a straightforward one-way funnel structure of data-->interpretation-->synthesis.

Or to take your example, your comment on the aphorism is "This is a characterisation of the general model of development." This seems so extremely atomistic that I can't imagine building arguments based on this level of Zettel. My first-level Zettel would be something like "A describes B as saying X, but B actually says 'Y' which is better understood as a claim that Z." Would you have that idea split into 3 or 4 individual notes and only put them together in a structured note?

I'll also comment that I'm concerned that I and @Sascha are engaged in a bit of nonconstructive debate. I'm saying that I'm trying to write analyses of back-and-forth arguments between multiple people who are discussing specific concrete things that can be described in statements, but also require interpretation, and are making complicated arguments that require comment at different levels, etc. It's true that you can come back and say "Nietzsche is philosophy, so with philosophy all you do is put aphorisms on individual cards..." Sure, Nietzsche is one philosopher, and discussing his work in aphorisms rather than in overarching systems is one kind of philosophical discussion, but it's not the kind of philosophy that I'm doing, and most philosophy isn't about categorizing aphorisms. So I hope it's clear my goal is not to try to disprove @Sascha's theory of knowledge or criticize the approach to his book, but actually to figure out a workflow that can help my writing productivity based on the kinds of things I'm writing about. I appreciate any and all suggestions!

• edited March 2020

I'm catching up on your questions and posts on the forum, @cobblepot, and I am enjoying your contributions. I want to offer few notes on my practice, as they diverge from some of the recommendations on this site, in case they might be helpful to you or others.

The principle of atomicity that is recommended on this site (as I understand it) seems to imply that each zettel should (and can) be distilled into a foundational, stable premise (fact), so that each zettel can be absolutely portable, potentially useful in any situation. But for the kind of work I do (academic, history of literature in America) I have not been able to imagine what such a zettel would look like, or how it would be useful. In my experience, that idea of atomicity is one that I don't find useful in practice.

So instead, I write zettels that are "atomic" not in any epistemological sense (because again, I don't know what would that mean), but only in the sense that they don't go on too long.

I conceive of each zettel instead in practical terms, as a bit of communication that would be "spoken" by me in the direction of my specific audience---scholars of English-language literature and literary history. Taking this approach, each zettel takes the form of a sentence or two, or possibly a loose paragraph, like what might appear in an article or a conference presentation. (The operations involved in transitioning from zettel to article/presentation is another subject entirely. It's messy, but can be quite fun, if it's the kind of thing you're into.)

I accept from the outset that the claim(s) made in each zettel are going to be highly contingent, because all claims in my field are highly contingent. But I support them, just as I would do in an article or presentation, with references to primary or secondary materials.

If I want to support the general claim made in a zettel with reference to a secondary source, I would simply write in that zettel "Writer A says such-and-such." I would not make that "Writer A says..." into a separate zettel, because there is no functional reason to do so. That piece of knowledge is serving its function within the zettel. And that's it.

An example from my work. Here's the title of a zettel from my kasten:

{2,4,a } Geographic discourse in early US served a civic function

I know based on my research that this statement has been considered true and pretty much non-controversial, and I cite some sources in the note to lend some support to that effect. I quote from two secondary sources, both of which say that geography textbooks actively tried to promote senses of national identity in the new nation. I also quote from a geography textbook from the 1790s articulating (arguably) its intent to serve that function.

This is all in one note. Is it "atomic"? For me, yes. It's about 250 words, and it functions as a unit. Maybe we could say that it is functionally atomic. Could I break it up into three different zettel, one for each source? I suppose, but that would only dissipate my archive and make it much more difficult to use. So I'll keep it as one, thank you.

From my research, then, I also know that this isn't the final word on the functions of geographic discourse in the early US. The authors of the secondary sources that I cite have a rather narrow definition of geographic discourse, given that they only look at geography textbooks that were used in schools. They presume that these textbooks are the embodiment of geographic discourse in the early national period. But they are---as scholars tend to be---wrong. Or, rather, just not completely right. They overlook geographic discourse in other forms---like magazines, for instance. It turns out, the way modern scholars understand geographic discourse is fundamentally different (narrow, scientific, institutionally defined) from the way people in the early US understood geographic discourse (diffuse, non-professional, ad hoc).

{2,4,a,1 } Geographic discourse in early US was not exclusive to school textbooks

And it turns out that these non-textbook forms of geographic discourse were not as interested in consolidating national identity as they were in---get this---actually communicating the nature of the world. So, the supposedly non-geographical texts were actually more geographical than the so-called geography books. Well that's a new note:

{2,4,a,2 } Geographic knowledge in early US was communicated in non-textbook media

There is plenty of room for further clarification, examples, justifications, etc, "underneath" each of these zettel. So none of them is the absolutely final word on the assertions they make.

For example, if I wanted to differentiate between the different "civic functions" of geography, I could link from {2,4,a } to a new note, "{2,4,a,3 } Geography's civic functions in early US," and go on to clarify/elaborate with new notes "underneath" that note, with "{2,4,a,3,a }" and so on.

Similarly, if I wanted to talk about the kind of geographic discourse found in magazines, I could link from "{2,4,a,2 } Geographic knowledge in early US non-textbook media" to "{2,4,a,2,a } Geographic discourse in early US magazines," and go on from there.

If I realize that I need to make a conceptual distinction, for example, between "discourse" and "knowledge"---which I absolutely do---I could add a new note somewhere here around here, like "{2,4,a,1,a } Distinction between discourse and knowledge." But it actually turns out I have already articulated that distinction elsewhere in the zettelkasten. So I include a link to "{1,3,a,2 } Distinction between discourse and knowledge" in a relevant note here.

When I go to draft, I read through all the notes, maybe print them out, maybe not, and I rearrange their contents in a way that makes sense for presentation, since the order of discovery is not necessary the order best suited for presentation. I might start with what I see as the most surprising claim---geographic knowledge in early US was communicated in non-textbook media---and then go back to the "beginning" to lead up to that claim. I might make the distinction between discourse and knowledge in the text proper, or in a footnote. Whatever works best in the circumstance.

Annnnnyway, I'm pretty convinced that there is no single zettelkasten method that can be considered generally applicable. I think instead we need to look for and borrow those methods that have been useful to others whose goals we consider similar to ours. Maybe these methods will be applicable to others here, or maybe not. I guess we'll see.

• edited March 2020

I agree with @argonsnorts fully on this point (and also the folgezettel-like implementation). I’m a historian of ancient history and I’ve found it hard to follow all the principles presented on this site to the letter. I’ve had to develop my own heuristics for preserving pieces of primary sources in the zetteln. For what I do, the exact phrasing of the Latin or Greek or Syriac can be essential, but I’m not always sure at the initial time of reading in what ways. It has been helpful for me to preserve quotes so that the combinatory power of the Zettelkasten unlocks the subtle nuances locked away in these ancient passages. If I only summarized in my own words I would lose this advantage. Though it is easy to see how a lazy application of this heuristic would simply lead to collector’s fallacy (and sometimes does).

• @argonsnorts , your specific examples here were incredibly helpful. I went back and caught up on all of your and @pseudoevagrius 's previous contributions to the site!

The principle of atomicity that is recommended on this site (as I understand it) seems to imply that each zettel should (and can) be distilled into a foundational, stable premise (fact), so that each zettel can be absolutely portable, potentially useful in any situation. But for the kind of work I do (academic, history of literature in America) I have not been able to imagine what such a zettel would look like, or how it would be useful. In my experience, that idea of atomicity is one that I don't find useful in practice.
So instead, I write zettels that are "atomic" not in any epistemological sense (because again, I don't know what would that mean), but only in the sense that they don't go on too long.

We seem to be on the same page here that the notion of zettels as atomic facts or individual statements is a non-starter. However, I'm currently thinking of zettels as being atomic not just in terms of length, but in the sense that they are points that are independent enough that I can imagine them being included in papers about different topics. Do you tend to work on one writing project at a time or several?

Also, I'm not sure if I should just continue here or start a new thread, but I want to ask about your use of a Luhmann-style hierarchical index rather than arbitrary ID numbers. Can you describe your thinking in doing this? Have you ever felt inclined to reorganize this overall structure?

To use a specific example, let's say that you have a long sequence under a general category {1} of theorizing differences between expert and popular books. Later, you realize that you need to expand this category to include expert and popular magazines, and that most (but not all) of the subcategories of {1} are also relevant to magazines. Do you make magazines a topic {2} and recreate/link all of the subcategories to similar subcategories of {1}? Or do you revise the top categories of {1}?

@pseudoevagrius, I also saw you describing some sort of both/and use of these involving YAML. I wasn't able to quite follow the discussion (some of your comments are about Mac-only software, which I don't use) but I'd love comments from you (and anyone else) on the following type of problem:

If I have a topic with many overlapping distinctions I find myself concerned about how to relate the distinctions together in ways that allow me to usefully connect ideas between them depending on the particular topic I'm writing about. For example:

writing domain: author (creation) / audience (reception)
expertise level: expert / popular
publishing format: book / magazine
topic of writing: film / theater / novel
mode of writing: review or evaluation of a work / neutral discussion of a work

My goal is to codify knowledge about these related domains such that I can write about some subset of these domains in a way that answers a research question. And I imagine wanting to draw from these notes for articles about several related research questions. So I might want to write an article about differences between film reviews in magazines aimed at experts versus popular magazines. Or, I may want to write about how popular books on theatrical works differ from popular books about films. Or, I may want to write about how popular art in the early 1900s (including works of film, theater, and novels) was evaluated differently by experts as compared to popular-press journalists (regardless of where the evaluations were eventually published). The very reason I started looking into ZK methods was due to problems in organizing notes and sources in ways that I could use across projects such as these.

Of course I agree that any ZK workflow must be tailored to the individual. But, can anyone suggest ways that folgezettel-style note structures, YAML, or other approaches might help me deal with these issues?

• edited March 2020

Honestly, I started using Luhmann-style numbers simply because using arbitary numbers (UIDs) just wasn't working for me---after two years of trying. Part of this was that I really did not like having to create a lot of organizational meta-notes (so-called "structure notes"), in order to keep my archive structured. That felt like housekeeping, and man, I really didn't want to do that. Because I didn't want to think about my archive, I just wanted to think in it, by adding to it bit by bit as I read and researched. As I've said elsewhere, the structure notes I made just ended up being undifferentiated lists of notes, without easily legigble internal structures. It is difficult (if not impossible) to see the difference between [202001231200] and [202003231400], unless you organize them in an outline format, either through tabs or by using some kind of outline numbering (ie, I, a, i). But, I thought, if I were going to go through the trouble of organizing my notes that way anyway, why not incorporate that organizing procedure, and the organizational marker itself, into the creation of the note, by assigning it a Luhmann number? Why separate the note and from its "address"?

Someone might object: "But wouldn't that prevent your ability to place that note anywhere in the archive! Isn't this method all about preserving flexibility and infinite possibilities?" To which I would say two things: first, I don't think pinning the note's initial location to the note itself is all that limiting, but I guess I'll take my chances, and second, imposed limitations can actually be quite useful in driving creative work, as limitations force one to make decisions and move on, building up and out rather than continuously rearranging what you've already got.

In switching to Luhmann-numbering, I found, to my absolute delight, that I no longer had to think about the structure of the archive---ever. The archive became self-structuring. Sweet relief! No more meta-notes! No more structure notes! No more going back and trying to group notes together into structures that may or may not be meaningful later. Rather, structures got created one, two, three notes at a time, as individual notes were "placed" within the existing structure. Also, as my previous example illustrates, if I wanted to connect two notes that had seemingly distant addresses, like {2,4,a,1,a} and {1,3,a,2}, I could simply link from one to the next, and suddenly they became neighbors, just like when using arbitrary UID numbers.

I will say, the fact that I found structure notes unhelpful was probably a result of the way I work. I typically don't have a clear idea of what project I am working on until I have already done a good deal of reading, researching, gathering, re-thinking, etc. So, I couldn't create a useful "structure note" for a project that didn't exist yet. What would I even call it? Also, I couldn't impose any organizing framework on notes within the structure note, through semantic headings/groupings for example, because I wouldn't know what groupings would end up being important. Trying to do so was putting the cart before the horse---or maybe, putting the horse inside the cart? I don't know, but I like that image... Anyway, by using non-semantic "headings"---Luhmann numbers---structures emerged as I added each note. And, importantly, any note could potentially become a so-called "top level" note, with branchings of notes "underneath" it. Again, the archive becomes self-structuring.

You ask, @cobblepot, about how to deal with later realizations, and the impulse to go back and reorganize "top categories." First, I would say: try not to think about categories and subcategories, or even about hierarchies of notes at all. You're not trying to create a database or a wiki, where higher level notes are more general, followed by more specific subordinate subcategories and notes.

Before I illustrate, I will say that I do follow Luhmann in creating sort of general categories at the "highest level"---"{2} World-Documentation," "{3} Literature"---but this is mostly for convenience, giving me a place to start putting notes in general areas I know are important for my long-term work.

But the notes "under" those headings---{2,1}, {2,1,a,1}, {2,2}, {2,2,a,1}---are all effectively on the same "level." Note {2,1} isn't a category or a heading---it is simply an observation that I had that fit better in the "World-Documentation" group than in the "Literature" group. It is titled "Early documenters of America lacked critical awareness of the objectivity-subjectivity divide." That is not a category or subcategory. ANd it is hardly the most important thought or the most general note. It's simply where I started. And a note that I write after might very well end up being more "general." For example, {2,1,b,1} is titled "Distinction between documenter and geographer." This note is not subordinate in any meaningful (categorial or hierarchical) sense to the notes "above" it, those with shorter addresses {2,1,b} or {2,1}. It is simply one of the notes I created afterwards, each following from a thought in the one before it. It is entirely possible that a note with a long address {2,3,a,1,a} has no obvious relation its supposed "top level" category, "World-Documentation." It simply followed from thoughts that followed from thoughts that did have some relation to that category. It is related by relation. And if I end up talking about a topic that seems related to another "top level" category, like literature, I can simply link to whatever note in the "{3} Literature" pile seems related, assuming there is one.

To address your specific example, then, I would say that once you realize that there are similarities between magazines and books, you can just add a single note that states that realization, and then branch off from there to examples from magazines you've found. There's no need to go back and revise previous notes, because those previous notes are not categories that govern or predicate what goes on "underneath" them. They are just notes that (might) lead to other notes.

So, presuming you've got a "top-level" note titled something like "{1} Expert and popular discourse," you might then have a note "{1,1} There are many differences between expert and popular books." Then you might make {1,1,a} {1,1,b} and {1,1,c} as separate examples of those differences. After your magazine realization, you could link from {1,1} to a new note: "{1,2} Some expert/popular distinctions found in books are also found in magazines." You could then have "{1,2,a} Differences between expert and popular magazines." You could simply list those differences in that note, or if they are meaningful enough, you could make a new note for each---{1,2,a,1}, {1,2,a,2}, etc. But only if it's useful. You might then have note "{1,2,b}" that is an observation on important distinctions between books and magazines, along the lines of the general expert/popular distinction you are observing. This note might seem to be a more "top-level" note than the others, in the sense that it is more general, being an observation on the broader distinction that you are working through. And it's "address"---{1,2,b}---may not directly signal its importance in your thinking, becuase it seems to be "lower," than {1,2}. But it is often our later thoughts that are the most significant, as they synthesize our earlier fragmentary gatherings. A shorter address does not mean more a general or more important note. You're not building a database, you are recording and building relations between your thoughts and observations.

Phew. I blacked out. What happened.

• edited March 2020

@argonsnorts said:
Part of this was that I really did not like having to create a lot of organizational meta-notes (so-called "structure notes"), in order to keep my archive structured. That felt like housekeeping, and man, I really didn't want to do that.

[...]

In switching to Luhmann-numbering, I found, to my absolute delight, that I no longer had to think about the structure of the archive---ever. The archive became self-structuring.

[...]

Also, I couldn't impose any organizing framework on notes within the structure note, through semantic headings/groupings for example, because I wouldn't know what groupings would end up being important.

[...]

[The] note is not subordinate in any meaningful (categorial or hierarchical) sense to the notes "above" it, those with shorter addresses {2,1,b} or {2,1}. It is simply one of the notes I created afterwards, each following from a thought in the one before it. It is entirely possible that a note with a long address {2,3,a,1,a} has no obvious relation its supposed "top level" category, "World-Documentation." It simply followed from thoughts that followed from thoughts that did have some relation to that category. It is related by relation.

[...]>

Fabulous @argonsnorts, @cobblepot, and @pseudoevagrius, and thank you, especially @argonsnorts for the thorough explication and articulation of an issue that's nagged at me for a year or more now. And thanks too for teasing out the implications of each note's relation to its neighbour. I find myself falling into (clearly long-habituated) patterns of wiki-style thinking when I think through the relationship between one note and the next. It's far too easy to fall into the trap of semantic/thematic hierarchies.

I too have laboured over the messy and increasingly unproductive task of creating structure notes to try and capture related/emerging ideas, but for the reasons you so artfully demonstrate, it always feel like I'm lapsing back into rigid categorical thinking rather than allowing the notes to 'think' their own way into new intellectual territory.

Many thanks for demonstrating, once again, why I love this place.

Where in the title/UID string to do put your Luhmann-style numbers?

Started ZK 4.2018. "The path is at your feet, see? Now carry on."

• @argonsnorts can you offer some screenshots of how this looks?

• For what I mean by Luhmann numbering and for the compelling argument that convinced me to give it a try, see: https://sociologica.unibo.it/article/view/8350/8270#fnref12

Maybe one of my biggest misconceptions about the zettelkasten was that it was a drafting tool rather than a thinking tool. I think the suggestion that the zettelkasten eases drafting is true and also a good way to get people in the door (certainly hooked me!), but drafting is a process that happens outside the zettelkasten. In Luhmann's words, drafting is communication with the zettelkasten, and this communication is reciprocal. It doesn't just move from zetteln to drafts. Sometimes my drafts themselves serve as maps to the raw materials in the zettelkasten. And on and on the cycle.

Structure notes as outlines are useful, but these are too draft-specific for following emerging patterns under many many layers of sedimented notes. The ability to see emergent patterns is where the zettelkasten shines. This is the "serendipity" or what I've called the "combinatory power" of this method. I want my zettelkasten to be poetic, not mimetic. I want my zettelkasten to produce new insights, not just reproduce the world outside of it (it's texts, articles, and categories). My use of structure notes, however, tended in the direction of creating hierarchical lists, like comprehensive bibliographies. This approach muted the subtle and more interesting relations between my notes by emphasizing what is essentially the relationships of other author's notes. I don't think that structure notes inherently cause this issue, but my usage of them kept running up against this problem in principle- particularly when I was tracking ideas without a specific draft in mind.

The only problem is that outlining the relationships between notes is an important part of thinking through them. If you are not zooming out from the atoms to see the patterns then you are losing something. But the patterns are not just there either. You have to be the one to trace them out (through note links, outlines, folgezettel, or whatever other method you follow). This is where I see the value of the Luhmann numbering (and to a lesser extent my folgechains), in addition to all of what @argonsnorts has already presented. The nesting is not hierarchical or categorical so much as it is relative to: a) what you have already noted and b) what you are noting now.

The magic is in the kind of relationships this numbering can denote beyond unique id’s (which I still use as well).
They also denote depths in notes and topics. This ability to show depth, or strata, in notes I think is crucial as answer to what @cobblepot is asking.

Different audiences are operating on different depths. Their questions are being asked of different ranges of issues with different specificities. If your zettelkasten is already suggesting to you these strata, it will be easier for you to discriminate what paths to take depending upon how deep or how broad you need to go.

title:
zettel id:
tags:
folgezettel:
conceptual:

I use YAML headers as a way of visually distinguishing metadata from the rest of the note, and I use textexpander to easily replicate the header each time. Is it a little clunky to use metadata? Probably. But here's why I do it.

My yaml header is just a simple text header that give me two benefits:

1. It allows me to be flexible with how I use the zettelkasten.
The yaml stores titles (multiple), zettel ids (multiple), tags (when necessary), links to other zetteln, and other categories/tools that I am experimenting with. I'm allowed to play with titles and different organizing principles because the yaml preserves older titles and principles in the search.

For example, in the past, I have experimented with an idea called "folgechains" (my own spin on folgezettel). I've spent some time with this tool and it isn't generating the kind of results that I was expecting (too draft specific - this was when I was using zettelkasten as a drafting tool!). Now I'm trying to employ a Luhmann # system like what @argonsnorts is using. I don't feel a fear of losing anything and I don't feel a need to completely restart my zettelkasten because all the relations are preserved. Now I can just pick up where I left off and make new trails with them.

1. It gives me a sense of dynamic note development.

My notes do not enter the zettelkasten fully independent and fully developed. I let their relation to other notes and the demands of outside circumstances tell me whether I should put more time into a note and exactly how much. The Yaml gives me a sense to track some of this development. I want to progressively fill each section. Except for tags: I want to progressively remove tags.

I've been struggling quite a bit with the exact problem you are raising @cobblepot. How do we are articulate what is in our zettelkasten, and how do we best organize the zettelkasten to make that articulation? I would characterize this problem as the difference between giving a lecture to undergrad students on a subject versus writing an article on a subject for my field.

Theory over practice. Here is what I did today. Maybe it will change tomorrow. I need to write a lecture for students. I assembled a network of related notes on the day's reading (the opening of Augustine's Confessions) by searching for “Augustine”. This resulted in a collection of reading notes on the particular portion of text and branches of concepts in philosophy and theology going on and on. Too much! I used the app scapple to visually display these notes and their relationships so I could see everything together. Pen and paper would work fine too. What's important is I can see connections between ideas that I haven't made yet. Further, making it visual helps me more easily filter out what would be more salient for the students.

I then string together an outline from this network (which would require another post. nothing is magic), fill in my zettel text, edit and streamline for the lecture. Now I have a lecture draft.

It was important for me to have a zettelkasten with conceptual relations that are not draft specific. My Augustine notes were all within relative reach. A lot of these notes go into the deep depths of Late Antiquity, far afield from the scope of my current class (translation theory, embyrology, phenomenology, divergent religious groups, etc.). But with Luhmann numbering (or something like it) I was able to easily trace these notes to the surface and see how those surfaces connected with other material we had discussed this semester and even modern issues that might ground the stakes of our class session in what matters to the students.

I would have to do much different grounding for an article on Augustine in my field. I could either make a separate structure note to track that, or those relations could be tracked in the same notes already. Their distance in the zettelkasten suggesting their topical distance and topical depth.

What do I do with the lecture draft? This is a question I have. Especially since a lecture is something broader than another kinds of drafts. It is designed to help understanding. Seems like it would belong in the zettelkasten (despite its size). I'm thinking I might put the lecture in as a note under the Augustine readings and linked to the concepts.

• @pseudoevagrius questions for you.

When you see a list of folgechains and see something like?

{3142,1}
{3142,1,a}
{3142,1,a,1}
{3142,2},
{3142,2,a}
{3142,2,a,1}

Describe what you see and what they mean to you.

Or show a shortlist of some of your folgechains and share what you know just by seeing the sorted list.

The more deliberate assignment of a Zettel's identifier provides additional value and information because you can see those Zettels are connected? Whereas with just a plan uid, 202003231743, you could not?

Is that the heart of the discussion?

• edited March 2020

@MikeBraddock said: The more deliberate assignment of a Zettel's identifier provides additional value and information because you can see those Zettels are connected? Whereas with just a plan uid, 202003231743, you could not?

If you don't mind me inserting myself here, Mike, I've been playing around with a Luhmann-style numbering system this afternoon and I can say already that one of the benefits is that I'm being forced to think much more pointedly about the relationship between notes. I've opined elsewhere on this board about my struggle with linking notes. As I did this brief exercise, I realized the rather slipshod and amorphous relationality I've established between the notes in my ZK, relations whose inner logic has now been blown away on the winds of time. I'm left to second-guess my own (evidently clear-at-one-time) decision to link Note A with Note B and so on. Sometimes the connection is clear to me still, but most often, it's not. I've also discovered I have a tendency to create 'link loops', where my chains of connected notes end up linking back to a note near the 'head' of the chain.

A Luhmann style numbering system, it seems to me, not only guards against this redundancy, but also preserves the fidelity of the original thinking that aligns particular notes with one another, thus lessening the need for connecting notes.

Additionally, as @argonsnorts so eloquently describes upthread, I've been confronted with the thorny problem of atomicity, which as he indicates, is one of context and contingency given my own immersion in research areas with similarities to his.

Phil

Started ZK 4.2018. "The path is at your feet, see? Now carry on."

• @Phil thank you for jumping in.

I hope the discussion continues it has been eye-opening to me.
Also the distinction between drafting and thinking, the pull of one interfering with the other.

I have had the feeling that using identifiers more deliberately had underutilized potential.

These folgechains, Luhmann-style numbering system, are very much like IP addresses for your surfing your thinking. Jump in anywhere and catch the wave.

This seems so fundamentally powerful and worked so well for Luhmann how have we gotten so far away from using this as a mainstay in our toolkit?

What do we get from UID's that we can't get from folgechains and more?
Am I missing something?

You definitely right about the eloquence.
There has been a steady flow here on this topic from many.

• First, thanks for a wonderfully in-depth analysis of your workflow. There are ideas worth stealing here for sure.

@pseudoevagrius said:
The magic is in the kind of relationships this numbering can denote beyond unique id’s (which I still use as well).
They also denote depths in notes and topics. This ability to show depth, or strata, in notes I think is crucial as answer to what @cobblepot is asking.

Not sure I see it this way. If a note has a number, say 6,3,e,1 or a UID 202003231443 What does each tell me?

The 6,3,e,1 number tells me that it is the note on the 6th subject in my zettelkasten, 3rd referenced note, further referenced to the e note, and finally the 1st reference to the e note. This only tells me this note relative position in the chain of notes started with 6... I still have to read and study the notes links to get a feel for its value. Also, I have found notes that are buried deep in one chain of notes to become more relevant in a different note chain. How could this be conveyed with an identifying number that shows the note to buried deep in the Zettekasten? How is this not hierarchical or categorical? Isn't the note put in a hierarchical position under the category "6,3,e"?

The UID 202003231443 does say too much more except it does age the note and whether a note is old or young could provide information as to its value.

@pseudoevagrius said:
For example, in the past, I have experimented with an idea called "folgechains" (my own spin on folgezettel).

@pseudoevagrius said:
2. It gives me a sense of dynamic note development.

My notes do not enter the zettelkasten fully independent and fully developed. I let their relation to other notes and the demands of outside circumstances tell me whether I should put more time into a note and exactly how much. The Yaml gives me the sense to track some of this development. I want to progressively fill each section. Except for tags: I want to progressively remove tags.

Wonderful. I look for anything that will help with the sense of note development. I love the idea of progressively work adding to the YAML. I'd never before considered progressively removing tags, but that is intriguing.

Thank you.

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

• I talk about my folgechains/folgezettel here in the forum: https://forum.zettelkasten.de/discussion/comment/3151#Comment_3151
and here, originally (and haphazardly) on twitter:

(with images).

These are not hierarchical because the numbers do not designate topics for me. The focus is note relationships. 1,1 follows 1. This does not mean that 1 is ontologically prior. I just happened to note 1 first 1,1 might have a lot more to say (or be a better note), but if I judge that it is elaborating on 1, then I make it 1,1. So there are just articulations and elaborations. 1 and 2 might even be related, but 2 is not elaborating on 1. If I just use links I could see how 1, 2, and 3 could all be related. This system however tells me a little bit more about their relationships by their proximities. Some of the distance is incidental, but the more I put in, the more decisions I make, the more patterns emerge. Ultimately, though, most of this is communicated in the writing. Numbers and note titles can only leverage so much.

My goal is to flatten my search in the zettelkasten (ie without layers of obscuring structure notes), but I still want to see topography. A topographical map actually works as a great meatphor here for what I mean by "depths."

@will I think your point about notes that cross sequences is critical. I think the beauty of the sequence is that it reveals these irregularities. When I see them, I know that either something is wrong in my framing or something interesting is going on here... or both. I take these as important starting places for digging in and beginning the writing/drafting process. Writing begins with an undecidable problem or aporia that we need to work through. That's the interesting work to be done.

Sometimes I'll make a new sequence/chain out of that point. I can do this by moving it, by copying it and renaming it, or by slightly re-articulating it for that context.

• edited March 2020

@MikeBraddock There isn't really anything to take a screenshot of, except for maybe my Index, which is just a list of notes like what is show in this post.

I also show there my file-naming scheme, @Phil --- UID Luhmann-number Title.txt, eg:
202002290921 {2,4,a } Geographic discourse in early US seems to have served primarily a civic rather than geographical function.txt

NB: in order for notes to auto-sort properly when using this KM macro, it is necessary to include commas between each character, and insert a space before the closing bracket. I don't know why, but, practice over principle, I guess. ¯\_(ツ)_/¯

Thanks for articulating your practice, @pseudoevagrius .

drafting is a process that happens outside the zettelkasten

This is such an important point for my practice too. And you illustrate this well in your example of mining the same archive to produce texts for two very different audiences: lectures for undergrads and articles for peers. The difference comes down to selecting notes at the appropriate level of complexity to suit your audience. *chef's kiss*

I am currently in the process of porting some of my lectures back into my archive, so that they might provide some grounding for articles/chapters. I anticipate that organizing them using Luhmann numbers will reveal the gaps I jumped over---topics and depths that undergrads didn't need to worry about---and thereby show what areas I can expand on and seek more grounding in, for my own work. I'll let you know.

As for what to do with lecture drafts afterwards? I don't know. For now I will keep doing what I've always done, organizing them by class in a file structure outside of my archive, in the same way that I keep article/chapter drafts separate. But who knows. I'll keep thinking about it.

I'm interested in continuing to hear from you @pseudoevagrius @phil @cobblepot and others, if you find the Luhmann-numbering helpful. Maybe we should split this off into another thread? Thoughts, mods @ctietze @Sascha?

• edited March 2020

This morning I took a step back and recalled that @Sascha pondered this very discussion in 2015 in his post, No, Luhmann was not about Folgezettel.

There Sascha says:

Luhmann, the godfather of the Zettelkasten Method himself, wrote that it is not important where you store a Zettel as long as you can reference it from every other point to the Zettelkasten. ¹

The possibility to create a direct reference, for example as a link, reduces the importance of the Zettel coming next in the sequence. The technique Folgezettel creates value from the position of a Zettel in the archive. But the technique of creating a link reduces the value of the position of a Zettel.

Looks like a paradox, doesn’t it?

1. Niklas Luhmann (1993): Kommunikation mit Zettelkästen, in: Universität als Milieu, Bielefeld: Haux. [↩]

He goes on to say:

Folgezettel are a technique to realize a principle. How you realize a principle, that is which technique you use, does not make any difference from a bird’s eye perspective. Of course they appear to be different down on the ground. Like a punch is different from a choke while both follow the principles of combat.

To have a Luhmann-esque Zettelkasten, you don’t have to follow his techniques. Just follow the principles he used and you’ll be fine. In fact, you’ll be better off because you don’t clutter your life through imitating other people’s scenarios.

And then:

Our solution

We have one and only one technique of direct and intended connection: We have links. We just insert a “§” with the ID of the Zettel we want to link to and there you have the link.

• If I want to see all the children of a particular note (read: Folgezettel) I just search for its ID with a “§”-prefix.
• If I want a Zettel to be an addition or a continuation of a Zettel I link to it and mark it as a continuation.

That’s it. Not additional features needed. No additional technique, either. It’s simply the application of the principle of connectivity.

I guess “§” was before The Archive, before [[ ]] style links, which we have now.

Losing the appearance of the relationships at the ground level is what folks are trying to regain?

It occurred to me that visualizing the relationships between the zettels as Christian does in the Epstein book processing videos would be a good solution for recapturing the ground-level view.

Hopefully, this mapping will be a feature of The Archive someday.

Until then @argonsnorts method of combining UIDs and Luhmann styled numbering and or "folgechains" seems reasonable if you want or need it.

In nearly every case of my own confusion or gap in understanding, I eventually find the answer somewhere here in the forums are the blog posts.

I think I was caught up in the paradox. I feel much better now.

Sascha blog post is well wrth a reread.

• @pseudoevagrius said:
These are not hierarchical because the numbers do not designate topics for me. The focus is note relationships. 1,1 follows 1. This does not mean that 1 is ontologically prior. I just happened to note 1 first 1,1 might have a lot more to say (or be a better note), but if I judge that it is elaborating on 1, then I make it 1,1. So there are just articulations and elaborations. 1 and 2 might even be related, but 2 is not elaborating on 1. If I just use links I could see how 1, 2, and 3 could all be related.

I appreciate your explanation. I feel like I'm getting glimpses of what your point at. Key for me is that if note "X" is an elaboration on note "Y" it is labeled "Y, X". It is so labeled for no other reason than it is an elaboration. It is placed behind the prior note in chronological order.

So far so good. I too want all my notes that elaborate on each other to be grouped together so they will surface together in a search.

This system however tells me a little bit more about their relationships by their proximities. Some of the distance is incidental, but the more I put in, the more decisions I make, the more patterns emerge.

Help me see the difference between a sort by date of UIDs and this? Maybe it has to do with the fact that you would always put a new note at the end of the sequence. A sequence might be 1,2,3,4 and a new note might elaborate on note 1,2, becoming note 1,2,4. But how would get this note to surface in a search? Maybe searching on 1,2 to bring up all notes that elaborate on 1,2.

Sound like an abstraction by using 1,2,3,4 instead of links and tags.

Ultimately, though, most of this is communicated in the writing. Numbers and note titles can only leverage so much.

Yes, I completely agree "most of this is communicated in the writing".

I hope this banter has been taken in the spirit of my wanting to understand and learn for those who have ventured where I haven't yet.

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

• To me UUIDs only communicate their uniqueness in time. Just because I created one note at a time stamp close to another, it does not imply to me that these two notes are conceptually, thematically, or semantically related.

For example, while feeding the Roman Empire section of my zkn, my wife might mention something about quiche and I’ll want to create a new note about quiche real quick and place it in the zkn. That temporal relation between the note is not a relation I’m too concerned in preserving.

This is a hypothetical example, but it represents the ad hoc and flexible way in which I communicate with my zkn.

I still use UUIDs to link but this is an early stage of their relationship to my zkn.

Another hypothetical:
This note makes me think of note x, y, and z so I link them.

But what is the direction of this link? Is the note an elaboration of x, y, and z? Or is note y an elaboration of it? Maybe z isn’t directly relevant but has a distant analogy I want to make.

Further is the notes relation to x, y, and z equivalent?

To tackle this issue when it comes time to draft, where should I begin? What direction should I go?
You don’t need this system to make those decisions, but as you write they are happening all at once. When you employ this numbering system you are making more of these decisions before the time of writing or outlining.

If you have certain heuristics for note placement this might not be a problem for you.

What I am trying to solve is a problem with how to communicate with the zkn from the outside. How to find a starting point. Somebody may ask me, could you give a lecture on Christianity in the Roman Empire tomorrow? When I enter search terms I want to have the results already demonstrate some of my thinking and tell me where to start. If it just returns a higher network of notes, it’s only a little better than a book shelf of cross references, and I’ll have a lot of work to do to wrangle it into an outline for the lecture.

• As helpful as the above posts are, I'm still trying to understand the thinking behind the different ID systems.

@argonsnorts: Although the post describing your workflow is fascinating, I was confused by what seemed to me to be some mixed messages within them. Please help me understand your thinking.

You celebrated Luhmann-style numbering as providing an organizing structure with essentially no effort on your part, freeing you from having to manually create structure notes. You also said, "And, importantly, any note could potentially become a so-called "top level" note, with branchings of notes "underneath" it. Again, the archive becomes self-structuring." Yet, later in your note, you advise against thinking of these Luhmann-style IDs as hierarchical, saying that notes with longer IDs are not necessarily conceptually subordinate to those with shorter IDs, and that notes with shorter IDs are not more general or more important than ones with longer IDs. I'm having trouble reconciling the notion that Luhmann IDs generate or reveal a useful structure but at the same time do not indicate any hierarchy. What is the useful thing I learn by seeing a flat "structure," even with links? Especially since you say that notes with subordinate Luhmann IDs need not even be related in content to ones with superordinate Luhmann IDs?

You also criticized time/date ID numbers as uninformative, presumably, in contrast to Luhmann-style IDs which I take it you think are informative. But what do you know from Luhmann IDs that date/time IDs don't tell you, other than the note's very broad top-level category?

@argonsnorts also wrote:

I am currently in the process of porting some of my lectures back into my archive, so that they might provide some grounding for articles/chapters. I anticipate that organizing them using Luhmann numbers will reveal the gaps I jumped over---topics and depths that undergrads didn't need to worry about---and thereby show what areas I can expand on and seek more grounding in, for my own work.

How does organizing notes according to Luhmann numbers reveal gaps if the numbers are not conceptually hierarchical or do not aim to collectively, exhaustively describe a topic?

@pseudoevagrius wrote:

These are not hierarchical because the numbers do not designate topics for me. The focus is note relationships. 1,1 follows 1. This does not mean that 1 is ontologically prior. I just happened to note 1 first 1,1 might have a lot more to say (or be a better note), but if I judge that it is elaborating on 1, then I make it 1,1. So there are just articulations and elaborations. 1 and 2 might even be related, but 2 is not elaborating on 1. If I just use links I could see how 1, 2, and 3 could all be related. This system however tells me a little bit more about their relationships by their proximities. Some of the distance is incidental, but the more I put in, the more decisions I make, the more patterns emerge.

To me this is a similar sort of paradox. If the numbers do not designate topics, and a preceding note is not necessarily ontologically prior to a subsequent note, then what does it mean to say that the subsequent note is an elaboration of the preceding note? It cannot simply mean immediately prior in time, correct? And how do looking at the note connections reveal a structure?

• @cobblepot said:
@Sascha , I suppose one note per aphorism works for a philosopher whose writing is primarily studied in aphorisms, like Nietzsche or Wittgenstein, but what about the case I mentioned: Writer B has a theory of interactivity based on analysis of several works. Non-aphoristic modern writer A has a developed criticism of writer B. A's argument has several steps and assumptions essential to the criticism. I think that A misunderstands B, and thus the criticism is wrong. I also think that B's original argument is wrong for different reasons. How to structure the Zettels? It doesn't have a straightforward one-way funnel structure of data-->interpretation-->synthesis.

Or to take your example, your comment on the aphorism is "This is a characterisation of the general model of development." This seems so extremely atomistic that I can't imagine building arguments based on this level of Zettel. My first-level Zettel would be something like "A describes B as saying X, but B actually says 'Y' which is better understood as a claim that Z." Would you have that idea split into 3 or 4 individual notes and only put them together in a structured note?

It depends. If you take it to the extreme it would mean that one Zettel = one file = one statement. But that is not appropriate when you really adhere the concept of capturing a thought.

For something that is more like contempory philosophy I'd have a Zettel for every position and list all the arguments there unless cluster of arguments get to big and then I'd create a seperate Zettel with a link.

The first layer would be the basic arguments. The Gettier Problem would be on one Zettel. The second layer would be all the direct solutions to the Gettier Problem. The third layer would be everything adressing the adressing of the Gettier Problem.

I think it will be clearer what I mean when I can make time and space for more practical examples and lay them out in actual blogposts. At this moment, many of our (and many other) difficulties stem from the theoretical nature of discussing principals with next to no actual examples.

The three layers for example are used (I argue: always present because of their analytical) in all of the fields I am working in that are

• analytical philosophy
• empirical research in health and fitness
• ficition writing
• literature analysis
• aphoristic style analysis like Nietzsche or religious texts (I argue that Zaratustra and most of his other work is religious not philosophical)
• psychological profiling (for my work)
• world building (for my writings)

So, I am actually doing it in all the fields you are questioning the usefulness. You can still argue that the three layers are not right for you. Then I'll say that you can refuse to accept the law of excluded middle but you will never act like you really do.

I am a Zettler

• @cobblepot The suggestion that an elaboration note necessarily follows the note it elaborates on in time assumes that I engage with my zkn chronologically and process every note fully the first time, but this is not the case.

Example. I was trying to better express the concept of andreia in Greek texts on Christian asceticism. This new problem led me down a path to older zetteln on 4 Maccabees I made years ago. I marked these notes as elaborations on notes I made yesterday. These notes are not chronological, or are they exactly topical either (4 Maccabees is a Jewish Second Temple text that precedes the ones that originally raised my question). The numbers more show my own trails through the zkn than they communicate something ontological about the note.

Another person might have organized these notes a completely different way. And maybe that’s the benefit of the kind of work historians do.

Can UUID links do that? Sure. They just do it differently. I’ve just found that as my zkn has increased in size these links become less helpful because there are just so many of them.

Some of us design zkn with particular projects in mind, and so maybe this approach seems perplexing. I need my zkn to help me with problems I haven’t anticipated yet. It’s already thinking partially about questions I haven’t asked yet. It has a life of its own.

A lot of this is making abstract what is really a preference. I think the key thing to ask is: when you search for a subject (when you are looking for a starting point in your zkn to address a new question) how do you do that? When you enter “Nietzsche” what do you see in the search pane? This approach is a way to find entry points like this without having to be constantly updating structure notes. Of course, you are updating the Luhmann numbers so what’s different? You aren’t having to use other notes to do it. So there are less layers. It is flat, but with bumps.

• I would also note that the decision to give a Luhmann number is more akin to a paper zkn. In a paper zkn you ask: does this note x go before or after note y. This is the question you ask with digital Luhmann numbers in practice.
You are not asking whether this note belongs to this collection of notes you have assembled (structure note). You are asking whether this note x relates to note y. Of course if I mark y as an elaboration of x it implies other relations to notes w and on up the Luhmann chain but this is a different sort of relationship than a structure note and a different sort of question in practice.

Maybe if you make structure notes of structure notes you accomplish something similar.

My problem with how I used structure notes was that I was always asking how note y related to the title/topic of the structure note more than I was asking how y related to individual notes in that structure note.

• I think I'm coming around to seeing Luhmann's numbering system's value. It seems a tool for creating relationships between notes like tags, links, structure notes, UID's, and visual tools like @ctietze's Ruby script.

I stole the seed of an idea from @pseudoevagrius about adding to my YAML section in my notes that is a place to add a quick phrase answering the question "In what context was this note created?" I think this would help with future searching.

@pseudoevagrius said:
When you enter “Nietzsche” what do you see in the search pane? This approach is a way to find entry points like this without having to be constantly updating structure notes. Of course, you are updating the Luhmann numbers so what’s different? You aren’t having to use other notes to do it. So there are less layers. It is flat, but with bumps.

I rarely just search on a single word. Search is your friend and using a more sophisticated query focuses the result without relying on structure notes. It is trivial to capture a list of notes from the search query and make them into a table of contents or the first draft of a paper or class lecture.

@pseudoevagrius said:
I would also note that the decision to give a Luhmann number is more akin to a paper zkn. In a paper zkn you ask: does this note x go before or after note y. This is the question you ask with digital Luhmann numbers in practice.

Ok, I think I get it. No need to update the Luhmann number. The new note just gets the next number in line from the old note the is most relevant, period.

How does this scale? Could you find the appropriate place to number a title when you have 100 notes, 1000 notes, 10,000 notes? How did Luhmann do this with so many notes? Maybe this where his keyword index guided him. Remember he didn't have the advantage we have now - the magical Omni Bar search function.

@pseudoevagrius said:
Maybe if you make structure notes of structure notes you accomplish something similar.

My problem with how I used structure notes was that I was always asking how note y related to the title/topic of the structure note more than I was asking how y related to individual notes in that structure note.

I use my structure notes differently than @pseudoevagrius. I don't update my structure notes. By that I mean they are updated as they are worked on but not indefinitely, continuously. I update them while reading a text or while a class is in session but they have a definite finish. I create them around a specific reading, lecture, video, or class. Not broad categories like philosophy or physics, that begs constant updating. Sure it would be a huge task to continuously update structure notes if you have a structure note for every root of Luhmann's numbering system.

Below are terms ripe for tags as there might be 10's or 100's of notes related to each one. The more notes in each tag cloud the more refined your search must be to surface the wanted notes. I say "wanted note" but really what comes up is a serendipity of notes with links outward.

#andreia
#Greek_texts
#Christian
#asceticism
`

The processing of the "4 Maccabees text" sounds ripe for a structure note. As it is a document that would be appropriately linked to from many notes and have more the one or two tags.

Topics of interest to my thinking are tagged appropriately, so all the books I've read on Zen come to my attention when I search for #zen #structure_note. Then as I explore more I may refine my search to #zen #hiaku.

Please, I don't know if my biases have blinded me. I do feel like I'm getting the spirit of using Luhmann's numbering system. I do see where it can help find cool entry points into my zkn.
I don't find structure notes problematic, on the contrary creating them is exciting and thrilling. I look for readings, lectures, videos that grab my interest and stimulate my zettelkasten itch then dive into creating a structure note and linked notes.

Please take my comments in the spirit of friendship and because I desire to learn. Funny aside, just thinking about how to respond, the writing of a comment, teaches me so much. Thanks for the opportunity to share.

Writing is thinking!

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

• @Will yes, I agree whole-heartedly about why it is so fun to engage with discussion here. If I'm terse I hope it doesn't come off in a negative tone. I just try to brief because I don't want to fall into the trap of spending more time writing about my zettelkasten than actually using it!

And I'm also doing the same thing. I'm less trying to persuade you than I am trying to explore the own limits of my own method, and similarly (I suspect) you are engaging with mine (et. al) only to the extent that it might be useful for you.

My zettelkasten is currently around 12,000 interconnected notes. So that's the kind of size I'm dealing with for reference.

As for one word text searches. I don't try to employ this as my main way of searching. However, as a teacher and an academic, I'm used to getting questions from left field that I wasn't already working on. I want my zettelkasten to be able to help me answer these questions to.

Tags are helpful like linking UUIDs but they only do so much work. In principle, to me, they are still like structure notes.
Clicking on a tag summons a list of notes in the search pane, but how are those notes interrelated? I can of course go through everyone and follow the links, and that would work. With the Luhmann system though I have a more instant overview of these notes are related in the search pane and from there I can click on notes to and follow the other links and paths they each contain.

Why these topographies are important is that I have two jobs going in opposite directions that I need my zettelkasten to help me facilitate: explaining generalities (lecture example) and producing specific particularities (thesis example).

In the case of tags, using tags to process my 4 Maccabees note might be helpful for the lecture example. In this way I'm archiving those notes into well-wrought discourses that others have worked out. But if I want to produce my own theses about texts like these I cannot rely on other's categories and definitions of these texts. This would only reproduce their results. I need to acknowledge theirs, acknowledge others, and produce my own.
This work requires much more theoretical granularity than would be salient for a general lecture.
Still, I need to do both!

I really do think Schmitt gives a much clearer explanation of Luhmann's numbering system than I can give: https://sociologica.unibo.it/article/view/8350/8270#fnref12

Schmitt acknowledges the limitations of Luhmann's paper zettelkasten but also shows the benefits of doing it that way anyway.

• @pseudoevagrius said
What I am trying to solve is a problem with how to communicate with the zkn from the outside. How to find a starting point. Somebody may ask me, could you give a lecture on Christianity in the Roman Empire tomorrow? When I enter search terms I want to have the results already demonstrate some of my thinking and tell me where to start. If it just returns a higher network of notes, it’s only a little better than a book shelf of cross references, and I’ll have a lot of work to do to wrangle it into an outline for the lecture.

I am trying to understand the use case and have been experimenting more with this using @argonsnorts notation scheme and found that I can create search links to connect/associate a Zettel to an entire folgechain or parts of a folgechain by ending the search string with a comma (,) and leaving off the right braces:

[[{2,]]

Am I heading in the direction you have been describing?

I am also seeing how you can associate Zettels together without necessarily having to use a structured note.

Including them in the titles also provides visual cues, another valuable part of the toolkit.

Thank you for helping me see this.
It is a joy to hang out on the forums.

• In some ways, you may understand it better than I do! I think this way of marking folgechains could be usefully repurposed at any party of the chain use {2 until that becomes too bloated and not as useful. Then use other parts of {2 instead.

The way you use search is really game changing and I haven’t fully wrapped my head around that yet.

I’ve also been reluctant to say too much about what I meant by folgechains because I think they were unsuccessful. They were trying to make the zkn into a writing machine rather than a thinking machine. Too linear. The folgechain was trying to be a draft (maybe too early).

But your implementation is making me rethink that a little. Will report back after some appropriate experimentation.

• @pseudoevagrius said:
The way you use search is really game changing and I haven’t fully wrapped my head around that yet.

Me neither, I keep finding new surprises.
Example
I'm revising a paper I wrote on the High Trail in the Hells Canyon Wilderness. It includes a wolf encounter. Today while thinking about this conversation thread and the relationships that could surface via a search and how the naming convention (Luhmann's numbering system) might help, I stumbled onto this gem.

Let me explain.
The gif starts with my default note list.
Then I type into the Omni Bar, the broad search term "wolf."
Then I see that there is a structure note as indicated by the "★"
A structure note is a point of entry into initial exploration, but to go deeper and discover the serendipity is the plan.
Then I exclude the structure note and its associate backlinks (which had been already thoroughly mined for their ideas) by typing in the Omni Bar "NOT high" and am left with notes with a wolf 'context' that are unrelated to the High Trail.
Low and behold!! I stumbled onto a super relevant note! Tale of two wolves from ** precisely one year ago today** (03-28-2019 - 6:28 AM). The Tibetans would call this an auspicious event. We're off to revise my paper integrating this surprising meme.

@pseudoevagrius said:
I’ve also been reluctant to say too much about what I meant by folgechains because I think they were unsuccessful. They were trying to make the zkn into a writing machine rather than a thinking machine.

You point to an idea that I too question. I am finding that sometimes I use The Archive as a writing machine and sometimes a thinking machine. Will this lead to confusion? Maybe, maybe not. For now, I leaning towards NOT. I think that this is the best use of my zkn. Writing is thinking. What feeds one feeds the other.

@pseudoevagrius said:
But your implementation is making me rethink that a little.

This conversation/exploration has me rethink a little too.

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

• Good questions, @cobblepot . I'll see if I can clarify some things, but just know that I'm basing my description on practice, not on prior theorization. Whatever this system does for me, it may do much more or much less for another practitioner. Have you tried using Luhmann IDs vs. time-date IDs?

@cobblepot said:
I'm having trouble reconciling the notion that Luhmann IDs generate or reveal a useful structure but at the same time do not indicate any hierarchy. What is the useful thing I learn by seeing a flat "structure," even with links? Especially since you say that notes with subordinate Luhmann IDs need not even be related in content to ones with superordinate Luhmann IDs?

The question seems to presume that all useful (interesting, valuable) structures are hierarchical. Or, put another way, that non-hierarchical structures are not useful. Using Luhmann IDs allows notes to be related in different, sometimes undefined or even undefinable ways. It does not prioritize hierarchical relations. It allows for them, but it does not require them.

By assigning a Luhmann ID to a new note, all you are doing is establishing a relationship---some relationship, any kind of relationship---between that note and another note, which has itself, in turn, an established relationship (of some kind) with another note. So when adding a new note, you are simply extending a web of relations. Those relations are not necessarily hierarchical in nature (general-to-specific, or even specific-to-general, as the case may be). They certainly can be hierarchical, one way or the other. But all the system demands is for some relationship to be established between a new note and an existing note. The relationship does not have to be defined or articulated. It can be, but it does not have to be. They are simply relations, and that's it.

You also criticized time/date ID numbers as uninformative, presumably, in contrast to Luhmann-style IDs which I take it you think are informative. But what do you know from Luhmann IDs that date/time IDs don't tell you, other than the note's very broad top-level category?

Luhmann-style IDs are not particularly informative on their own, no. They are most informative to the extent that they index a relationship (some relationship) to an "adjacent" note. The nature of that relationship is not articulated by the Luhmann ID.

There could be some dissonance, given that the alphanumeric sequencing of Luhmann IDs seems to suggest hierarchy (a is prior to b, 1 is prior to 2), since that's a common way of organizing. But there is hopefully an appreciable difference between hierarchy and sequence.