Zettelkasten Forum


Principle of Atomicity - File-based Zettelkasten vs. Roam Research

I've been using a zettelkasten based on @rene Sublime Zettel for more than 2 years. It has over 2000 notes in it covering diverse topics from neuroscience to cooking to clinical medicine. It's worked well, and as it enlarges, I get more value from it. But when new and interesting tools come along, I like to check them out. I know Roam Research has gotten a lot of attention around here and around similar corners of the web. So I started playing with it a few weeks ago.

I think one of the most striking and important differences between Roam and Markdown-based systems is that in Roam, the atomicity is even more granular. The fundamental unit is the bullet or line of text. Every bullet has a unique ID that allows it to be linked to anything else.

In Markdown-based ZKs, like mine, or those that use the Archive, the zettel is a markdown file, which usually has several lines of text in it (and sometimes a lot of text). This is helpful in that it consolidates related sentences but it doesn't allow the level of granularity that Roam does. Sometimes I really do only want to link to a single sentence and not an entire note. Roam allows that. A markdown-ZK could allow this if one made zettel-files line by line (one line per zettel), but this becomes really tedious, mostly because you have to explicitly title each zettel file.

In contrast, in Roam, you can just open up a Daily Note and start writing. There is no cognitive overhead in having to title a zettel. You can basically do a brain dump and if you didn't feel like it, you could leave that brain dump right there and never move it or properly place those lines into their own pages (groups of related bullets).

But, then, as @sfast talks about in his latest post, you can build infinite hierarchical "Structure Notes" or whatever you want to call them elsewhere.

So all this to say, I think one of the most enabling features of Roam is the size of the atom can flexibly go from a single line up to whole blocks of hierarchical bullets, without the cognitive overhead of having to create distinctly titled pages.

The other things like the back-linking and all that are cool too, but I think the fact that bullets are the unit of operation is perhaps the most powerful difference. I don't want to jump onto the Roam Wagon, since it's proprietary and I like my system, but I'm wondering if I'm losing something by sticking to the file-based ZK.


Here is a real example.

I was reading some medical articles on an autoantibody found in a type of neurological disease. I typed that single line into my Daily Note (which is basically an inbox)

You'll also see that I was reading @sfast 's post, and I dropped that in there too, very easily, without any concern that those two pieces of information should be separated.

Then, later, I went to my 'Structure Note' on FGFR3 Antibodies and referenced that single line from my Daily Note

That's the beauty of Zettelkasten in general, the modularity of knowledge which allows you to combine and recombine infinitely. But in this case, the level of granularity I needed was better served by Roam than my file-based ZK.

But in this case, it was very easy for me to reference that single line.

If I were to do this in my Sublime Text (file-based) system, how would I have done this?

  • Should I have made a zettel for that single fact "15% of patients...." ?
  • Or make a larger note that collected multiple facts from the original article, and then referenced that whole note?
    • If I did this, it'd work out, but I'd also be linking to some extra information that I don't need there.
    • Moreover, what's nice about the the way Roam does it is that you see the actual content, not your title. So it fits right into the outline, rather than being something like "2020042601 - Findings from Tholance et al. 2019"

Comments

  • edited April 28

    Ya it’s weird. I find it fascinating but between the cost and always online part of it, I just couldn’t get myself into it. If down the line it provides a crazy level of utility that matches my personal workflow then I’d reconsider.

    I found that the concept resonated with me a lot because I traditionally have “thought in a bullet pointed manner”. When thinking about a concept on my iPad I create bullet point notes to make sense of models. Before learning about Roam I also tried implementing a collapsible markdown system in a github wiki. Trying to figure out how one would do scaling complexity, that allows you to engage material on the level of prior knowledge that suits you.

    I’d just make sure to be careful about getting caught up in a collecting and connecting obsession. Luhmann was crazy productive with a paper based one, lol. Which to me is a reminder to not overly focus on collecting information at the cost of improving my ability to think.

  • @achamess has actually brought up three related but distinct issues: transclusion, content atomaticity, and note-creation friction.

    First:

    @achamess said:
    I think one of the most striking and important differences between Roam and Markdown-based systems is that in Roam, the atomicity is even more granular. The fundamental unit is the bullet or line of text. Every bullet has a unique ID that allows it to be linked to anything else.

    Roam calls these bullets 'blocks', and they only 'lines' if they are just one sentence long. If they are longer, they word wrap. As you mentioned, each has an individual ID, which is similar to the way most people do ZK notes - one per file, each with a UID.

    One difference, however, between Rome references and most ZK software references is that Rome can show the content of the link to note instead of the link itself. Apparently, this feature is called transclusion and I think it has the potential to strongly affect how people write with their ZK. With transclusion, you can have an outline whose content includes both notes and the transitional material needed for an actual essay draft, but is actually readable straight through like an essay. With links only, you have to navigate through the notes one-by-one to "read" the outlined content or export it into a different system.

    I don't know of other software that allows transclusion - everyone, please post if do. I think there may be a way to do it in VS Code or emacs/org-mode, but I haven't found it yet.

    Second, atomaticity:

    In Markdown-based ZKs, like mine, or those that use the Archive, the zettel is a markdown file, which usually has several lines of text in it (and sometimes a lot of text). This is helpful in that it consolidates related sentences but it doesn't allow the level of granularity that Roam does. Sometimes I really do only want to link to a single sentence and not an entire note. Roam allows that. A markdown-ZK could allow this if one made zettel-files line by line (one line per zettel), but this becomes really tedious, mostly because you have to explicitly title each zettel file.

    You are highlighting what you think of as the usefulness of granularity, but if your notes are really only one sentence long, I think there is another danger lurking in your future: radically atomic claims, such as facts, have very little context. If you are just aiming to capture facts, that is not really a problem, but if you are aiming to capture claims or ideas within the single sentence that you right now and that is understandable to you now will become useless when you come back to it in five years or 10 years and are a separate person. Remember the principle that ZK notes should make sense to someone else reading them, not just yourself at the time you create them.

    There is some sort of advantage in being able to compile facts in different structures, but truthfully, you really don't need a ZK note to capture a fact. Facts can be stored in reference materials, looked up later online, or put into a spaced repetition system to be memorized (and people in the forum have discussed ways to include these systems in ZK systems). I do not see much value in making individual facts alone into ZK notes to be linked in various places. Of course, if you have an observation or an idea that relies on that fact, the fact might be mentioned in the context of a more important point.

    Third, note-creation friction:

    In contrast, in Roam, you can just open up a Daily Note and start writing. There is no cognitive overhead in having to title a zettel. You can basically do a brain dump and if you didn't feel like it, you could leave that brain dump right there and never move it or properly place those lines into their own pages (groups of related bullets).
    So all this to say, I think one of the most enabling features of Roam is the size of the atom can flexibly go from a single line up to whole blocks of hierarchical bullets, without the cognitive overhead of having to create distinctly titled pages.

    I think Roam's seeming advantage here can be replicated with an 'inbox' or 'capture' file that you process later. It's true that you can't link to the notes in that file right away, but maybe that's a good thing. Fast capture of ideas is good, but there is a different between an ephemeral note that you capture quickly so you don't forget it and a permanent note that is written out in full and thoughtfully connected to your ZK.

  • @achamess Thanks for the translation of what Roam does to the rest of the Zettelkasten folks! This is the first time I read about Roam and could imagine what happens and how to work with it. Great job!

    Interesting to see how the practice of transclusion is translated to a live preview, too. Zkn3 kind of does the same when you can attach branches multiple times, but I never thought of this feature in terms of transclusion -- I only know this from MultiMarkdown and Wikipedia as a way to affect the rendered output (you specify which other file/page to transclude, and then the HTML output generator will copy & paste the contents inside for you during rendering.)

    Lemme share a hack that might help in some cases where you wished you had the ability to link anything to everything.

    Sascha and I collaborate on a lot of writing projects; be it blog posts, book manuscripts, or the video course script. We found that IDs also work really well as jump-marks. (The stuff programmers use for GOTO; or how table of contents in a webpage make your browser scroll down to the heading.) So an "identifiable" object can be a Zettel, or just any other position in a text. Brought to the extreme, you could add Ids to every paragraph or bullet point, but the noise would be unbearable, I think. It can help to create a finer granulation, though. And if it already has an ID, it might just as well be extracted from the parent list of bullet points and elaborated in its own paragraph -- after all, the links to the ID-as-jump-mark won't break and don't care where the ID is put, as long as it is findable in the Zettelkasten.

    Example:

    # 198711091115 True facts
    
    - True facts
        - ... about the land snail
        - ... about the tarsier (202004280915)
        - ... about bananas
    
    # 200105262349 My amazing Zettel
    
    I like tarsiers and know many true facts about them.[[202004280915]]
    

    With a link to [[202004280915]], you will end up at the true facts about the tarsier list item. When your fact collection grows, and you lose overview in the list, you may want to extract it into its own Zettel and use the same ID:

    # 198711091115 True facts
    
    - True facts
        - ... about the land snail
        - ... about the tarsier  [[202004280915]]
        - ... about bananas
    
    
    # 200105262349 My amazing Zettel
    
    I like tarsiers and know many true facts about them.[[202004280915]]
    
    # 202004280915 True facts about the tarsier
    
    Tarsiers have noses.
    

    This is waaaaaay less nimble than having each bullet point be a link-able object by default, though.

    Author at Zettelkasten.de • https://christiantietze.de/

  • edited April 28

    @achamess, many thanks for starting this thread! I wanted to do the exact same but you did describe it much better than I could have done!

    I agree that Roam's feature of block-addressability is very interesting, and being able to link to individual blocks or lines of text sounds very useful. One core aspect of Zettelkasten-style note-taking is the idea that each thought should be in its own note item (with its own ID so that it can be referenced). In that sense, having even more granularity (down to the line or bullet point level) sounds great.

    However, during the last weeks I've been thinking whether this is actually helpful. Questions:

    1. How does this scale in the long run?

    Roam-style block-addressability allows for very quick note-taking, as you say it can get close to a "brain dump". This is great and has its uses, esp. as an "inbox". But what about the long-term aspect? My fear is that this feature will trick me into getting sloppy/lazy. I.e., I would likely use this for most of my note-taking (since it's so much faster and easier, and time is always rare). So I would take less proper notes. But will this cause a big mess in the long term?

    2. Is omitting the note title a good idea?

    Taking proper notes also means that I have to think more about what I want to state with this note. E.g., being forced to write a title for the note means that I have to come up with a short summary (or core statement) of what this note is about. IMO, performing this little extra step is truly helpful in the long-run, since it helps to increase the clarity of my notes, and sharpens my senses.

    This is even more true for the act of "transcribing" the note's thought with your own words in the comments of the note.

    3. Can I easily expand a line/block later on?

    In a proper Zettel note, I'd not only add tags but many other elements, like my own comments, any quoted text, cross-links, reference information, or links to external sources. The Zettel note keeps all these elements together, and it makes it clear to me that these belong to this very Zettel note. I think that's a good thing. If the need arises, I'm sure that I could make Roam-style lines/bullet points into proper Zettel notes. But would I take the time to do so, later on?

    4. Can I easily reuse lines/blocks in external applications?

    Personally, I think it's crucial that a zettel note is self-contained / self-standing. I.e., each note should contain all necessary information so that I could copy this note somewhere else, or export it into another app w/o the loss of any relevant information. This means, e.g. for literature notes, I want to include the article's citekey and reference information, and ideally also a web link or DOI for the article. Additionally, I want to include a link to the article's full text (including the page info for the quoted text, or even its exact annotation info).

    Besides self-contained Zettel notes being truly helpful when working with your Zettel notes in external applications (like DEVONthink or Scrivener, etc), this also makes each Zettel note more robust. I firmly believe that this helps my notes to survive in the long run.

    Example

    As an example and to illustrate the above, here's a screenshot of a Zettel note that I've taken for your "literature note" example above (from the "Tholance 2019" paper):

    Please ignore the formatting-specific peculiarities in that screenshot and focus just on the elements of this note. This note hopefully illustrates my points above and should give an example of what I mean with "self-contained notes".

    Post edited by msteffens on
  • Perhaps, you need an outliner? I'd rather not store thoughts in a messy manner like this. But in the process of figuring things out, this is very acceptable.

    Sometimes, I use OutlineEdit to scratch some ideas out. After they have settled I decide what to do with them. Either store them as outlines in my Zettelkasten, write a blogpost/article (which gets processed back to my Zettelkasten after it is finished) from it or create a Zettel.

    Version 2 of The Archive will include features that heavily support such a desk version.

    Or did I misunderstood you?

    If you use SublimeText there should be some plugins to support outlining.

    I am a Zettler

  • @achamess said:
    I think one of the most striking and important differences between Roam and Markdown-based systems is that in Roam, the atomicity is even more granular. The fundamental unit is the bullet or line of text. Every bullet has a unique ID that allows it to be linked to anything else.

    Maybe I misunderstand. I can do this with The Archive software. I do this by placing in-line links where I want to jump to for one note to another. I can place in-line links in sentences or before or after bullet points. Maybe Roam makes this easier or prettier but this works as demonstrated in yjr short video video I created.

    Now to make this work I've created a Keyboard Maestro macro to create a link in the form of [[UID]], but the link could be in another form as it just refers to a section of a note, not a separate note. The macro creates the UID and then copies it to the clipboard for pasting at the other end.

    Excuse my stumbling, but here is a short screencast showing this in action. This also works within a note and while the software doesn't "jump" to the link, the link is highlighted by the theme and stands out visually as you see in the video.

    I also thought while editing the video that this technique could be expanded and there is no reason this "pseudo jump" link couldn't refer to 2, 3, or more particular spots within the Zettelkasten. The link itself doesn't have to be a UID but I use a Keyboard Maestro generated UID to avoid collisions and to reduce choice paralysis. Can Roam do that?

    Will Simpson
    I'm a Zettelnant.
    Research: Rationalism, Zen, Non-fiction Creative Writing
    kestrelcreek.com

  • edited April 30

    Thanks for the thoughtful replies here everyone.

    @ctietze - Yes, one could make every line or block addressable with a UID if one chose to, and @Will touches on that. One doesn't even really need separate files. A single text file could be one's entire ZK and all the linking is internal to that one file. It'd be a mess to work with but it's possible.

    @msteffens Hi! You also make good points about titling, and I thought the same. Maybe the title forces me to be judicious about what goes in and also to have to think about what I'm putting in.

    @sfast I do write things in outline format most the time, but outliner software is not what I need (e.g. OmniOutliner) since I can't link easily as one does with a ZK (although, see discussion above about inline links). I don't really like writing in prose for many things anymore. I write in bullets even in my zettels because it makes information retrieval easier. Anything can be an outline as long as there are hierarchies of text, and can be interchanged with prose, like this article from @ctietze shows https://zettelkasten.de/posts/how-i-use-outlines-to-write-any-text/

    @cobblepot Excellent dissection of my post into separate issues. Thanks for putting a name on the concept I was referring to. I didn't know about transclusion.

    Regarding the capturing and using singular facts: It depends on your line of work, but for scholarly work in the sciences, much of our outputs (articles, reviews, grants) relies on the assembly and presentation facts from across sources to make our arguments or interpret data. So sometimes, there is no getting around it. Like in my example above, knowing the prevalence of a diagnostic test outcome in a population is something you just have to know in order to say other things about that disease or test. Leaving those things in their original sources (the article from which it comes) makes it very laborious to do any work since you have to remember where those facts reside and you have to go back to hunt them. Having a system that allows me to collect and group facts into higher level structures is supremely useful and a common use case.

    Post edited by achamess on
  • Suppose for a moment we thought about this discussion like this.

    Copy some text, any text, assign a UID to it, save it somewhere for future access.
    Invoke a script, macro, that presents the lists of available transclusions.
    Select one to insert at your current cursor position with the assigned TCUID.

    Like maybe a homegrown custom approach for doing transclusion like this:

    1. Add a transclusion folder to The Archive directory similar to the media folder.
    2. Highlight some text and or copy to the clipboard such as some bullet points.
    3. Invoke a KBM macro that assigns a new TCUID creates a file in the transclusion folder with that TCUID and the contents get written to that file with the text from the highlighted text or clipboard.
    4. Replace the highlighted text with the text in the file in the transclusion folder just created by step 3 appended the TCUID link [[TCID-202004301450]].
    5. Create another KBM macro that allows you to select from a list of available transclusions from the transclusion folder that pastes the selected transclusion text with the same TCUID link [[TCID-202004301450]] as above.

    Search Link example (these are comments without the <> so they would render)

    !-- TCID-202004301450 Transclusion text Begins here --

    • Transclusion Hello World. - [[TCID-202004301450]]

    !-- TCID-202004301450 Transclusion text Ends here --

    Tag Link example (these are comments without the <> so they would render)

    !-- TCID-202004301450 Transclusion text Begins here --

    !-- TCID-202004301450 Transclusion text Ends here --

    These are just suggestions feel free to take this and run with it.
    Does anyone think they can build something like this easy enough for everyone to use?

  • Or even grep The Archive for the any available TCUD's and present those. No need for a folder. I wish I had thought of that sooner.

    Really all we need is a common format convention to follow.

    Grep for these

    !-- TCID-202004301450 Transclusion text Begins here --

    Transclusion Hello World. - [[TCID-202004301450]]

    !-- TCID-202004301450 Transclusion text Ends here --

    Present as a list. Select one and insert it.
    I know you folks can whip something like this up in no time at all.

  • Or maybe this:

    - [[TCID-202004301450 Begin]]
    - Transclusion Hello World.
    - [[TCID-202004301450 End]]
    
    - #TCID_202004301450_Begin
    - Transclusion Hello World.
    - #TCID_202004301450_End
    
    1. KBM that wraps the TCUID Begin and End links to the text for transclusion creation.
    2. KBM to then grep The Archive directory for the TCUID's Begin and End Markers, present to the user, the user selects, KBM pastes accordingly. anywhere else in any other zettel at the cursor position.

    No need for a separate transclusion folder. Simply mark the text you want as part of your transclusion system. Basically create them by convention and use KBM to create and then use them.

    Sorry for the progressive comments.

  • [ It‘s probably best to discuss transclusion in its own thread ]

    As far as I understood transclusion, it is about including one Zettel note within another. I.e, there should be no need to explicitly mark some content as content that’s available for transclusion. One should instead be able to include any note within any other note. MultiMarkdown already has a syntax for it:

    https://fletcher.github.io/MultiMarkdown-6/MMD_Users_Guide.html#filetransclusion

    In a similar fashion, I could imagine:

    {{ ZETTEL_ID }}
    

    as a marker to include the content of the note with ID ZETTEL_ID.

  • @msteffens That is a good step to include the complete note in the Zettel, but I would wish, The Archive could allow me to incorporate a line of thought out of one note into another, and therefore into another context. Effortlessly, that could greatly facilitate the connectibility – and would be, obviously like RoamResearch partly shows, a milestone in grinding down even an atom ;-))

  • So I tried a proof of concept prototype for @achamess's example.
    @Martin is this similar to what you might work for you?

    Step 1 - The 15% entry today's inbox:

    Step 2 - Invoke KBM macro and select the line item to include in the FGFR3 Autoantibodies zettel.

    Step 3 - The 15% entry today's inbox included in the FGFR3 Autoantibodies zettel.

    KBM Macro used for this.

    As I was testing this I realize I could also use this to also quickly add previously used bibliographies without having to leave my current zettel.

    KBM Macro used for this.

    Attached is my feeble attempt at these macros. Please experiment. Any improvements, refinements, additional ideas, and comments would be appreciated.

    @Will your demo help me along this path. Thanks.

    (Hopefully, I haven't fat-fingered anything with typos.)

  • An additional thought: The definition of an atom should be orientated towards a unit of knowledge not a unit of knowledge representation. A sentence or a paragraph is not a unit of knowledge. A statement, a thought and similar things are units of knowledge.

    It is knowledge work and not sentence ordering. It is knowledge structure and not paragraph relations. We aim to work on knowledge and its structure. This is the correct level of analysis.

    I am a Zettler

  • Transclusion is a very useful thing because in addition to seeing all the information you want in one place, it also has the advantage that if you edit the original note, it will change for each transclusion.

    TiddlyWiki is a very good example of using transclusion because the entire user interface is literally based on transclusion: https://tiddlywiki.com/#Transclusion

    I can also show through an example:

    • Go to https://tiddlywiki.com/
    • Create a new note ("tiddler") with the plus button in the sidebar
    • Type any text you want
    • Add the tag $:/tags/SideBar in the "tag name" field below the title of the note
    • Save the note with the check mark in the upper right corner
    • A new tab has appeared in the sidebar, which is the note you just created. Try modifying the note and check the sidebar again.

    In practice, an article can be compiled with transclusion, for example, I display notes for the "Editor toolbar" tag:

    <$list filter="[tag[Editor toolbar]]">
    
      <h2><$link/></h2>
    
      <$transclude/>
    
    </$list>
    

    There is a very useful plugin for TiddlyWiki that allows you to view your notes with virtually no clicks:

    https://tobibeer.github.io/tw5-plugins/#preview

    Hover your mouse over one of the links and leave it there for a while.

  • edited May 1

    @sfast said:
    An additional thought: The definition of an atom should be orientated towards a unit of knowledge not a unit of knowledge representation. A sentence or a paragraph is not a unit of knowledge. A statement, a thought and similar things are units of knowledge.

    It is knowledge work and not sentence ordering. It is knowledge structure and not paragraph relations. We aim to work on knowledge and its structure. This is the correct level of analysis.

    Respectfully, this seems arbitrary. What is knowledge, in your view?

    Is it the grammatical structure or length of a text that defines it? Or is it content?

    Let's take the follow:

    Water is comprised of 2 hydrogen atoms and 1 oxygen atom.

    That has the grammatical form of a sentence. It is a statement (proposition) about the world that verifiable and likely agreed upon by the majority of people. It is concise, yes. It has the grammatical form of a sentence. In your view, what would you call this? Information? Data? Knowledge? And would would it qualify as something worthy of being it's own zettel or should it be part of a more encompassing collection of such sentences, maybe a zettel entitled "Water" that has multiple pieces of information about water.

    I'm not trying to be confrontational but I am just curious as to your rationale.

  • (1)

    @sfast,

    I think a thought is no longer than a sentence or two. For example, you also divided your comment into two paragraphs, because although it is about one topic, there is an invisible difference between the two paragraphs and I think that is what determines how long one thought lasts and where the other begins.

    (2)

    @achamess,

    In your view, what would you call this? Information? Data? Knowledge?

    In this sense, I think it is unnecessary to categorize thoughts. Treat what you feel as a separate whole and put it in a separate zettel. If you see that this is a whole that can be interpreted on its own, regardless of the surrounding text, then I think it can be stored separately.

    (3)

    It is my own experience (although I am still a beginner with Zettelkasten) that I am not basically writing zettel longer than a short paragraph. If, however, I began to elaborate on the idea (for example, that water consists of 2 hydrogen and 1 oxygen atoms), I would move the non-directly related sections into separate zettels so that, for example, the zettel of water would be a kind of structure zettel containing the original sentence, but also links to hydrogen and oxygen zettels.

    I marked the sections of this comment with numbers. I would divide this line of thought into such small parts because these parts make sense on their own.

  • @sfast said:
    It is knowledge work and not sentence ordering. It is knowledge structure and not paragraph relations. We aim to work on knowledge and its structure. This is the correct level of analysis.

    YES! Focus on tools, note formating, naming conventions, 'sentence ordering', and 'paragraph relations' is not knowledge, at least not the type of knowledge I want to spend my limited time left developing. This ultimately distracts me from the project of growing and nurturing knowledge that changes me or my world in a positive fashion. (How do the idiosyncrasies of a notes format make a positive impact on my world?)

    Confession
    I am guilty as anyone for being distracted by the new and shiny. I've spent too much time fiddling around the edges and pledge to dig in and try and get 0.001% better each time I encounter my zettelkasten.

    I for one am anxious to get my hands on @sfast's book. I pledge not to distract him from this important work anymore.

    ps - the baseline of reading and writing in complete sentences is a form of knowledge that I was very focused on when I was 3 but not so much today, maybe to my detriment.

    Will Simpson
    I'm a Zettelnant.
    Research: Rationalism, Zen, Non-fiction Creative Writing
    kestrelcreek.com

  • @achamess said:

    @sfast said:
    An additional thought: The definition of an atom should be orientated towards a unit of knowledge not a unit of knowledge representation. A sentence or a paragraph is not a unit of knowledge. A statement, a thought and similar things are units of knowledge.

    It is knowledge work and not sentence ordering. It is knowledge structure and not paragraph relations. We aim to work on knowledge and its structure. This is the correct level of analysis.

    Respectfully, this seems arbitrary. What is knowledge, in your view?

    Is it the grammatical structure or length of a text that defines it? Or is it content?

    Let's take the follow:

    Water is comprised of 2 hydrogen atoms and 1 oxygen atom.

    That has the grammatical form of a sentence. It is a statement (proposition) about the world that verifiable and likely agreed upon by the majority of people. It is concise, yes. It has the grammatical form of a sentence. In your view, what would you call this? Information? Data? Knowledge? And would would it qualify as something worthy of being it's own zettel or should it be part of a more encompassing collection of such sentences, maybe a zettel entitled "Water" that has multiple pieces of information about water.

    I'm not trying to be confrontational but I am just curious as to your rationale.

    Could you please answer @achamess' questions, @sfast? I don't have any knowledge about epistemology/metaphysics, so I don't really understand what you mean. :)

  • @inquisitiv3 I will have to write a blogpost on this topic.

    But I think this will take me very long. I would not satisfied with my (very) rough English answer in a forum reply and figuring this out in English is quite a hassle.

    It is hard to use philosophy without becoming a phrase philister and get to a concept that can properly act as a tool.

    Perhaps, I can hint you in the direction I am going with this: Language is a tool of representation with specific advantages and disadvantages. If you use technical language with models and hard definitions you'll engage more the left brain which tends to view the world from bottom-up. It tries to figure out the whole from the pieces. If you use prose with metaphors and concepts of sufficient familiarity you'll engage more the right brain which tends to view the world from top-down. It to accept the world as it is and views the whole with blurry lines between its pieces.

    If you use this to understand philosophers you'll need to feel the text first, reflect on it, seperate those two layers and then bring them back together. Nietzsche does use, for example, a lot of metaphors and can therefore transport meaning in an organic and alive manner. There is the other layer still present but this is not what Nietzsche is about.

    Both hemispheres have their own way of representing. I assume we agree that both scientific papers on neurology and novels by Hemingway have knowledge in them. But if we use a too technical, prescriptive definition of knowledge you cannot assess what Hemingway wants to tell us. Or even better (or worse) religious texts. Perhaps religious texts are better than Hemingway to show the problem. It is quite dumb or arrogant to assume that people became aware about the problem of Adam and Eves offspring (incest would be necessary) recently. People had sex and avoided incest quite a while ago and new about the problem. You cannot understand what the bible tries to say if you take it literally.

    At the same time, it is obvious that we want the thing that is represented by Nietzsche, the bible, Hemingway, scientific papers etc. That is knowledge that takes certain shapes. Language is one shape. But isn't it knowledge if one ape shows the other how to crack a nut? Without any language they are giving each other (practical) knowledge. The Zettelkasten Method is one way -- the BEST, OBVIOUSLY!!!111 -- of representing what is already represented. If you use text you want to represent a text with you words which already was a representation.

    If you assume a sentence is the atomic unit you are operating on a superficial level: On the level of representation. But the level of knowledge is one level deeper. Practically: If you want to understand what a person is saying you don't focus on the language she uses and how but you want to access what is in her head as acurate as possible.

    Our modern western thinking seems to prioritize the way of the left hemisphere. We live in a world of representation and not just in the world. That is one of the reasons we romanticise living in the moment because we do not want to represent all the time but want to be present. And, I think, one of the reasons why it is quite common to ask how to connect Zettel instead of how to make meaningful connections between bits of knowledge.

    I am a Zettler

  • edited July 24

    I've actually been moving away from the principle of atomic notes. When I think about it, there are three reasons to make a note atomic

    1. Easy understanding. If you limit the size of a note, then you don't have to waste time reading it, looking for the relevant information.
    2. The limited size of a note makes it easy for linking and referencing.
    3. In a physical zettelkasten, limited note size allows you to realistically elaborate on any note or idea through the process of sticking a new note between any two notes.

    The third option in a digital zettelkasten isn't an issue because every note essentially has unlimited expansion. The first two reasons do stand though. Where I've settled is to let a note expand to a max of around 5-10 paragraphs and if I find myself needing to reference any given paragraph in multiple notes than I will excise that section of the note and turn it into an embed.

    Another workaround, depending on how much you link individual sentences, is to do what zettlr does. It allows you to easily generate a time ID anywhere in a document. So if you wanted to link an individual sentence, you can just generate a time ID at the end of a sentence and then use that as the reference point. I do not use the Archive (because Mac only) so I'm not sure if they also do that.

    P.S. also just realized I already posted here at the start lol. Now that I realize this is an old post from April, @achamess I wonder where you stand now. Do you feel the same way or have you moved past Roam?

  • edited July 24

    @sfast said:
    @inquisitiv3 I will have to write a blogpost on this topic.

    But I think this will take me very long. I would not satisfied with my (very) rough English answer in a forum reply and figuring this out in English is quite a hassle.

    It is hard to use philosophy without becoming a phrase philister and get to a concept that can properly act as a tool.

    Perhaps, I can hint you in the direction I am going with this: Language is a tool of representation with specific advantages and disadvantages. If you use technical language with models and hard definitions you'll engage more the left brain which tends to view the world from bottom-up. It tries to figure out the whole from the pieces. If you use prose with metaphors and concepts of sufficient familiarity you'll engage more the right brain which tends to view the world from top-down. It to accept the world as it is and views the whole with blurry lines between its pieces.

    If you use this to understand philosophers you'll need to feel the text first, reflect on it, seperate those two layers and then bring them back together. Nietzsche does use, for example, a lot of metaphors and can therefore transport meaning in an organic and alive manner. There is the other layer still present but this is not what Nietzsche is about.

    Both hemispheres have their own way of representing. I assume we agree that both scientific papers on neurology and novels by Hemingway have knowledge in them. But if we use a too technical, prescriptive definition of knowledge you cannot assess what Hemingway wants to tell us. Or even better (or worse) religious texts. Perhaps religious texts are better than Hemingway to show the problem. It is quite dumb or arrogant to assume that people became aware about the problem of Adam and Eves offspring (incest would be necessary) recently. People had sex and avoided incest quite a while ago and new about the problem. You cannot understand what the bible tries to say if you take it literally.

    At the same time, it is obvious that we want the thing that is represented by Nietzsche, the bible, Hemingway, scientific papers etc. That is knowledge that takes certain shapes. Language is one shape. But isn't it knowledge if one ape shows the other how to crack a nut? Without any language they are giving each other (practical) knowledge. The Zettelkasten Method is one way -- the BEST, OBVIOUSLY!!!111 -- of representing what is already represented. If you use text you want to represent a text with you words which already was a representation.

    If you assume a sentence is the atomic unit you are operating on a superficial level: On the level of representation. But the level of knowledge is one level deeper. Practically: If you want to understand what a person is saying you don't focus on the language she uses and how but you want to access what is in her head as acurate as possible.

    Our modern western thinking seems to prioritize the way of the left hemisphere. We live in a world of representation and not just in the world. That is one of the reasons we romanticise living in the moment because we do not want to represent all the time but want to be present. And, I think, one of the reasons why it is quite common to ask how to connect Zettel instead of how to make meaningful connections between bits of knowledge.

    Thanks for the explaination @sfast! I think that I now understand what you mean with knowledge and knowledge representation. If I've understood you correctly, you're saying that texts can only try to represent knowledge. All knowledge is "metaphysical"1, and to represent it you need to use a tool (demonstration, text, etc.) to describe it.

    I thought of the phenomenological layer2 while reading your post. Is it correct to describe the first layer as were you you try to represent the knowledge in your own words using text?

    EDIT: I'm interested in reading some books on metaphysics and epistemology. Does either of them have a dependency on the other? I'm thinking of where to begin.


    1. I've no idea if I use the term "metaphysical" correctly here. What I mean is something that can't be directly represented in any way. ↩︎

    2. As described in your post Three Layers of Evidence↩︎

  • @Nick your three points are spot on. That's what I understood from the ZK style of note taking but I couldn't put it into words. I made a new zettel citing you verbatim :]

  • @Nick said:
    I've actually been moving away from the principle of atomic notes. When I think about it, there are three reasons to make a note atomic

    1. Easy understanding. If you limit the size of a note, then you don't have to waste time reading it, looking for the relevant information.
    2. The limited size of a note makes it easy for linking and referencing.
    3. In a physical zettelkasten, limited note size allows you to realistically elaborate on any note or idea through the process of sticking a new note between any two notes.

    The third option in a digital zettelkasten isn't an issue because every note essentially has unlimited expansion. The first two reasons do stand though. Where I've settled is to let a note expand to a max of around 5-10 paragraphs and if I find myself needing to reference any given paragraph in multiple notes than I will excise that section of the note and turn it into an embed.

    Another workaround, depending on how much you link individual sentences, is to do what zettlr does. It allows you to easily generate a time ID anywhere in a document. So if you wanted to link an individual sentence, you can just generate a time ID at the end of a sentence and then use that as the reference point. I do not use the Archive (because Mac only) so I'm not sure if they also do that.

    P.S. also just realized I already posted here at the start lol. Now that I realize this is an old post from April, @achamess I wonder where you stand now. Do you feel the same way or have you moved past Roam?

    Glad to see this discussion continues.

    I am not using Roam at the moment because I don't want to get locked into a proprietary piece of software. But the principle of very granular linking still appeals to me. I think it is a distinct advantage of Roam. And I'm seeing some very creative add-ons and development on Twitter for Roam. I think even if people don't use Roam, some of the tinkering and workflows that people are developing will spill over into other PKM solutions like ZK.

  • @achamess because of the ability to create tweet threads I thought it would be a really good method for the creation of a latticework of note sequences, with the pinned note being the index. I experimented with the idea but it didn't work out. The primary problem that I ran into is that twitter doesn't do versioning where it would always give me the ability to edit a note (mainly for the purposes of linking to new notes in old notes).

  • @inquisitiv3 said:
    Thanks for the explaination @sfast! I think that I now understand what you mean with knowledge and knowledge representation. If I've understood you correctly, you're saying that texts can only try to represent knowledge. All knowledge is "metaphysical"[^1], and to represent it you need to use a tool (demonstration, text, etc.) to describe it.

    Yes.

    I thought of the phenomenological layer[^2] while reading your post. Is it correct to describe the first layer as were you you try to represent the knowledge in your own words using text?

    The phenomenological layer is more akin to information. The phenomenon just is.

    EDIT: I'm interested in reading some books on metaphysics and epistemology. Does either of them have a dependency on the other? I'm thinking of where to begin.

    Yes, they do. Metaphysics concerns itself the nature of being. Epistemology concerns itself with our access to this being. Philosophy of language is also thrown into the mix.

    [^1]: I've no idea if I use the term "metaphysical" correctly here. What I mean is something that can't be directly represented in any way.

    [^2]: As described in your post Three Layers of Evidence.

    I am a Zettler

Sign In or Register to comment.