Thursday, December 3, 2015

My Zettelkasten

I have written many times about Luhmann's Zettelkasten (see Luhmann on "Taking Note"). I have also written many times on ConnectedText and how I use it as a Zettelkasten (just click on ConnectedText or Zettelkasten in the word cloud). But I have never explicitly made clear how the approach I take in ConnectedText is related to that of Luhmann's. Here it is.

I believe the most central aspect of Luhmann's Zettelkasten was the radical rejection of hierarchical organization as you might find it today in outliners, for instance where the information is arranged under headings, like so:

   objective idealism
   subjective idealism
etc., etc. 

Instead, Luhmann opted for a non-hierarchical organization by giving every note (Zettel) a unique number that corresponded and indicated the physical place of the note (Zettel) in the boxes in which they were stored. This is in some ways like the call numbers according to which libraries organize their book collections, like the Dewey decimal system or, better, that of faceted classification by Ranganathan.

This approach allowed him both to refer to each of the notes by a fixed number and to find the physical piece of paper by its location in the Zettelkasten. So 1 was followed by 2, 3, 4, etc. His convention allowed him to have have several notes that continue any note. So 1/1 and 1/2 continue 1. Obviously, an electronic version of the Zettelkasten that has no built-in limitation on the length of notes can do without such continuations (but it does not have to do without continuations either).[1] Daniel Luedecke's Zettelkasten sports "Folgezettel" or "continuations." They seem to implement that feature.

Secondly, Luhmann's approach allowed him to branch off any topic that seemed necessary to him. So, a note on "objective idealism" could branch to both "idealism" or "transcendental idealism" or "subjective idealism." Luhmann indicated branches by letters (a, b, c etc.). Combinations of continuations, branches, and continuations of branches could lead to expressions that are just as forbidding as the call numbers of library books. There are no limits to the number of physical objects you can refer to by these scheme. Luhmann's system is a freely expandable collection of interlinked notes. It resembles a hypertext system for storing and modifying information where each page is easily reachable from any notes. It is as well as you could do in a paper-based system, I would say.

How Luhmann actually assigned numbers is not important, as far as I am concerned. It was important for his physical implementation of the Zettelkasten. It is far less important for an electronic version as a database in which every record automatically gets assigned a number already. If you decide to use a non-database version by implementing it in plain text, for instance, you need fixed numbers again. But the exact time and date of when the note was taken might be sufficient (see Christian Tietze's approach for this, i.e. search for him in this blog and follow the links).

I have decided for a database or a personal wiki which is "a freely expandable collection of interlinked ... 'pages,' a hypertext system for storing and modifying information — a database, where each page is easily" reachable from any other page.[2]Furthermore, the notes which are static in a paper system are freely editable.

You reach any note simply by enclosing the name of the note by double brackets, like so "[[objective idealism]]". In other words, you can directly link to any other page directly. And without the interference of numbers, as Tietze's system still seems to oblige you to do, or the interference of keywords or tags, as Luedecke's Zettelkasten and many other note-taking applications require you to do. Let my quote Caulfield again:
At the heart of wiki is a simple idea that names matter. Page names in wiki are not locations. They aren’t a place where a document lives. Names identify ideas, patterns, theories, and data in wiki that can be recombined with other ideas, patterns, theories, and data to make complex meaning not expressible in a normal text.

If you’ve ever had a good wiki experience, you know what this feels like in practice. Groping towards an idea on one page you realize its relation to another page and quickly make a [[Bracketed Link]] or CamelCaseAssociation to pull that idea into your web. But most non-wiki environments frustrate this fluidity. They don’t want to know the name of the page — they want to know its location, which is like asking someone to give up using variables in their code and start addressing memory directly. It can be done, but it is going to kill your flow.
The same holds for having to add reference numbers or key words. They break the flow, slow you down and get between you and the ideas, patterns and theories--or so I would hold.

I have nothing against tags or categories per se. ConnectedText, the personal wiki software I use, allows you to freely assign as many as you want. But they are not the primary way of organizing your stuff. The same could be said of numbering schemes. Nor do I have anything against search. ConnectedText has a very capable search engine. But, again, I would not want to have to have to rely on search alone to navigate a system of 10,000 notes, or more.

In any case, as I have said many times before, I believe a personal wiki, or, more generally, a personal hypertext system best captures the spirit of Luhmann's system because it allows names to "identify ideas, patterns, theories, and data in wiki that can be recombined with other ideas, patterns, theories, and data to make complex meaning not expressible in ... normal text." It is no accident that many people discussing Luhmann's Zettelkasten have characterized it as a precursor to hypertext.

I should perhaps add that I consider my approach in no way as the alleinseligmachende (or exclusively salvatory) approach to note-taking. It may, indeed, be just my highly idiosyncratic way of doing things. So, take my advice (just as the advice of anyone else) cum grano salis.

1. I do believe that if you write more than 500 words, you are usually not going on in the same way on the same subject, but are making different points that deserves a new note.
2. The quote is from Bo Leuf and Ward Cunningham.


Simon said...

You're really talking about linking your ideas together in some form of chain. How do you overcome the parochial nature of only the thought you are looking at in one note with links to others? You may well have 50 notes linked to an idea. Is an index page still valid that will essentially link to every note on a particular idea and do you use such?

MK said...

It's not a chain but a network, and that is the answer. You can easily create an index by searching and/or embedding a search in a page.

Parochialism is always a problem, of course. The only way to overdome it I know is by thinking again, and again ...

zoe said...

My question throughout all these systems is, how do you implement cross-linking as you write, when your list of topics becomes very large? Doesn't it require you to have a mental index of all the possible other entries? Otherwise, how will you know to connect an entry to [[this relationship]] and [[that relationship]] AND [[the other relationship]]?

One thing that I like about the personal knowledge software PiggyDB is that it auto-detects words in your entries that are the titles of other entries. So even if I forget that I have a card called [[dispensationalism]], if I'm writing another card and use the word "dispensationalism," PiggyDB will turn it into a wikilink. Unfortunately, PiggyDB has many limitations that make it less than ideal for other reasons.

But in the end, I'm still mystified as to how to work with many, many entries without having to carry around a complete mental index of all entries to properly interlink each one... Is there something I'm missing here?

MK said...

Autodetection results in over-linking. That's bad in my book, very bad.

ConnectedText (which I use) is smart enough to auto-suggest links that exist. So, if you type "[[" and then "obj", it will suggest "objective idealism". You can either accept it or reject it. In other words, you do not need to "carry around" a complete mental index of all entries.

There are other tools that help with the relation to other entries, like maps. See this post, for instance.

For my view on piggydb, see piggydb.

MK said...

By the way, Luhmann's paper system did have precisely this problem. He had to have in his head a (more or less) complete index. He solved this problem by emphasizing the essential arbitrariness of connections. And I agree that one should not obsess overly much about the one correct connection. There are several, usually.

I also have special pages that have an embedded search for persons, ideas, or theories, like this: "[[$ASK:Piaget|INDEX]]".

They also have special categories, like this "[[$CATEGORY:Person|Psychology]]" But perhaps this is too specific to ConnectedText.

Angry Thinker said...

I fail to see how your type of links are superior than links in e.g. Evernote or OneNote. Taking OneNote as an example, you can link a note to another using a name or a number for that link. So "you can directly link to any other page directly", to use your phrase. In fact, in OneNote you can even use the double square bracket method to link directly to another page. However, I fail to see the advantage of that because you still have to know the name or number of the page you want to put between those square brackets, which is also the case for "normal" manual direct linking.
And as for the mental index about notes, I agree that that can be handled with keywords or categories as they are called with CT. With the non-wiki apps you can also use categories to find notes. So here too, I don't see the advantage of a wiki app.
The only advantage of CT in terms of a Zettelkasten I can see is the ability to produce those note relationship maps.
So, all-in-all I do not get the elation you experience with CT, but I am probably missing something essential.

MK said...

The links in Evernote are very different from those in OneNote. ConnectedText works pretty much like OneNote. (I think I commented on that before.)

"However, I fail to see the advantage of that because you still have to know the name or number of the page you want to put between those square brackets, which is also the case for "normal" manual direct linking." -- I don't understand. Don't know whether this helps: In ConnectedText, you don't really have to know, just typing "[[" brings up a list of all available topics, typing "a" after that narrows the list to all entries that start with "a", etc. (Works only for links internal to ConnectedText.

"I do not get the elation you experience with CT." What can I say?

Angry Thinker said...

2nd para of your last comment: yes, that explains it. Indeed, that is useful & more efficient than in other apps, incl. OneNote.
Might be useful to clarify that in some of your discussions with Christian Tietze because it seems to me he does not realise what opening [[ does & how it helps.

Juan Diego said...

I have been following Christian Tietze’s method (unique ID and title) and I agree with your remarks on how that number may interfere with your work. So I’m now moving all my notes to dokuwiki.

I’m curious about the ways in which you keep the principle of “one thought, one note” in a wiki. It seems that in Tietze’s view, titles are more like commentaries on the content but they can change. Instead, wiki titles may be meant to be shorter and fixed and the topic/article/page should be broader. I might have this idea because the obvious example of wiki use is Wikipedia (which is encyclopedic in nature). TiddlyWiki may be more suited to our note-taking purposes *by design*. I don’t know how much the design of dokuwiki (or Connectedtext) affects its use.

MK said...

Wiki titles (in ConnectedText, at least) can be fairly long. They are not in any way fixed. In fact one of the strength of any true wiki system is that any reference to a "wiki title" gets updated when you change it. So, if you change "xyz" to "tytxtt," all the references to "xyz" get updated to "tytxtt".

That is one of the reasons why you don't have to spend much time thinking about the tiles when you first create a topic.

This is not true of some systems that claim to be be wiki-like. In nVALT, the connections between linked topic gets broken, if you rename the topic. (This is also true of applications that automatically detect whether there is a topic with a certain name (like Devonthink).

Obviously, I like ConnectedText best, but I have a feeling that TiddlyWiki and dokuwiki (with which I have some limited expperience) will work as well.

MK said...

I had not used TiddlyWiki in a while, and I remembered incorrectly. TiddlyWiki does not seem to update the references to a tiddler when its name changes. See also: here.

There was a reason why I found it less useful than ConnectedText. I am not sure either how well it would handle more than 10,000 tiddlers, but I obviously did not try.