Writing Slowly

While studying the cataloguing and indexing systems of the early Twentieth Century I noticed that the index box was originally supposed to be a key to the records held elsewhere. In other words it was like a library catalogue.

The library catalogue doesn’t exist for its own sake. Rather, it’s the key to finding something else - the books stored on the library shelves. From the late 19th Century onwards, any bureaucratic organisation typically stored its records in filing cabinets (Robertson, 2021), usually numbered consecutively (numerus currens) as they arrived or were created. Then alongside these records the index box contained a parallel catalogue entry for each item (Byles, 1911).

Without a comprehensive index, it would be very hard to find anything in the records. Sure, you could find something at random, provided you didn’t care what it was, but without the index, finding a specific item would be like looking for a needle in a haystack. If the records system used running numbers it would be easy to find the most recent documents, and the longest-held, since these would be located at the end and beginning of the records, respectively. But finding anything in between these points, without an index, would be extremely time-consuming.

As I read about this approach to record-keeping, I realised that Niklas Luhmann’s system of personal academic notes, his Zettelkasten, displayed some similar features to this, as well as some crucial differences.

Luhmann was a German sociologist who famously published a massive academic output by relying on his large collection of handwritten notes, his Zettelkasten (note-box). His Zettelkasten was similar to the standard filing system in that it was a box of numbered index cards. But it differed in two important places.

First, on the whole, there were no other records. The index box simply referred to itself. The records and the index of the records were the same thing. (There were of course references to external academic sources, but I mean that Luhmann’s Zettelkasten didn’t refer to a separate location where he kept his notes such as a set of notebooks or another separate filing cabinet; rather, his notes were the records, and his records were the notes.)

Secondly, the numbering system wasn’t exactly consecutive (numerus currens), it was instead associative. In other words, the notes weren’t placed strictly in order of writing, but were arranged instead according to how their contents related to other notes. He started consecutively but then branched off by adding letters and numbers to the notation. For example, he would add a note that related to note 9 by creating note 9a, and so on (this is a slight simplification, but basically sound).

These two features, (‘records=index=records’ and ‘associative-not-consecutive-numbering’) taken together, meant that Luhmann’s Zettelkasten was effectively almost self-indexing. It was an index of itself, and there was hardly any other indexing work, other than adding cards in relevant locations, with a suitable ID number.

True, Luhmann’s second Zettelkasten did also have a keyword index, but this index of 3,200 keywords was quite limited relative to the large number of notes (67,000) it supposedly indexed. And this index didn’t reference every instance of his keywords. On the contrary, it didn’t need to be exhaustive because by means of their ID numbers the notes were arranged in long chains so Luhmann could jump from one relevant note to another, without needing to keep referring back to an index.

According to Luhmann scholar Johannes Schmidt (2018: 58):

“the file’s keyword index makes no claim to providing a complete list of all cards in the collection that refer to a specific term. Rather, Luhmann typically listed only one to four places where the term could be found in the file, the idea being that all other relevant entries in the collection could be quickly identified via the internal system of references described above.”

I really like the idea of a self-indexing system, and it somehow felt familiar, but I couldn’t think where I’d heard of this idea before.

Then I realised it’s been staring me in the face this whole time. One of my chief long-term inspirations is Christopher Alexander’s A Pattern Language (1977), which was also a key influence for Ward Cunningham, creator of the first ever wiki, the Portland Pattern Repository.

A Pattern Language has enjoyed a cult following and has been identified as “one of the most widely read architectural treatises ever published” (Dawes and Ostwald, 2020), though not without criticism (Dawes and Ostwald, 2017). Anyway, here’s what the preface of that book said about the concept of a pattern language:

“A pattern language has the structure of a network. […] The sequence of patterns is both a summary of the language, and at the same time, an index to the patterns.” — ‘Summary of the language’ p. xviii https://patternlanguage.cc/

It’s interesting to compare and contrast these two examples of a pre-Web, analogue hypertext. Both demonstrate significant elements of this self-indexing aspect. What Christopher Alexander said about his pattern language can also be seen in Niklas Luhmann’s Zettelkasten. The sequence, whether of patterns or of notes, is both a summary and at the same time an index of itself.

What I take from all this is the reminder that I don’t need to work too hard at indexing, provided my notes include a few links to other relevant notes.

If every note had just one link, then all the notes would be connected. In contrast, though, a note with no links (an ‘orphan’), will be very hard to find again, except via a full-text search. So I need just enough indexing to be useful, and no more. As Luhmann himself said:

“The decision where to place what in the file can involve a great deal of randomness as long as I add references linking the other options” (Luhmann, 1987, p. 143, quoted in Schmidt, 58)

Well, now it’s time for a confession: my own collection of notes has no keyword index at all.

But I wonder how much of a keyword index would be useful for my own collection of notes. My feeling is that there’s little point in creating a separate keyword index for two main reasons, as folllows.

First, In the digital age we have at our fingertips something Luhmann never had: a full-text search capability. This means that any time I want to find all my notes containing a particular word, I can easily find them almost instantly. No need for an index just to find notes.

Second, it’s useful and efficient to do all my work on my notes in my notes. This means documenting my searches. Let’s say I want to find all my notes relating to a particular word, as in the case above. Rather than just doing the search, I also document it by creating a new note, perhaps named after the keyword I’m searching. The point is that there was a reason I wanted to do this particular search, and If I don’t document it, that particular information is forever lost. In contrast, by documenting my search I create a new note which may prove useful in future and which also acts as a kind of hub for future searches relating to this particular keyword.

This is similar, at least in spirit, to Luhmann’s ‘hub notes’ which Schmidt identifies:

“The cards containing a collection of references are furthermore of interest because they represent so-called “hubs”, i.e., cards that function as nodes that feature an above-average number of links to other cards so that these few cards provide access points to extensive parts of the file.” (Schmidt, 58)

There’s a brief but useful section on hub notes in Chapter 6 of Bob Doto’s book, A System for Writing, which clearly shows how these differ from structure notes. I found this distinction subtle but helpful.

Since I haven’t even got one, I’m not sure about giving advice on indexing, but if I was sure, I’d say this:

Make a keyword index if it pleases you to do so, especially where the keyword doesn’t otherwise appear in your note. But observe over time how much use you gain from your index. The concepts of ‘self-indexing records’ and ‘working on your notes in your notes’ may provide new insights into the value of your index-work.

References

Alexander, Christopher. A pattern language: towns, buildings, construction. Oxford university press, 2018.

Byles, R.B. 1911. The card index system; its principles, uses, operation, and component parts. London, Sir I. Pitman & Sons, Ltd.

Dawes, Michael J., and Michael J. Ostwald. “The mathematical structure of Alexander’s A Pattern Language: An analysis of the role of invariant patterns.” Environment and Planning B: Urban Analytics and City Science 47, no. 1 (2020): 7-24.

Dawes, Michael J., and Michael J. Ostwald. “Christopher Alexander’s A Pattern Language: analysing, mapping and classifying the critical response.” City, Territory and Architecture 4, no. 1 (2017): 17.

Doto, Bob. A System for Writing. New Old Traditions, 2024.

Luhmann, N. Biographie, Attitüden, Zettelkasten. In N. Luhmann, Archimedes und wir. Interviews, edited by D. Baecker & G. Stanitzek (pp. 125–155). Berlin: Merve, 1987.

Robertson, Craig. The filing cabinet: A vertical history of information. U of Minnesota Press, 2021.

Schmidt, Johannes FK. “Niklas Luhmann’s Card Index: The Fabrication of Serendipity.” Sociologica 12, no. 1 (2018). Reprinted as Ch 10 in Practicing Sociology: Tacit Knowledge for the Social Scientific Craft, pp. 101-115. Columbia University Press, 2024.

Stay in the Writing Slowly loop and never miss a thing (unless I forgot to press ‘publish’, in which case, yeah, you might miss a thing. Anyway: