What a Zettelkasten Is Actually For

My friend Brennan Kenneth Brown has written a post asking what note-taking PKMs have actually accomplished, and the honest answer, in my estimation, is that he is mostly right. Brennan is one of the best people on the indieweb, a Queer Métis writer and developer in Calgary who builds things and shares them freely, and this piece has the quality I associate with him: he went and looked. He does not sneer at people who like Obsidian; he asks for evidence, finds very little, and reports the finding.

While Brennan is right that PKM (personal knowledge management) has made no one measurably smarter, I will argue that the doctoral student is the one reader with a serious context of use, by showing 1) that the Luhmann legend collapses under archival scrutiny even more completely than Brennan suggests; 2) that the sharpest line in this debate, Andy Matuschak's remark about contexts of use, points toward graduate education rather than away from note-taking; and 3) that the disciplines never teach the skill and then call the resulting struggle personality. The stakes here are pedagogical rather than aesthetic; I teach graduate students, and I assign this method in an undergraduate course.

Where Brennan is right, and then some #

The Luhmann story has become a productivity fable, and the archival record does not support the fable. Consider what the record actually says.

Johannes F. K. Schmidt is the scientific coordinator of the Luhmann project at Bielefeld, which acquired the estate in 2011 and began publishing the card index online in 2019. Schmidt has counted the cards. There are about 90,000 of them across two collections, so that famous number survives checking; almost nothing else does. Luhmann did not publish seventy books. The Bielefeld archive puts him at roughly 600 publications and just over forty monographs, and the seventy-book figure that circulates in blog posts traces to an uncited sentence on Wikipedia rather than to any scholar or to Sönke Ahrens, who gets blamed for it.^[1]

More damaging to the fable: Schmidt titles his article "Niklas Luhmann's Card Index: The Fabrication of Serendipity," and the operative word is fabrication. He finds that a referencing discipline engineered the surprises; Luhmann did not stumble into them. In his words, it was "specifically not (only) the paths that Luhmann tread in his initial readings and note-taking" that made the system generative, "but rather the special filing technique and the (selective) relations established between his notes."^[2] Luhmann did that work. Every day, for forty-five years. Schmidt reports that the file "preoccupied so much of his time," and Luhmann said as much himself in interviews.^[3] The box did not think for him; he thought, in writing, into a box.

Likewise, the architecture resembles nothing the apps sell. Luhmann's keyword index held about 1,250 entries for the 23,000 cards of the first collection, roughly one index entry for every twenty cards. The index was a door, not a filing scheme, and the actual work happened by following links from wherever you entered.^[4] Full-text search inverts this design completely: when your software can find any string instantly, you never build the links, and the links were the entire machine.

Then there is the part almost nobody mentions. Luhmann built the entire first collection, all 23,000 cards from 1951 to 1962, while working in public administration with no academic position at all. He came to the academy sideways in his mid-thirties, had two already-written books accepted retroactively as dissertation and habilitation at Münster in 1966, and then spent twenty-five years (1968–1993) at a brand-new sociology faculty, working on a single thirty-year project he had declared in advance. The box is one line on that résumé.

Schmidt's own conclusion is that the file "can be seen as — drawing on the words of Erving Goffman — the backstage of his theory and therefore as Niklas Luhmann's intellectual autobiography."^[5] An autobiography is not a method you can install.

Brennan's other points land too. Alberto Cevolini, who edited the standard volume on early modern knowledge management, states as the main hypothesis of an entire article that in the use of a card index as a surprise generator "there is nothing particularly surprising if one considers the evolution of knowledge management in early modern Europe."^[6] Schmidt says plainly that Luhmann adopted the technique. Erasmus, Bacon, Milton, Locke, Newton, and roughly every literate schoolboy in Renaissance Europe kept commonplace books, because commonplacing was a compulsory school exercise with a printing industry behind it selling pre-ruled blanks. If a method is universal, it cannot explain why one person is extraordinary. And by the seventeenth century, critics attacked commonplacing for producing exactly the derivative, second-hand learning that Brennan sees in the PKM blogosphere. This critique is four hundred years old, and the practitioners themselves made it.

The collector's fallacy is real, and Christian Tietze named it well in 2014: "to know about something isn't the same as knowing something."^[7] Brennan is right that the PKM industry's main verified output is PKM content; I would add only that this failure mode has haunted note-taking since Umberto Eco warned about photocopies.

The one place I want to push #

Brennan quotes Andy Matuschak's note that people who write extensively about note-writing "rarely have a serious context of use."^[8] In my estimation this line is the sharpest in the whole debate, and it points somewhere Brennan did not follow it.

The complaint is not that notes are useless. It is that the loudest advocates have no external work the notes must serve, so the system becomes the work (the collection, in short, takes on a life of its own). Take the line seriously and it becomes a test: who has a serious context of use?

A doctoral student has one by definition. She has a multi-year project with a required output; a comprehensive exam that asks her to hold a field in her head; a literature review that must synthesize rather than summarize; and a citation obligation that will require her, in year six, to find the page number of something she read in year one. The National Science Foundation puts median time to doctorate at 7.2 years from entry into graduate school.^[9] No working memory covers that span; no pile of PDFs does either.

Brennan's counterexamples are all people who already know what they are writing. Robert Greene's cards are theme-sorted extracts waiting to be shuffled into chapters of a book he has already sold; John McPhee's index cards are structural labels for a piece he has already reported; Susan Cain, in Brennan's description, dumps everything into one giant Word document, which works beautifully when you know what the document is about. Greene, McPhee, and Cain give us methods for people who have found their subject. A PhD student's central problem is finding one, and finding it out of several thousand pages of other people's arguments; the note system fills the lacuna between reading a field and having something to say to it.

Barbara Lovitts names this transition exactly: being a good course-taker is not enough.^[10] Programs admit students because they excel at coursework, then ask them to do something structurally different. The evidence backs her up. Across the studies Bair and Haworth synthesized, undergraduate GPA predicted doctoral completion in only five of twenty-three studies, and GRE Verbal in three of twenty; prior academic ability barely predicts who finishes.^[11] I should be precise about one thing here, because it cuts against the story I am telling: most doctoral attrition happens early, before candidacy, not at the dissertation. But among students who do reach the dissertation, one study in that same synthesis found half of them naming lack of structure at that stage as their central obstacle. Coursework supplies deadlines and syllabi. Then it stops, all at once, and what replaces it is whatever you built yourself.

The training vacuum #

Consider what convinced me; it comes from my own field.

Ithaka S+R has run interview studies of how scholars actually work, and Brennan cites their brief on scholars as collectors. Their 2017 report on religious studies interviewed 198 scholars and found that "scholars generally develop idiosyncratic approaches to organizing and storing the information associated with their research," systems that are "highly personal and developed over the course of their careers," with many admitting to "not having much of a system at all and being disorganized."^[12] Brennan characterizes this literature accurately.

But read their historians report and one sentence goes further. Managing notes and sources is "a primary challenge," the report says, and then: "although most struggle with this process, it is not addressed in a formal (or informal) way in the education of historians. Historians are expected to develop a personal approach to this process." In most cases, they "rely on their peers, from their dissertation cohort, for tips and tricks."^[13]

That sentence does not describe a discipline that has considered note systems and wisely declined them; it describes a discipline that never teaches the skill, watches everyone struggle with it privately, and calls the struggle personality. The same report catches historians photographing entire archive boxes without reading them, then never processing the images: "If I don't process it then, annotate, decide what's important, it just goes into a big pile that never gets figured out. You don't know what you've got at the end, and you have to essentially go through it all again."

So when Brennan says scholars work without formal branded systems, he is right, and the data he points at also says they are drowning. Those two findings are compatible; they are, I would say, flip sides of the same coin. The absence of adopted systems is not evidence that systems do not help. It is evidence that nobody teaches them.

What I actually claim #

I do not tell my students a slip-box will make them prolific. I tell them four rules, and each one is a delivery mechanism for something with better evidence behind it than Zettelkasten itself has. 1) One idea per note, three to five sentences; the constraint forces a decision about what the idea is, which is the smallest possible unit of synthesis. 2) Write it in your own words, never copied; this practice is elaborative encoding, and putting a claim into your own language is the moment you discover whether you understood it. 3) Record the source (author, title, page); in year four you will need it, and you will not remember. 4) Link it to something you wrote earlier; that link is the actual thinking, and it is what Schmidt says made Luhmann's box work.

The strongest empirical support here is not for the card system at all; it is for the habit the card system enforces. In Robert Boice's small 1983 study (twenty-seven faculty in three groups of nine), the faculty assigned five scheduled writing sessions a week averaged 3.2 pages a day against 0.8 for those who wrote when in the mood, and went about one day between creative ideas against five days for those told not to write at all.^[14]

I want to be careful with that paragraph, because Brennan's discipline applies to me. Those numbers reach us secondhand, through Helen Sword's reconstruction, and Sword's whole point is that this literature is unreliable. She went back to the original and found that widely circulated Boice statistics "appear nowhere in the original article."^[15] A body of writing about disciplined scholarship propagated figures nobody had checked; twenty-seven subjects is also not much to hang a practice on. What I take from Boice is a shape rather than a result: small, regular, low-stakes writing seems to beat waiting for the mood. In short, a modest claim, and I hold it loosely.

A permanent note is a small, regular, low-stakes piece of writing with a deadline attached to a reading you had to do anyway. In my estimation, that sentence is the whole trick.

And I will concede one more thing to Brennan that he did not even ask for. The permanence claim is oversold. An Ithaka historian said it best: "It's not like I can go to my notes from [my last] book, and put them together in a different order and write a different book." Notes organize around the thing you are writing, and they mostly do not survive into the next project; McPhee throws his cards away. This disposability is not a failure. A slip-box that helped you write one dissertation has done its job even if you never open it again.

So Brennan and I end up close. His advice to beginners is to stop hunting for the perfect system and start creating: "Start a blog! Learn and freewrite on your discoveries, the organization and sorting of those notes can come later, if at all." He is exactly right, and I would point out that his advice is itself a note-taking method: write regularly, in your own words, in public, about what you are reading. Whether you keep it in Obsidian or a shoebox matters far less than whether you do it.

The thing worth defending is not the software. It is the small daily act of turning someone else's argument into your own sentence, and the link you draw from that sentence to one you wrote in March; this act, repeated, is the whole method writ large. Everything else is furniture.

Notes #

---

1. Niklas Luhmann-Archiv, Universität Bielefeld, gives "annähernd 600 Veröffentlichungen, darunter über 40 Monographien." Sönke Ahrens, How to Take Smart Notes (2017), gives a figure in the high fifties. The English Wikipedia entry on Luhmann says he wrote "more than 70 books and nearly 400 scholarly articles," a sentence that carries no citation and appears to be the source of the number in circulation. ↩︎

2. Johannes F. K. Schmidt, "Niklas Luhmann's Card Index: The Fabrication of Serendipity," Sociologica 12, no. 1 (2018): 58. https://doi.org/10.6092/issn.1971-8853/8350 ↩︎

3. Schmidt, "Fabrication of Serendipity," 57. Luhmann's fuller account of the file as a communication partner is "Kommunikation mit Zettelkästen: Ein Erfahrungsbericht" (1981), and his remarks on how much of his time filing consumed come from a 1987 interview, discussed in Johannes F. K. Schmidt, "Niklas Luhmann's Card Index: Thinking Tool, Communication Partner, Publication Machine," in Alberto Cevolini, ed., Forgetting Machines: Knowledge Management Evolution in Early Modern Europe (Leiden: Brill, 2016), 287–311. ↩︎

4. Schmidt, "Fabrication of Serendipity," 54. The first collection holds roughly 23,000 cards against about 1,250 keyword entries, the second roughly 67,000 cards against 3,200. ↩︎

5. Schmidt, "Fabrication of Serendipity," 59. ↩︎

6. Alberto Cevolini, "Where Does Niklas Luhmann's Card Index Come From?," Erudition and the Republic of Letters 3, no. 4 (2018): 390–420. https://doi.org/10.1163/24055069-00304002. On the longer history see Ann Blair, Too Much to Know: Managing Scholarly Information before the Modern Age (New Haven: Yale University Press, 2010), and Ann Moss, Printed Commonplace-Books and the Structuring of Renaissance Thought (Oxford: Clarendon, 1996). ↩︎

7. Christian Tietze, "The Collector's Fallacy," zettelkasten.de, January 20, 2014. https://zettelkasten.de/posts/collectors-fallacy/ ↩︎

8. Andy Matuschak, "People who write extensively about note-writing rarely have a serious context of use." https://notes.andymatuschak.org/z51q8prEJzs5Jqa5WPThYoV ↩︎

9. National Center for Science and Engineering Statistics, Survey of Earned Doctorates, which reports median time to degree of 7.2 years from graduate school entry and 5.7 years from entry into the doctoral program. For completion, the Council of Graduate Schools' Ph.D. Completion Project, covering 29 US and Canadian universities, reported ten-year completion of 49 percent in the humanities against 64 percent in engineering, with an overall rate of 57 percent. Note that the CGS institutions volunteered for a project about improving completion, so these are probably favorable numbers. ↩︎

10. Barbara E. Lovitts, "Being a Good Course-Taker Is Not Enough: A Theoretical Perspective on the Transition to Independent Research," Studies in Higher Education 30, no. 2 (2005): 137–54. https://doi.org/10.1080/03075070500043093 ↩︎

11. Carolyn R. Bair and Jennifer Grant Haworth, "Doctoral Student Attrition and Persistence: A Meta-Synthesis of Research" (ASHE Annual Meeting, 1999), ERIC ED437008, synthesizing 118 studies drawn from 430 identified. The synthesis reports that "as much as two-thirds of doctoral attrition occurs prior to the achievement of ABD status." The lack-of-structure finding is reported there from Hugeley (1988). ↩︎

12. Danielle Miriam Cooper, Roger C. Schonfeld, et al., Supporting the Changing Research Practices of Religious Studies Scholars (New York: Ithaka S+R, 2017), 28. https://doi.org/10.18665/sr.294119. Brennan cites the related Ithaka brief on scholars as collectors, which draws on this report. ↩︎

13. Jennifer Rutner and Roger C. Schonfeld, Supporting the Changing Research Practices of Historians (New York: Ithaka S+R, 2012). "A primary challenge" at 23, the notes-reuse remark at 24, the training vacuum at 27, the archive photograph problem at 11–12. https://doi.org/10.18665/sr.22532 ↩︎

14. Robert Boice, "Contingency Management in Writing and the Appearance of Creative Ideas: Implications for the Treatment of Writing Blocks," Behaviour Research and Therapy 21, no. 5 (1983): 537–43. https://doi.org/10.1016/0005-7967(83)90045-1. The article is paywalled and carries no abstract. All figures above are as reconstructed in Sword (below), who describes "nine subjects in each of three groups" with outputs "averaging .2 pages for the control group, .8 for the spontaneous group, and 3.2 for the contingency management group." ↩︎

15. Helen Sword, "'Write Every Day!': A Mantra Dismantled," International Journal for Academic Development 21, no. 4 (2016): 312–22. https://doi.org/10.1080/1360144X.2016.1210153. Open copy: https://www.research.ucsb.edu/sites/default/files/RD/docs/Write-every-day-a-mantra-dismantled.pdf. Sword documents that statistics widely attributed to Boice by later writers "appear nowhere in the original article, except obliquely in a rather hard-to-interpret diagram," which is worth remembering every time someone quotes a productivity number at you, including me. ↩︎