한국어

Is Zettelkasten Good for Studying? What the Research Says

At a glance
  • No controlled study we could find has tested the Zettelkasten as a whole against other ways of studying.
  • Luhmann built his slip box for writing, and called it a second memory, not a memory trainer.
  • Writing notes in your own words does help. A meta-analysis of 86 studies found a reliable benefit.
  • Linking and organizing notes is not enough on its own. Closing the book and recalling kept more.
  • To study with it: write each note with the book closed, title it as a question, and recall it on later days.

No controlled study we could find has tested the Zettelkasten as a whole against other ways of studying. Its habits, though, have each been tested. Writing notes in your own words helps you remember. Linking and organizing notes on their own keep less, and closing the book to recall is what makes things last. So to study with a Zettelkasten, add one step: recall.

What is the Zettelkasten method?

A Zettelkasten is a slip box: one idea per note, with notes tied to each other by number and by links. The German word means "note box".

The method comes from the German sociologist Niklas Luhmann. In a 1981 essay he described his box. Wooden drawers hold A6 paper slips. Notes are not filed by topic; each gets a fixed number. A new note can branch off any existing one (57/12 is followed by 57/12a), and a note lists the numbers of related notes elsewhere. A separate keyword index helps him find things.

About 90,000 slips That is how many notes Luhmann wrote between 1951 and 1996. Bielefeld University's Luhmann archive now holds them.

One point is easy to miss. Luhmann built the box to write and to develop ideas, not to pass exams. In the same essay he calls it a kind of second memory and a partner to talk with, and he notes that some slips you will never see again. What the box remembers and what you remember are two different things.

Is there research showing a Zettelkasten works for studying?

We found no controlled study that compares the Zettelkasten itself with another study method. The top search results for this question describe the method or share experiences, but do not cite experiments.

So this note takes the method apart. Each part has been tested for decades in learning research.

Zettelkasten habitClosest researchWhat it found
Write in your own words, do not copyGeneration effectWhat you produce stays longer than what you read
Link and organize notesElaborative study, concept mappingKept less than recall did
Store it in the box and look it upRetrieval practice (this step is missing)Recalling like a test kept more days later

Do you remember more when you rewrite in your own words?

Yes. That what you generate yourself is remembered better than what you only read, the "generation effect", has held up for decades.

Bertsch and colleagues (2007) pooled 86 studies. The average benefit of generating was 0.40, almost half a standard deviation. McCurdy and colleagues (2020) went further, with 126 papers and 310 experiments. In a typical study people either read word pairs as given, or produced the second word from a cue such as its first letters, and were then tested.

On every kind of test, what people produced themselves stayed longer

  • Read only
  • Generated
0%25%50%75%100%Read only · Recognition · 62.2%62.2%Generated · Recognition · 76.7%76.7%RecognitionRead only · Cued recall · 52.7%52.7%Generated · Cued recall · 65.9%65.9%Cued recallRead only · Free recall · 26.8%26.8%Generated · Free recall · 36.3%36.3%Free recallWith no hint,9.5 points more
Estimated mean proportion remembered, pooled across experiments. Recognition means picking out what you saw; cued recall means recalling with a hint; free recall means recalling with no hint.Source: McCurdy et al. (2020), Table 1, model 5
See the numbers
Read onlyGenerated
Recognition62.2%76.7%
Cued recall52.7%65.9%
Free recall26.8%36.3%

The same analysis has a result that matters more for students. The less the answer was fixed in advance, the bigger the benefit. The authors call this generation constraint. A cue that allows only one answer is highly constrained; one that allows many is not.

The less fixed the answer, the more people recalled

0%20%40%60%Generated items · Less constrained · 48.4%48.4%LessconstrainedGenerated items · Medium · 35.4%35.4%MediumGenerated items · Highly constrained · 29.5%29.5%HighlyconstrainedThe more fixed,the less recalled
Estimated mean proportion of generated items recalled with no hint (free recall), pooled across experiments.Source: McCurdy et al. (2020), Table 1, model 9b
See the numbers
Generated items
Less constrained48.4%
Medium35.4%
Highly constrained29.5%

This is why the Zettelkasten tells you to write in your own words rather than copy sentences from the book. Copying a highlighted line is close to reading. Rewriting it in your own words is closer to low-constraint generation. Keep in mind that most of these experiments used words, not summaries of long texts. The same analysis also found smaller effects when people wrote or typed what they generated than when they said it aloud.

Does the benefit last past a day?

Yes. It lasted past a day, and the gap grew. McCurdy et al. (2020) also split the studies by when the test was given.

The gap for generated material was largest after more than a day

  • Read only
  • Generated
0%25%50%75%100%Read only · Immediately · 47.9%47.9%Generated · Immediately · 57.9%57.9%ImmediatelyRead only · 5 min to 1 day · 48.7%48.7%Generated · 5 min to 1 day · 58%58%5 min to 1 dayRead only · Over 1 day · 44.4%44.4%Generated · Over 1 day · 60.8%60.8%Over 1 day16.4-point gap
Estimated mean proportion remembered by delay between study and test. The over-a-day group has few comparisons (40 to 43). Point differences are our arithmetic.Source: McCurdy et al. (2020), Table 2, retention delay
See the numbers
Read onlyGenerated
Immediately47.9%57.9%
5 min to 1 day48.7%58%
Over 1 day44.4%60.8%

On an immediate test the gap was 10.0 points; after more than a day it was 16.4. The authors report that long delays produced a significantly larger generation effect. A note written in your own words still pays off days later, not only on the day you write it.

Does linking notes make you remember them?

Not on its own. Building relations with the material open kept less than closing it and recalling.

Karpicke and Blunt (2011) compared students who studied science texts by drawing concept maps with students who closed the text and recalled it. Recall produced more learning, on questions that asked for inferences, and even when the final test was to draw a concept map. Linking notes is close to concept mapping. We charted that study's numbers in Is Note-Taking a Waste of Time?.

Roediger and Karpicke (2006) found something similar. Students who reread a passage did better on a test 5 minutes later, but on tests 2 days and 1 week later, those who had taken recall tests remembered much more. The rereaders were more confident they would remember.

A Zettelkasten has steps for writing notes, linking them, and finding them when needed. The box does the finding, and there is no step where your head does the recalling. For Luhmann, who was writing books, that was enough. A student cannot bring the box into the exam, so one step is missing.

How should you change it for studying?

Write each note with the book closed, title it as a question, and recall it on later days. You can try this today with paper; no app is needed.

  1. Write the note with the book closed. Read a paragraph, close it, and write one or two sentences from memory in your own words. That is generation and recall at once. Open the book again only when you are stuck.
  2. Title each note as a question. Instead of "spacing effect", write "Why do wider review gaps last longer?" Later the title alone becomes a question to answer.
  3. Write one line on why two notes link. Not "related to this", but "both say harder recall makes things stick". The link becomes low-constraint generation too.
  4. Recall from the title the next day, a few days later, and a week later. Cover the body, say the answer, then check. When to come back is in our spaced repetition schedule note.
  5. Do not turn whole lectures into notes. Leave details you can look up in the source, and make notes only of ideas you need to be able to explain.

In Aloom

Aloom is like a Zettelkasten in that notes link to each other. You ask AI a question, keep only the answers worth keeping, and each one joins what you already know in a knowledge graph. The difference is the step this note says is missing. Each day you recall the number you chose, 1, 3 or 5, as questions before you see the answers. The more you recall a node, the darker its ink, so you can see what you remember, not only what the box holds. See how Aloom works

FAQ

Is Zettelkasten worth it for students?

It is worth it if you want to understand material well enough to explain and connect it. On its own it is not a review system, so pair it with closed-book recall on later days. For subjects built on problem solving, notes do not replace practice problems.

Do I need Obsidian for a Zettelkasten?

No. Many people use apps like Obsidian because notes can link to each other, but Luhmann used paper slips in wooden drawers. What matters more than the tool is whether you write in your own words, and whether you later recall what you wrote with the note covered.

Does the Zettelkasten method work?

For writing and developing ideas, the main evidence is Luhmann's own report after 26 years of use, and that is one person's account. For studying, nobody has tested the method as a whole. Its own-words habit is supported by research; its linking step is weaker than recall, so add recall.

References