Keeping a note you will use again
95 min
Two hosts talk the lesson through. The voices are synthetic; the script was written from this lesson and checked against it, and asserts nothing the lesson does not.
- Design a review routine for a body of notes, using retrieval and spacing from How to Learn Anything rather than re-deriving them
- State what makes a note findable later, and what that costs at the time of writing
- Evaluate a personal knowledge system against the time it costs to maintain
Six lessons have been about what a note is and what the evidence does and does not say. This one is the constructive half, and it starts where the evidence points rather than where the advice literature does.
Start from the finding
Lesson 1 quoted the field's own summary and it is worth having again, from the 2024 meta-analysis: "Such findings indicate that the primary benefit of taking lecture notes emerges from the review of notes."1
That sentence was measured on college students taking lecture notes, tested inside a week, which is the scope lesson 1 set out and the scope every figure in this course carries. This lesson applies it to a meeting, a report and a notebook kept over a year. That is a step past the evidence and the course is taking it deliberately, because the reader is not sitting in a lecture and a course that only spoke about lectures would be no use to them. Where the step matters, the lesson says so again.
So the question this lesson asks is not how to write a note. It is what you do with the note afterwards, and almost nobody has ever measured whether they review anything, which is what the course project is built to test on your own week.
Everything below is about the second half of lesson 1's two functions.
The routine, which you already have
How to Learn Anything lesson 3 is retrieval practice and lesson 4 is spacing. This course uses both and re-teaches neither, and if you did that course you've already got the two things a review routine needs.
What is new here is the object. Those lessons were about material; this is about a document, and the difference shows up in one operation. Material is either learned or not. A document is also a record you have to trust, so a note can need checking rather than testing, which is what the meeting note gets below and what no amount of retrieval practice would do for it.
A note you reread is a page you are looking at. A note you test yourself against is a question. Same page, two operations, and that course's lesson 3 argues that the second one is much stronger.
So a review routine has three parts, and this course's contribution is only the third.2
- A retrieval step, from How to Learn Anything lesson 3. Cover the note and say what is on it, or write the question the note answers on the other side.
- An interval, from that course's lesson 4. Long enough that you have to work, short enough that you still can.
- A scope you will actually keep. Not everything you have ever written. One project, one subject, one notebook.
The third is where routines die and it's the one nobody plans for.2 A routine covering everything is a routine covering nothing by the second week, and that is this course's own observation rather than a finding anybody has reported.
One week, with a routine attached
This case is constructed and the notes in it are invented.2
Monday. A meeting about a supplier. Four decisions, who owns each, one number.
Tuesday. Reading notes on a report, two pages, mostly paraphrase.
Wednesday. Somebody explains how the billing system handles refunds. One page, written fast.
Thursday. Nothing worth keeping.
Friday. An idea of your own about the refunds problem, half a page.
Now the routine, fifteen minutes on the following Wednesday. Four notes, and it does something different to three of them.
The billing explanation gets the retrieval step, because it is the one you will need and the one you understood least at the time. Cover it. Say out loud how a refund is handled. Look. Note what you missed.
The meeting note gets a different treatment, because it is not material to be learned. It gets checked: are the four decisions still what you think they were, and has anybody acted?
The reading notes get nothing this week. They are storage and you will come to them when the report matters.
And the idea gets five minutes of writing, because an idea you don't develop within a week is usually an idea you've lost.
Notice what the routine did not do. It did not touch everything. It did not reread. And it matched the operation to what the note was for, which is lesson 1's two functions doing work.2
Before the next section. You have a note you wrote eight months ago and you know it exists. What would have had to be true, at the moment you wrote it, for you to be able to use it now?
Show the answer
Two things, and most people name the second and miss the first.
You would have to be able to find it. Which means you knew, at the time of writing, where this kind of note goes, and you put it there rather than wherever was open.
And it would have to make sense out of context. Eight months ago the meeting, the argument and the reason were all in your head, and none of them is now. A note that assumed them is a note about something you can't now reconstruct.
Both of those cost time at the worst possible moment, which is while the thing is still happening and you are least willing to spend it. That's the real trade in note-taking, and it isn't the one the famous study is about.2
And this course has read nothing about it. Nothing in its sources measures findability, context, or what makes a note usable a year later. What follows is reasoning, and the lesson says so rather than dressing it as evidence.2
Findability, and what it costs
A note you cannot find is a note that does not exist. Everything below is this course's own reasoning from that, and no source read here tests any of it.2
One line of context at the top. What this is, when, and why you wrote it. Fifteen seconds, and it is the thing this course would put first.
One place, decided in advance. Not the best place: the place. A system where everything goes to one inbox and gets sorted later works, and a system where notes go wherever was open does not, because "wherever was open" is not searchable by the question you will actually ask.
Something to search by. Whatever you will type in eight months. A project name or a person's name is usually a better bet than a topic.
And the cost of all three is at the time of writing. That is why they do not get done, and saying so is more useful than recommending them again.
The systems
You have probably met at least one of the personal knowledge systems: Zettelkasten, second brains, linked notes, atomic notes. They are the maximal version of the external-storage function, and they are what most people mean when they ask about note-taking.
What this course can tell you about them is less than you want.
None of the books that teach these was opened for this course, so what follows is a description of them rather than a reading of them, and it comes from general knowledge rather than from anything in the research file.3
What they are. A Zettelkasten is a collection of small notes, each holding one idea, each given an identifier, and linked to each other so that reading one leads you to the related ones. The second-brain family is the same idea with the emphasis on capture and retrieval rather than on linking. Both are ways of making a body of notes navigable by association rather than by folder.
What is claimed for them. That over years they compound: that the links do work you did not do deliberately, and that a large well-linked collection produces connections you would not otherwise have made.
What the evidence is. A search for peer-reviewed work measuring whether any personal knowledge system improves anything returned nothing.3 That is the finding and it is worth stating as one rather than as a gap.
What the advice literature offers instead is the output of Niklas Luhmann, whose card index of about ninety thousand slips the method is named after, and who was extraordinarily productive.3
So is a Zettelkasten worth building or not?
Show the answer
The honest answer has three parts and only the third is a recommendation.
A practice with no evidence is not a practice shown not to work. Nothing read for this course tests these systems in either direction, and a lesson that said "no evidence, therefore no" would be making the mistake lesson 5's two non-significant rows exist to prevent, in its stronger form: there somebody measured and found nothing, here nobody measured.
And the case its advocates make is a sample of one. One person, no comparison group, and that person is the one whose practice it is modelled on. Memory lesson 7 taught the sample question and this is its sharpest form: not a small sample, a sample chosen because of its outcome.
Which is not an argument that it does not work. It is an argument that his output is not evidence that it does, which is a different sentence and the one this course can support.
The recommendation is narrow, and it comes in two halves that are not equally backed. Going back to notes on purpose is review, and that is the one part of these systems this course can point at a finding for. A note that makes sense on its own is context, and nothing read here measures it: it is this course's own reasoning from the findability section above, said there and repeated here so the two are not taken together. Neither needs an app.
The parts this course cannot back are the parts people spend the time on: the identifiers, the link graph, the taxonomy. Nothing read here says they're worthless and nothing read here says they work. If you enjoy building one, that's a real reason and it isn't the reason you're usually given.
Before the numbers. Two systems, one careful and one rough. Which of them is still running in November?
Show the answer
The rough one, and the reason is the whole of this section.
A careful system's cost is paid every day and its benefit arrives over years, so the first week when the daily cost isn't available is the week it stops. You can probably name that week in advance: a deadline, an illness, a fortnight away. A system that has stopped returns nothing whatever the quality of what is in it.
The numbers below are invented to show that shape. What they cannot show is your own November, which is what the second exercise asks you to price.
Two systems, costed
These numbers are invented, and the point is the shape rather than the figures.2
System A takes twelve minutes a day. Every note gets an identifier, a set of links and a place in a structure. The notes are genuinely better: a year in, you could find anything.
System B takes two minutes a day plus fifteen minutes once a week. One line of context at the top, one inbox, and the weekly review itself.
A is better per note and that is not the question. Over a year, A costs about seventy-three hours and B about twenty-five: twelve for the daily minutes and thirteen for the weekly review. A has no review step at all, which is the first thing wrong with it.
And A does not run for a year. It runs until the first week when twelve minutes a day is not available, which is a week you can probably name in advance. Then it stops, and a system that has stopped returns nothing, whatever the quality of what is in it.
That is why maintenance cost is the variable that decides, and it's the course's own reasoning rather than something anybody has measured.2 A system's benefit arrives over years and its cost arrives every week.
Four things people get wrong about this
The corrections below are this course's reasoning from the two functions rather than findings, and the last is a correction of this lesson's own likeliest misreading.2
"The right app will fix it." Every system above is a habit with software attached. The habit is the part that fails.
"A system that works for a famous person will work for me." One person, no comparison, and the famous person is the one whose practice the method is modelled on.
"If I capture everything I will be able to use it." Capture is the storage half and review is where the benefit is. More capture without more review is more maintenance for the same return.
"There's no evidence for it, so it doesn't work." A practice nobody has tested is not a practice that has failed a test, and that sentence is one this lesson would refuse in either direction.
Practice
Take 30 minutes. Then take 15 minutes in each of the next two weeks, which sit outside this lesson's sitting.
Pick one body of notes. One project, one subject, one notebook. Not everything.
Write the routine down in four lines before you start.
- The scope. What is in and what is out, named specifically.
- The retrieval step. What you will actually do to the page. Cover and recall, write the question on the back, explain it to somebody. Not "read".
- The interval. When the first review happens, how you will be reminded, and what the second gap will be if the first one goes well. That second number is How to Learn Anything lesson 4 being used rather than named: a gap that worked is a gap you can stretch.
- The stopping rule. What would tell you this is not worth continuing.
Then run it twice.
Then two lines. What you found you could not recall, and what the routine cost in real minutes against what you budgeted.
The fourth line is the one people leave out and it's the one that makes this honest. A routine with no stopping rule is a routine you will abandon without deciding to.
Take 20 minutes.
Write down every part of your current note-taking system and what it costs you in minutes a week. Capture, filing, tagging, linking, reviewing, tidying, moving things between tools. If you have no system, list what you do instead, which is also a system and also costs something.
Then two things against each part.
- Minutes a week, honestly, including the tidying you do instead of working.
- What it returned in the last month. Specific occasions on which you got something out of it.
Then one line: what is the most expensive thing on that list that returned nothing?
It is usually an organising step rather than a capturing one, which is this course's expectation rather than something it has measured.2 The project measures the return side of this properly, and this exercise is the cost side.
Connections
Back. Lesson 1 is the two functions, and this lesson is the external-storage half of them. How to Learn Anything lessons 3 and 4 are retrieval and spacing, used here and not re-taught. Memory lesson 7 is the sample question, which is what the Luhmann case runs into. Memory lesson 6 is the template for how this course treats a technique with narrow evidence and a wide reputation.
Forward. Lesson 8 is how to read the next claim you meet about any of this, and it is where the four books get named.
Go deeper
- Typed Versus Handwritten Lecture Notes and College Student Achievement: A Meta-Analysis (Educational Psychology Review, 2024). Read in part by this course. Read its introduction for the sentence this lesson starts from, which is the field's own summary of where the benefit sits.
- Retrieval practice, a free resource run by researchers in that field. Not read for this course, and named because the routine above is retrieval practice applied to a document, and somebody who wants the evidence behind the retrieval step should get it from people who study it rather than from a note-taking course.
Sources
- Abraham E. Flanigan, Jordan Wheeler, Tiphaine Colliot, Junrong Lu and Kenneth A. Kiewra, "Typed Versus Handwritten Lecture Notes and College Student Achievement: A Meta-Analysis", Educational Psychology Review 36, 2024, article 78. Read in part, at the level lesson 1's scope table records. Supports: the quoted sentence about where the primary benefit of taking notes emerges, and nothing else in this lesson.
- Almost everything in this lesson is the course's own reasoning and is labelled as such where it appears, which is a larger share than in any other lesson here and is worth saying plainly: nothing read for this course measures findability, context, maintenance cost, or what makes a note usable a year later. That covers the three-part routine and especially its third part; the constructed week and everything in it; the account of what a note needs to survive eight months; the three findability recommendations; the two costed systems and their invented figures; the claim that maintenance cost is what decides; and the expectation about which step returns nothing.
- The two best-known books on these systems, and Niklas Luhmann's own writing about his card index, are on the research file's unread line: none was opened for this course. The build refuses a citation to any of them, and this lesson is exempted for Luhmann alone so that it can say whose card index the case rests on; it cites him for nothing, and lesson 8 is where all four are named. The figure of about ninety thousand slips is at search-summary level. The description of what a Zettelkasten is and what is claimed for it is not in the research file at all, and the body says so where the description begins. The statement that a search found no peer-reviewed work measuring whether any of these systems improves anything is a finding of this course's own Stage 1 search, recorded in the research file, and it is a statement about what that search found rather than a claim that no such work exists anywhere.
Check your understanding
This lesson has a 6-question quiz. Pass it and the questions come back on a schedule in Review, so what you learned stays learned. Your progress is saved in your browser; no account needed.