I keep a shelf in my apartment in Setagaya with seven half-used notebooks on it. There is a Cornell-format pad from my last year at Waseda, a Leuchtturm1917 dotted that I filled with mind maps during a year I was convinced visual thinking would fix my refactoring problems, a cheap A5 outline pad I use during standups at work, and three nearly identical Midori MD notebooks that I bought because a senior engineer I respected used them. The shelf is, on one level, embarrassing. On another, it is the most honest summary of my note-taking life: I have tried everything, I keep coming back to the same two or three methods, and I get asked about this constantly by juniors at work and by friends still finishing university.
The question is always some version of “which note-taking method should I use.” I have spent enough years optimizing this across Obsidian, Notion, plain markdown, and paper that I have a strong opinion. The best method depends on what you are capturing, what you intend to do with the notes afterward, and how your brain prefers to organize information — and most people pick a method for cosmetic reasons and quietly stop reviewing the notes a week later. Below is what I have actually learned running each of the four major methods on real material: lectures, technical books, conference talks, code reviews, design docs.
Before I get into the individual methods it is worth saying how they compare in shape. Cornell is the gold standard for academic lectures and structured study material; it is medium-speed to capture and its review value is the highest of the four, because the format itself forces you to come back and process the page later. Mind mapping is slow during capture and only middling for review, but it earns its keep for brainstorming and surfacing non-obvious connections. The outline method is the fastest of the four to write and reviews well, which is why it has quietly become the default for almost everyone who takes notes on a laptop. Sketchnoting is slow and demanding but has the highest review value alongside Cornell, because drawing forces you to decide what matters before the pen touches paper. The right answer for any given situation is the intersection of the material, the time you have to capture it, and whether you will ever return to the page.
Cornell
Walter Pauk developed this at Cornell University in the 1950s, and despite seventy years of trends and tools it is still the most-recommended academic note-taking system in the world. The page is divided into three zones. On the right is the main column where you take notes during the lecture — bullets, abbreviations, your own phrasing, whatever keeps up with the speaker. On the left, a narrower column stays blank during the lecture and gets filled in within twenty-four hours with keywords, questions, or prompts that act as recall cues. At the bottom of the page, a horizontal strip waits for a two- or three-sentence summary of the page’s takeaway, also written after the fact. The real work is not the capture; it is what happens later.
I tried Cornell properly for the first time during an algorithms class in my second year at Waseda, and it was the first time in my life that my notes were actually useful for studying instead of just being a record of my own handwriting. The trick is the cue column. Within a day, you re-read the right side and ask, for each chunk, what is the question this would answer? Then you write that question in the left column from memory. You have built yourself flashcards without using any flashcard software, and the act of generating the cues is itself active recall — one of the two or three best-evidenced learning techniques there is. The summary line at the bottom does the same thing at a larger granularity, forcing synthesis you would otherwise skip.
Cornell shines on lectures with a clear structure, textbook chapters, and any material you will be tested on later. It falls apart fast when you try to use it for brainstorming, fast-moving meetings without room to keep up, or anything with diagrams that do not fit cleanly in the narrow right-hand column. I have also learned the hard way that Cornell on a laptop is mostly cosplay. The spatial separation between cue and note is the entire point, and reproducing it in a Notion template where the columns scroll independently kills the cognitive load gain. Get a notebook with the format pre-printed, or print templates at home. Pretending Markdown is paper is not worth it.
Mind mapping
Tony Buzan popularized mind maps in the 1970s, and the shape is familiar: a central topic in the middle of the page, branches radiating outward to subtopics, sub-branches forking from those, and lines or colors connecting concepts that relate across branches. The idea is that the spatial arrangement mirrors the way the brain stores associative information, which is a stronger claim than the evidence really supports but a useful enough metaphor to work with.
I went through a mind-mapping phase about three years ago when I was rewriting an authentication module at work. The codebase had grown organically for five years and the dependencies between services were tangled in ways no UML diagram in our wiki actually captured. I sat down with an A3 sheet and mind-mapped the system from the central node outward, and within ninety minutes I had spotted two circular dependencies and one race condition that had been hiding in plain sight for two years. That was when I understood what mind maps are actually for. They are not a capture format. They are a thinking tool for material where the relationships matter more than the sequence.
Where mind maps fail is the exact place people most often try to use them, which is taking notes in real time during a lecture. By the time you have drawn one section’s branches the speaker has moved on, and the page is usually illegible halfway across. They are also bad for closed-book exam review, because reading top-to-bottom is harder than reading a linear page. The places I keep using them are pre-lecture previews, where I spend five minutes mapping what I already know about a topic to activate prior-knowledge anchors; post-lecture synthesis, where I take ten minutes after a session of normal notes to map the key concepts and see what connections I missed; and essay or design-doc planning, where I want the conceptual landscape laid out before I linearize it. Treat the mind map as a complement to other methods, not a primary capture format, and the disappointments stop.
Outline
The outline method is the classic numbered or indented format you see in every academic textbook: major topics at the highest level, sub-topics indented below them, supporting details indented below those. It looks like the example below, which is roughly how I would outline a high school biology refresher on photosynthesis:
I. Photosynthesis overview
A. Inputs: CO2, water, sunlight
B. Outputs: glucose, oxygen
C. Two-stage process
1. Light reactions
a. Occur in thylakoid membrane
b. Produce ATP and NADPH
2. Calvin cycle
a. Occurs in stroma
b. Uses ATP/NADPH to fix carbon
The outline is by far the fastest of the four methods to write, especially on a laptop where indentation is a single Tab keystroke, and the hierarchy is visually obvious without you having to think about how to draw it. It reviews linearly without effort and works for almost any structured content. This is why it has quietly become the default format for everyone I know who takes notes digitally — Obsidian, Notion, Bear, plain markdown, OneNote, Roam, the tool barely matters because outlines transport across all of them without losing structure.
The weaknesses are real but overstated. Outlines do not capture cross-branch relationships well — if a concept under Section I.B.2 connects to a concept under Section III.A, the outline will not show you that without an awkward parenthetical. They can feel rigid, because everything has to fit a hierarchy. And it is easy to drill into one sub-branch and lose the big picture, which is why I now always write a one-sentence summary at the top of each outline page before I start taking notes.
The variation I actually use day to day combines the outline format with a Cornell-style cue margin. After a meeting or a study session, I add a column down the left side — physically with a margin in a paper notebook, or as a blockquote callout in Obsidian — with prompts and questions generated by re-reading my outline. Outline speed during capture, Cornell review value afterward. It is the closest thing to a default I would recommend to anyone who has not yet found their groove.
Sketchnoting
Sketchnoting is the visual end of the spectrum. You capture ideas with a mix of words, simple drawings, frames, arrows, and varied typography, and the result looks more like an illustrated infographic than a traditional notes page. The theory behind why it works is dual-coding, which is one of the better-supported ideas in cognitive psychology: information stored both verbally and visually is recalled more accurately than information stored as either alone. Drawing also forces you to interpret and decide what matters, because you literally cannot transcribe a slide into a sketch — you have to pick what to keep. And visual hierarchy, bigger words for important ideas and smaller for supporting ones, communicates emphasis at a glance in a way that no bullet list manages.
I have only really used sketchnoting in one context that worked, which was a design conference I attended in Yokohama in 2024. The speakers were talking about systems thinking and software architecture and the material was almost entirely conceptual — no formulas, no code, no facts I needed to memorize. I bought a B5 sketchbook in the lobby that morning and spent the day drawing the talks instead of transcribing them. Three weeks later I could still recall the structure of each talk in a way that my outline-format conference notes from previous events had never given me. Six months later I still could. That is the dual-coding effect made tangible.
Where sketchnoting falls apart is anywhere the material is fact-dense or moves fast. I once tried to sketchnote a technical talk on Rust’s borrow checker and spent the entire hour drawing one decent diagram of ownership while the speaker covered four other concepts I missed entirely. STEM material with formulas and proofs is brutal for sketchnoting, language vocabulary needs to be precise rather than evocative, and any lecture pacing that does not give you breathing room will leave you behind. There is also the aesthetic trap, where you spend more time on whether the drawing is pretty than on whether it is correct, and end up with beautiful pages that taught you nothing. If you want to start, the highest-leverage thirty minutes you can spend is learning a basic visual vocabulary — a person, a building, a thought bubble, an arrow, a question mark, a few container shapes — that you can draw fast and without thinking. Without that vocabulary, every drawing decision becomes a stall.
What the research actually shows
The empirical literature on note-taking is reasonably consistent across studies. The most cited paper is Mueller and Oppenheimer’s 2014 work showing that students who took handwritten notes outperformed those who typed, especially on conceptual questions, because the slower pace forced summarization rather than transcription. Bui, Myerson and Hale in 2013 found that organized note structures, Cornell and outline among them, outperform free-form notes on retention because the structure itself cues memory. Stacy and Cain in 2015 found that actively reviewing your notes within twenty-four hours improves retention by roughly thirty percent compared to only re-reading at exam time.
The pattern that emerges from these studies and the dozen or so others I have read alongside them is that organization plus active review beats any specific format. If you take Cornell notes but never go back to fill in the cue column, you are doing slow outline notes with extra steps. If you take messy outline notes but always do active recall on them within a day, you will outperform someone with beautiful unread Cornell notes every time. The method is the scaffolding for the behavior. The behavior is what actually moves retention.
Choosing for your situation
When a junior at work asks me which method to start with, I give them a short decision procedure. If it is a lecture or reading you will be tested on later, use Cornell — the review value is the highest of the four and the structure forces the behavior. If it is a meeting or a fast conversation where you need to keep up, use an outline because it is the fastest to capture. If it is brainstorming or essay planning where the relationships between ideas matter more than the sequence, use a mind map. If it is deeply conceptual or visual material that you want to internalize and re-read months later, sketchnote it. And if you genuinely do not know which category your material falls into, default to an outline, because it is the fastest to capture and easiest to review, and you can layer a Cornell cue column onto it later if it turns out to be worth the investment.
After a few years of experimenting, most of the people I know have ended up with the same hybrid I have. The bulk of notes goes into outline format because nothing beats it for raw capture speed. A Cornell-style left margin gets added afterward for cues and questions, which buys you most of the Cornell review value without the lecture-pace friction. Occasional mini-diagrams or rough sketches go in the middle of the page when a concept is genuinely visual. And a one-sentence summary at the bottom of each major section keeps the big picture from disappearing into the indentation. This hybrid works on paper, in Notion, in Obsidian, in OneNote, in plain markdown — the format is more important than the tool, and once you have the format the tool is just a vehicle.
Pairing notes with the right audio
For most note-taking, lofi works well as background because the auditory texture is steady enough not to interfere with listening to a lecture or processing dense reading. Anything with lyrics is a different story — lyrical music demonstrably interferes with verbal note-taking because the language centers of the brain end up shared between the lyrics and the words on your page. The science of ambient music post has the research on why, if you want to read the actual studies.
When I am writing design docs or processing a technical book, I leave the 24/7 lofi stream on in the background. It is calibrated to be low-arousal enough that I can sustain a sixty to ninety minute reading-and-noting block without my attention going brittle, which is roughly the longest stretch I can do good work before I genuinely need a break.
What to do tomorrow
If you currently take messy linear notes and never review them, the single highest-leverage upgrade is adding a Cornell-style cue column to whatever you already do. Five minutes spent filling in cues per page roughly doubles the long-term value of the notes, in my experience and in the literature, because it converts a passive transcript into an active recall surface. Try it for one week on whatever you are currently working on and notice how much more material you actually remember when you come back to the page.
My honest take, after a decade of fiddling with this, is that almost everyone is overthinking the method and underthinking the review. The best note-taking system in the world will not save you if you write the notes once and never look at them again, and a bad system reviewed daily will beat a great system you abandoned after a week. Pick whichever format you will actually use, build the twenty-four-hour review into your routine, and stop reading note-taking blog posts — including this one. You already know enough. Go fill in a cue column.
For the broader study system that surrounds note-taking, see our study techniques that pair with lofi post — note-taking is one piece of a larger workflow that includes active recall, spaced repetition, and structured time blocks.
References
Studies cited in this post:
- Mueller, P. A., & Oppenheimer, D. M. (2014). The Pen Is Mightier Than the Keyboard: Advantages of Longhand Over Laptop Note Taking. Psychological Science, 25(6), 1159-1168. The classic handwritten-vs-typed study.
- Bui, D. C., Myerson, J., & Hale, S. (2013). Note-taking with computers: Exploring alternative strategies for improved recall. Journal of Educational Psychology, 105(2), 299-309.
- Stacy, E. M., & Cain, J. (2015). Note-taking and handouts in the digital age. American Journal of Pharmaceutical Education, 79(7), 107.
