Three stages, three different jobs
| Stage | When | What good looks like |
|---|---|---|
| Capture | In the lecture | Fast, ugly, incomplete. Gaps marked, not filled. |
| Process | Same day or same week | Structured, selective, in your own words. Half the length. |
| Convert | Immediately after processing | Questions. The only form that survives to the exam. |
Most students do stage one and stop. A smaller number do one and two. The third stage is what distinguishes notes that get used from notes that get filed, and it takes ten minutes.
Capture: optimise for speed and gaps
- Write less than you think. Verbatim capture crowds out understanding, and the research on this is fairly consistent: transcribing produces worse comprehension than selective note-taking.
- Mark gaps with a symbol rather than stopping. A `?` in the margin means "I lost this" and takes half a second. Stopping to resolve it costs you the next two minutes of the lecture.
- Capture the lecturer's asides. "This comes up every year" and "don't worry about the detail here" are worth more than three slides of content and appear in no deck.
- Don't copy what's on the slide. You have the slide. Write what isn't on it — that's the whole value of being in the room.
- Photograph the board rather than copying a complex diagram badly while missing the explanation.
Process: the slot that everything depends on
- 1
Work from capture plus slides, same week
Ideally the same day, while the spoken half of the lecture is still in memory. After about a week you're reconstructing from a textbook, which is slower and gives you the textbook's framing rather than your examiner's.
- 2
Write full sentences with the causal words restored
Because, therefore, unless, only when. Bullet fragments drop exactly those, and they're the content. This is also where you'll discover which parts you didn't actually understand.
- 3
Resolve the gap symbols
Each `?` gets closed — from the slides, the recording, a textbook, or a note to ask in the next session. Unresolved gaps compound; three weeks of them makes a module feel impossible.
- 4
Mark the examinable claims
Five per lecture, starred. You're deciding this while the lecturer's emphasis is fresh, which revision-you cannot do.
- 5
Cut ruthlessly
Processed notes should be shorter than what you captured. If they're longer, you've expanded rather than processed, and you've built something you won't re-read.
Which method for which subject
| Subject shape | Capture method | Why |
|---|---|---|
| Definition-heavy (law, medicine, humanities theory) | Cornell or two-column | The cue column is a question bank in embryo. |
| Problem-based (maths, physics, engineering) | Worked examples with your own annotations | You need the method and the decision, not prose. |
| Process and mechanism (biology, physiology) | Diagrams with arrows | The causal chain is the content and lists lose direction. |
| Argument-based (history, philosophy, literature) | Claim–evidence–objection structure | You're capturing a position, not facts. |
| Fast, dense, slide-driven | Annotate the slides directly | Don't compete with the deck; supplement it. |
Paper or digital
Handwriting is generally better for comprehension, mostly because it's slower and forces selection — you physically cannot transcribe, so you have to decide. Digital is better for searching, backing up and reorganising, and it's the only realistic option for heavy diagram-plus-equation subjects on some setups.
The pragmatic answer most people land on: handwrite in lectures where you can, photograph or scan afterwards, and process digitally. That gets the comprehension benefit of handwriting and the searchability of digital, at the cost of one photo per page. Note-taking apps vs paper covers the trade-offs in more detail.
The systems that don't scale
- Rewriting notes neatly. Feels productive, is transcription, teaches nothing. This is the single most common wasted study activity.
- Elaborate linking schemes across five modules of coursework. Zettelkasten is powerful for long-running research and heavy machinery for a lecture course.
- One giant document per module. Unnavigable by week six and never reopened.
- Colour systems with more than three colours. The taxonomy becomes a decision you make at capture time, which is what breaks capture.
- Notes you never convert into questions. The most common failure, and the most costly, because everything upstream of it was wasted effort.
Don't run it evenly across five modules
The full three-stage treatment for five modules is around eight hours a week, which most students don't have, and attempting it uniformly is why the system dies in week five rather than being scaled down deliberately.
Rank your modules by exam weight and difficulty. Run the full system on two, capture-and-questions on two more, and slides-only on the fifth. An uneven system you maintain all term is worth several times a complete one you abandon in October — the same triage logic that governs the weekly review.