Notes that survive the term

The test of a note

A note is good if, three months later, it reminds you of something you understood. That is the only criterion, and it rules out most of what students produce: transcriptions, highlighted PDFs, and AI summaries.

The failure of the first two is old. A transcript is a recording of a lecture happening, not of you understanding it. Highlighting marks text as important and produces recognition, which is why it consistently rates among the least effective study methods when tested.

The failure of the third is new and worth naming precisely. An AI summary is a note nobody encoded. Compression is the work. A summary you did not write is a summary someone else’s process performed, and what you retain from reading it is roughly what you retain from reading anything: a feeling of familiarity, and no structure of your own to hang the material on.

Notes in your own words, always

The rule underneath everything here: write it as you would say it to a friend. If a sentence in your notes could have been copied from the source, it probably was, and it will teach you nothing when you return to it.

This also solves a practical problem. Notes containing copied sentences are how accidental plagiarism happens: three weeks later you cannot tell which sentences were yours. Paraphrasing at note-taking time removes the risk entirely, and it makes the note better. Two problems, one habit.

If you must keep an exact quotation, mark it unmistakably: quotation marks, the page number, and something visual you cannot miss.

A structure that works

For a lecture or a reading, four sections:

• Question. What is this trying to answer? One line, written before or immediately after.
• Core. The three to five things that would let you reconstruct the rest. Not everything.
• Confusion. What you did not follow. This is the most valuable section and the one everyone omits. It becomes next session’s opening.
• Connection. What this links to elsewhere in the course, or contradicts.

Then, within a day, close the notes and write the summary from memory in three sentences. That is retrieval, it takes two minutes, and it converts a passive record into a study session.

Where the model belongs in note-taking

Not as the writer. As the interrogator.

• After you write. Paste your notes: what have I misunderstood, what is missing, and what would somebody who knew this subject add? Diagnosis on your work.
• Turning notes into questions. Turn these notes into fifteen questions, a third of which require applying the idea to a new case. Then answer them from memory.
• Filling a named gap. You know what you missed; ask about that specifically, with your notes pasted for context.
• Cards. Generating Anki cards from your notes, checked against the source before they enter the deck.

The boundary is straightforward: the model works on notes you wrote. It does not write them.

Recorded lectures and transcripts

Automatic transcription is useful for accessibility, for second-language students, and for going back to the one thing you missed. Whisper runs free and locally, and most platforms now transcribe automatically.

The trap is that a transcript feels like notes and is not. It is the lecture, in text, unprocessed. Use it to check a specific point, or as a source to make notes from. Never as the notes themselves: a folder of transcripts at revision time is a folder of lectures you now have to attend again.

Tools, all free

Obsidian: free for personal use, plain markdown files you own, works offline, links between notes. Joplin: open source and free, similar, syncs. Plain markdown files in a folder: free, portable, will still open in twenty years. Anki for the retrieval layer.

The tool matters much less than people spend time believing. Elaborate note systems become a productivity hobby that substitutes for studying. Plain files, four sections, own words, reviewed by retrieval.

What to keep

At the end of a module you should have three things: your notes, your confusion list with resolutions, and your unaided-check log from Module 3. Those three, together, are a genuine account of what you learned and how. They are also the process record that Module 7 is about, and unlike anything you could assemble after the fact, they exist already, because you made them while working.