NotesNudge the notes you never go back to

Spaced Repetition Beyond Cards

Spaced repetition software was designed for material with a right answer: vocabulary, dates, formulae, anatomy. You see a prompt, you produce the answer, you grade yourself, and the algorithm schedules the next showing.

Most of what is in a notes archive is not like that. Applying the machinery anyway is a common mistake, and abandoning the principle because the machinery does not fit is a more common one.

A related workplace concept is self-reporting bias, which offers a useful contrast with the personal knowledge-work problem discussed here.

Why the algorithm does not transfer

Grading requires a correct answer. The scheduling depends on you rating how well you recalled, and rating your recall of an argument about pricing is not a coherent operation. Without meaningful grades the algorithm is scheduling on noise.

Ideas do not have a retention target. A card exists so you will know a fact indefinitely. A note about a half-formed thought has no such goal, and scheduling implies one.

And most notes should expire. The algorithm assumes permanent value and schedules accordingly, which is exactly wrong for material that was true about a problem you no longer have.

For an independent external reference related to this topic, see RemNote.

The result of forcing it is a queue that grows, feels like an obligation, and gets abandoned — the same failure as an over-ambitious weekly review, arriving faster.

What does transfer

Three things, and they are the useful part.

Space it. Days rather than hours; weeks for anything you want to keep for months. You do not need an algorithm for that — you need not to review something the day after you wrote it.

Make it effortful. Attempt recall before looking. This is the mechanism, and it survives the loss of everything else.

And expect returns to fall off. Once something is genuinely part of how you think, further review adds little and the slot is better used elsewhere.

That is the principle without the software: spaced, effortful, and finite.

Where cards are actually right

Not an argument against flashcards, which are excellent for what they were built for.

Facts you want available cold. Vocabulary, key figures, definitions in a field you are learning, names.

Procedures with a fixed sequence.

And material you are being examined on, where the retention target is real and dated.

If your notes contain that kind of material, extract it into cards and use the software properly. Just do not put your thinking notes in the same system — the two need different treatment and mixing them ruins the queue.

The version that works for ideas

A small marked set instead of a scheduler. Perhaps forty notes you decided were load-bearing, reviewed a few at a time.

Rotation instead of intervals. Three or four per session, cycling through — which produces a real interval of a couple of months without anyone computing anything.

Retirement instead of graduation. When a note is genuinely absorbed, take it out of the set rather than pushing its interval to a year.

And a written line instead of a grade. Still true. No longer relevant. Connects to the thing I am doing now. More informative than a number and it is what makes the session produce something.

The honest position

Spaced repetition software is one of the few genuinely evidence-backed tools in this whole area, and its evidence is about the operation it performs on the material it was built for.

Applied to ideas, the algorithm's precision is false — the intervals in the research are not settled enough to license the schedules the software implies, and the grading it depends on has no meaning for a note about a half-formed thought.

Take the principle. Leave the machinery. The principle costs nothing and survives; the machinery imports an obligation that most people abandon within a season.

The short version