დასაწყისი · კატალოგი · პროფილი · ცხრილი
პროდუქტიულობა BODY HANDBOOK
პროდუქტიულობა · §704
Spaced Repetition
Most of what you learn is forgotten within a year. Spaced repetition is software that schedules flashcards to come back just before you'd lose them — you spend less time per fact than you would re-learning it later, and you keep the knowledge for decades instead of weeks. The basic effect has more than a century of lab evidence behind it, and the technique is now standard equipment for medical students, language learners, and anyone whose work depends on remembering things. The catch is the daily discipline — fifteen to thirty minutes a day, every day, for as long as you want to remember the material.
Do · Daily Evidence Strong თავი პროდუქტიულობა

The wins are real and the catch is the daily discipline. Commit twenty minutes a day and you keep most of what you study for years — language words, medical anatomy, legal cases, programming syntax, anything that fits on a card. The cost is daily contact, and the failure mode is the queue piling up the week you skip it. The evidence behind it is one of the deepest in learning science; the only real question is whether you need long-term recall enough to earn back the time.

You know how the language vocabulary you memorised for a high school exam is gone now? That's the math at work. Hermann Ebbinghaus measured it in 1885, and a careful 2015 replication reproduced his curve almost exactly — without rehearsal, about 60–70% of new facts vanish in the first day, and the curve keeps eroding for weeks Murre & Dros 2015. What spaced repetition does is hand you the same card again precisely when you'd be about to lose it. Each successful catch strengthens the memory more than the previous one did, and the next-due date moves out further. Five reviews over six months beats fifty re-reads in one weekend, for reasons documented across 184 articles and 317 separate experiments by 2006 Cepeda et al. 2006.

The active-recall half matters as much as the spacing. Re-reading a textbook page feels productive and barely is. Closing the book and trying to recall the page is what actually files it away.

One input sits underneath all of it: sleep. The review triggers consolidation, but the night is when it actually sets — so chronically short nights, the sleep-debt state, quietly blunt the very filing your cards are paying for.

What the long-term numbers look like

The most striking data comes from Harry Bahrick, who in 1984 tested 733 American adults on Spanish vocabulary — some of them had learned it in high school 50 years earlier. Beyond an initial 3–6 years of decline, words entered what he called permastore: a flat shelf of vocabulary that stayed accessible for 25-plus years without rehearsal. The strongest predictor of what was still there was how deeply it was learned the first time around Bahrick 1984. In a follow-up, 13 review sessions spaced 56 days apart produced the same five-year retention as 26 sessions spaced 14 days apart — half the work, same result Bahrick et al. 1993.

The cleanest real-world test is medical school. At Washington University, researchers controlled for MCAT scores, second-year grades, and test anxiety, then asked what additional study habits predicted USMLE Step 1 performance. Each additional 1,700 Anki cards a student had reviewed was independently worth one extra point on the exam Deng et al. 2015. A 2023 cohort study at the University of Minnesota found Anki users scored 6–13% higher across four medical school exams, including the closest available proxy for Step 1, and the benefit held regardless of underlying test-taking skill Wothe et al. 2023. Outside Anki, a randomised trial of spaced clinical-education emails — same effect, lower friction — showed retained knowledge in the intervention group versus controls Kerfoot et al. 2007.

The effect is not subtle. The question for anyone with a serious volume of facts to memorise is not whether to use spaced repetition but which tool to use.

How to actually do it

The dominant tool is Anki: free on Windows, Mac, Linux, and Android; $25 one-time on iPhone; free sync across devices. The developer doesn't run it as a business, so there's no advertising and no upsell. Pick Anki unless you have a specific reason not to.

The earlier you start, the better the compound. In the Minnesota data, students who began Anki in first year outperformed late starters even after controlling for total card count Wothe et al. 2023. The scheduler needs months of history to work well; jumping in two weeks before an exam gives you the testing-effect benefit without the spacing benefit.

What most guides get wrong

"More cards is more learning." Past your daily time ceiling, every extra card you add now is a review you owe yourself in two months. Cohorts with 30,000-plus cards underperforming peers with 10,000 well-curated ones is a recurring pattern in medical-school communities. The dose-response in the Deng paper is a population average, not a personal optimiser Deng et al. 2015.

"Anki teaches you the language." Anki holds onto facts you already learned somewhere else. Holding a conversation, writing a clinical note, debugging production code — these need reading, speaking, and problem-solving on top of the cards. Spaced repetition is the long-term storage layer; it isn't the learning by itself.

"I'll catch up on the weekend." The whole point of the scheduler is that it shows you each card at the moment closest to when you'd forget it. Skipping days pushes cards past that moment; you're now re-learning them more or less from scratch. The math doesn't reward the catch-up session.

Where it goes wrong

Almost every "I tried Anki and quit" story fits one of four patterns.

  • Review hell. Three days of life happens. You open the app to 540 cards due. The session feels like a punishment, you bail, and you don't come back. The fix is a low daily new-card cap — lower than feels heroic when you set it — so the queue never grows past what you can clear in twenty minutes on a normal day.
  • Card hoarding. The deck grows to thousands of cards the learner doesn't actually care about. Review time balloons, the important cards drown, and the streak becomes a chore. The fix is honesty about what's worth remembering, and willingness to delete.
  • Bad cards. A card whose cue is ambiguous ("Spanish: door") or whose answer is a list keeps failing — Anki flags these as leeches. They eat sessions. The fix is to rewrite or delete the moment a card has failed three or four times in a row.
  • Recognition, not recall. Studying a pre-made deck of facts you never properly encountered, and learning the wording of the card instead of the knowledge underneath. The fix is to do real first-pass learning — read the chapter, watch the lecture, write your own card — before the deck starts.

What else could you use

For casual learning — fifty Italian phrases for a holiday, a personal-interest topic — the overhead of card-making and daily review usually isn't worth the win. Duolingo, Babbel, and similar apps embed lighter spacing inside a friendlier shell; the ceiling is lower but you actually open them. SuperMemo is the original tool, paid, built on Piotr Wozniak's own algorithms — devoted following, steeper learning curve, harder to extend. RemNote and Obsidian-with-Anki integrations let cards live inside your notes rather than in a separate deck, which suits some workflows.

For high-stakes exams that test application more than recall — bar exam, math olympiads, the clinical Step 2 — spaced past-paper practice often beats card-based study. Same underlying machinery (spacing plus active retrieval), different unit of practice Pan & Rickard 2018.

What changes if you stick with it

The first thing you notice is at the two-week mark: cards you saw on day one come back and you actually remember them. It's a quiet shift but a real one. You used to study, forget, restudy, forget. Now you study, see the card again later, and it's there.

At six months, the deck holds a few thousand items you've genuinely retained. The foreign language you'd put down a decade ago is functional again. The textbook chapter you read in February is recallable in August. The clinical edge cases you crammed for last quarter's rotation are still available now. People around you notice the shift before you do — you stop reaching for your phone to look things up mid-conversation; the language sentence that used to stall by the third word now finishes.

At the year scale, Bahrick's permastore data starts to look less like a curiosity and more like a description of your own knowledge: things you reviewed enough have settled into a state where forgetting roughly stops Bahrick 1984Bahrick et al. 1993. The cost is the daily contact you've kept up. The win is a different relationship with what you've studied — what you put effort into remembering, you actually remember.

The broader study-skills literature — retrieval practice without software, the testing effect as a daily habit, the spacing effect inside curricula — works on the same machinery. Note-taking systems that pair with cards (Obsidian-Anki, RemNote) are worth a look if your work already lives in notes. Language learning beyond cards (comprehensible input, conversation practice) is the other half of the language story; this article is only the storage layer.

Related in the handbook
Helps
  • — The review cements a memory, but sleep is what files it — chronically short nights blunt the consolidation your cards depend on.
Related
  • — Both are daily-discipline productivity tools — one protects focus, the other protects memory.
  • — Reading is how you take material in; spaced repetition is how you keep it from leaking back out.
·
704