Spaced repetition explained with Keeplet

Spaced repetition is a study technique that shows you a piece of information again at ever longer intervals, just before you forget it. Keeplet applies it to every flashcard (called a "card" in the app) with the SM-2 algorithm: you rate how well you remembered, and Keeplet decides on its own when to show the card again, on the website and in the app for Android and iPhone.

What is spaced repetition

Studying a topic only once, even for a long time, leaves a memory that fades quickly. Spaced repetition changes how you distribute your study: instead of rereading everything in a single session, you revisit each piece of information at several moments, letting more time pass between one review and the next.

The idea is simple:

1. You learn something new.
2. You review it after a short time.
3. If you remember it, you review it after a longer time.
4. If you don't remember it, you start again from short intervals.

Every successful review makes the memory more stable, so the interval can grow: from a few days to weeks, then months.

Why it works: the forgetting curve and active recall

In 1885 the German psychologist Hermann Ebbinghaus published Über das Gedächtnis, in which he measured how quickly we forget what we have learned. The result is the forgetting curve: without review, a substantial part of what you learn is lost in the first days, then the loss slows down. Every time you review, the curve flattens and the memory lasts longer.

Two effects studied by psychology explain why the method is effective:

• Spacing effect. For the same amount of study time, spreading reviews over time produces better long-term retention than concentrating them. This is the conclusion of a review by Cepeda and colleagues published in Psychological Bulletin in 2006.
• Retrieval effect (testing effect). Trying to recall an answer, rather than rereading it, strengthens memory more than simple rereading. Karpicke and Roediger showed this in a paper published in Science in 2008.

Spaced repetition combines the two effects: it makes you retrieve the information, and it does so at the right moment.

The right moment: not too early, not too late

Reviewing too early is wasted time, because the memory is still fresh. Reviewing too late means starting over, because the information is already gone. The ideal interval is the point where remembering takes a little effort. Calculating it by hand for hundreds of cards is impractical: that is why an algorithm is used.

How Keeplet applies it

Keeplet uses a version of SM-2 (SuperMemo 2), a review-scheduling algorithm created in the late 1980s. In brief, here is how it works:

• Every card has a next-review date. When it arrives, the card appears among those "to review".
• After flipping the card, you rate it with four ratings: Not at all, A little, Somewhat (website) / Fairly well (app), Great (website) / Perfectly (app).
• The higher the rating, the further away the next review will be. With Not at all the card restarts from short intervals.
• Every card has its own "ease factor" that starts at 2.5 and adapts to your answers: cards you remember well move further away, difficult ones come back often.
• Keeplet also takes punctuality into account, with a late penalty that the original SM-2 does not have: if you review long after the due date you do not advance a level, even if you remember.

The precise values (ratings, intervals, ease factor) are in The SM-2 algorithm in Keeplet. With an account, the server calculates the schedule; the Android and iPhone app repeats the calculation on the phone to show you the preview under the ratings and to let you review offline (without an account, the calculation happens only on the phone).

What you do every day

You don't have to plan anything. Every day:

1. In the app, open the Today screen and tap Review now: the full steps are here. On the website, you start the review from the menu: see How to do a review session in Keeplet.
2. Rate each card honestly.
3. Done: the cards reschedule themselves.

The number of cards per day follows your daily goal, which by default is 15. The app estimates 20 seconds per card: 15 cards take about 5 minutes.

How to get the most from the method

• One card, one piece of information. Short, precise cards are easier to rate. Keeplet accepts up to 1000 characters for each question and answer, but that is not a target to reach.
• Rate honestly. If you always give the highest rating, cards come back too late.
• Be regular. A few minutes every day works better than one long session now and then. A reminder helps: see How to set a review reminder in the app.
• Create cards from your notes. In the app you highlight a sentence, on the website you turn a note into a flashcard: see How to create a card by highlighting text and How to create a flashcard from a note in Keeplet.

Frequently asked questions

What is the difference between spaced repetition and normal review?

In normal review you reread everything at the same time. In spaced repetition each piece of information has its own calendar: it comes back only when needed, so you spend less time on what you already know and more on what you forget.

How much time should I spend every day?

It depends on how many cards are due. With the default goal of 15 cards a day, the app estimates about 5 minutes. You can change the goal from the Profile in the app.

What happens if I skip a day?

The cards stay due and you find them again the next day. A little lateness costs nothing; but if you review after more than 1.5 times the planned interval, even remembering you do not advance a level, and after more than 2 times you go back one level: see The SM-2 algorithm in Keeplet.

Does Keeplet use the FSRS algorithm?

No: today Keeplet uses SM-2 and offers no alternative algorithms. The full picture of what Keeplet does and does not do is in the Keeplet fact sheet.

Is spaced repetition good for every subject?

It suits anything you need to remember precisely over time: definitions, dates, formulas, vocabulary, regulations. It works less well for practical skills, which are trained by doing.