Spaced Repetition Algorithms

In 1885, German psychologist Hermann Ebbinghaus memorized thousands of nonsense syllables, words like WUX and ZOF, and then tested his own memory of them at set intervals. He found that without review, he forgot 58% of what he had learned within just 20 minutes, and over 70% within a day. But when he reviewed the syllables right before he was about to forget them, each review made the memory last dramatically longer before it needed reviewing again. That discovery, the forgetting curve, is the foundation of every spaced repetition app you have ever used, from Anki to Duolingo.

- What the Ebbinghaus forgetting curve shows about how memory decays - How the SM-2 algorithm (used in Anki) decides when to show you a flashcard again - The difference between massed practice (cramming) and spaced practice - How to build your own spaced-repetition study schedule for a real class

Ebbinghaus's data showed memory does not fade in a straight line, it drops steeply at first, then levels off. Without any review, you might retain only 25-30% of new information after a week. But each time you successfully recall something right as you are about to forget it, the curve flattens: the next review can wait longer than the last one. This is why reviewing material once, the night before a test, barely works, you are fighting the steepest part of the curve every time instead of ever flattening it.

SM-2, developed by Piotr Wozniak in 1987 for the SuperMemo program (and later adapted by Anki), assigns every flashcard an ease factor starting at 2.5. After you review a card, you rate how hard it was to recall, usually on a scale from again (you forgot it) to easy. If you recall a card correctly, the algorithm multiplies the previous interval by the ease factor, so a card you last saw 6 days ago, rated good, might not reappear for 15 days. If you fail a card, the interval resets to 1 day and the ease factor drops, so struggling cards come back sooner and more often than easy ones.

Massed practice, studying one topic in one long block like cramming the night before an exam, produces fast short-term gains but rapid forgetting, often losing most of the material within a week. Spaced practice, the same total study time broken into shorter sessions spread across days or weeks, produces slower initial learning but dramatically better long-term retention. A well-known 2006 study by Cepeda and colleagues, reviewing over 800 conditions across decades of research, found spaced practice outperformed massed practice in the vast majority of comparisons, sometimes doubling retention months later.