The forgetting curve
The forgetting curve describes how quickly we lose newly learned information: memory decays fastest in the first hours and days, then more slowly. Each well-timed review flattens the curve, so the same fact is forgotten more and more slowly over time.
Hermann Ebbinghaus first mapped the forgetting curve in the 1880s by testing his own memory of nonsense syllables over time. He found that without review, recall drops sharply at first — much is lost within a day — then tapers off.
The practical insight is not the decay itself but what happens when you review at the right moment: each timely review resets and flattens the curve, so the next decay is slower. After several well-spaced reviews, a fact can stay retrievable for months or years with almost no further effort.
A spaced-repetition algorithm is essentially a forgetting-curve predictor: it estimates when each specific card is about to drop below recall threshold and schedules the review for just before that point — never too early (wasteful) or too late (already forgotten).
How EverFlip does this
EverFlip’s FSRS engine models a forgetting curve per card, using your own rating history, and schedules each review at the point of maximum benefit. Your /account stats show your real retention — the curve, flattened by your own practice.