Spaced Repetition, Explained — Review Schedules That Beat the Forgetting Curve

The forgetting curve — we lose most of what we learn

In 1885, psychologist Hermann Ebbinghaus measured how fast memory decays, using himself as the test subject. The results were brutal: nearly 40% of new material vanished within 20 minutes, and about two-thirds was gone after a single day.

His experiments used nonsense syllables, so real coursework fades more slowly — but the shape of the curve is the same. Unreviewed memories collapse fast. If you've ever thought "I knew this yesterday, why is it gone?", you were standing in the middle of the forgetting curve.

How spaced repetition works — recall at the edge of forgetting

Ebbinghaus also found the antidote. Each time you relearn material, the forgetting curve gets flatter. Repeat the cycle a few times and the memory effectively crosses into long-term storage.

Timing is everything. Reviewing while a memory is still fresh does little. Retrieving it right before it slips away — when recall takes real effort — strengthens it the most. Cognitive scientists call this a "desirable difficulty."

So spaced repetition stacks two of the best-proven effects in learning science: retrieval practice (recalling without looking at the answer) and the spacing effect (stretching the gaps between reviews).

The review schedule that actually works

A common starting point is expanding intervals: once on the day you learn something, then after 1 day → 3 days → 7 days → 14 days → a month, roughly doubling each time. Treat those numbers as a starting point, not a law.

The one rule that matters: lengthen the interval for cards you answer easily, and reset it for cards you miss. That principle powers SRS (spaced repetition software) algorithms like SM-2 — and its analog ancestor, the Leitner box.

• Leitner box: cards promote through boxes 1→5 and drop back to box 1 when missed — higher boxes get longer intervals.
• SRS algorithms: each card's difficulty is tracked and its next review date computed automatically — you just answer today's queue.
• The shared principle: the schedule lives in a system, not in your head.

Flashcards — the natural unit of spaced repetition

Spaced repetition works best when "what to review" is broken into small pieces. That's why the question-front, answer-back flashcard became the standard tool: one card equals one rep of retrieval practice.

A few rules keep your cards effective.

• One concept per card — if the answer runs past three lines, split the card.
• Make questions self-contained — not "What are its advantages?" but "Why does spaced repetition beat cramming for long-term retention?"
• Write both directions (term→definition and definition→term) to double your retrieval paths.

Why doing it by hand fails

The theory is simple; the practice breaks down for two reasons. First, with even three or four subjects, just tracking what's due today becomes a job of its own. Second, making cards eats the time you meant to spend studying — turning 50 pages of lecture slides into a good deck is hours of work by itself.

Sustainable spaced repetition is a tooling problem. Hand card creation and scheduling to a system, and keep your own effort where it counts: the retrieval practice itself.