Spaced Repetition, Explained

Most exam preparation fails in a specific, predictable way: material learned in week two is gone by week ten, so the final month becomes a panicked re-learning of things that already felt mastered once. That is not a discipline problem. Forgetting follows a curve — steep at first, flattening over time — that memory researchers have measured for over a century, and rereading does very little to change its shape.

Spaced repetition attacks the curve directly. Instead of reviewing everything on a fixed rotation, it schedules each fact for review just before you would probably forget it. Every successful recall flattens that item's forgetting curve, which earns it a longer gap before the next review. The practical result is that your daily review load stays roughly constant even as the number of things you know keeps growing.

Two well-established findings do the heavy lifting here: the spacing effect, which shows that the same minutes of study produce more durable memory when they are spread out rather than massed together, and the testing effect, which shows that actively retrieving an answer strengthens memory far more than re-exposing yourself to it. A spaced-repetition system is simply software that applies both findings to every card you own, every day.

When you review a card, you grade your own recall, and the schedule responds. A confident, correct answer stretches the interval — a card seen today might return in three days, then a week, then three weeks, then two months. A failed recall shrinks the interval back down so the card returns quickly while it is fragile. Over months, easy material drifts toward rare, cheap check-ins while stubborn material keeps getting the frequent attention it needs.

You do not need to understand the algorithm's arithmetic to benefit from it, but one mental model helps: the system is constantly sorting your entire collection by fragility and showing you only the slice that is about to slip. That is why a 2,000-card collection can cost twenty minutes a day while a 200-card collection reviewed haphazardly can still leak.

The single most common way people break spaced repetition is flattering themselves. Flip the card, see the answer, feel a warm sense of familiarity, and press Good — even though, two seconds earlier, the answer was not actually there. Familiarity is not recall. The exam will ask you to produce the answer with the card face down, so that is the standard your rating has to reflect.

A useful discipline: say the answer out loud, or write it, before flipping. If what you produced does not match, the rating is Again, full stop. This feels slower and stricter than it needs to be, and it is precisely why it works — every dishonest Good quietly schedules a future exam question you will miss.

The scheduler can only be as good as the cards you feed it. The cards that survive months of intervals share a shape: one question, one answer, minimal context needed, and phrased so that a correct answer is unambiguous. A card that asks 'Describe the cardiovascular system' cannot be graded honestly; a card that asks one retrievable fact can.

Write cards in your own words when the material allows it. The act of reformulating is itself encoding, and your own phrasing carries retrieval cues a pasted sentence never will. Where a list truly must be memorized, break it into one card per element plus one card for the count, rather than a single card you will half-recall forever.

Spaced repetition is a maintenance system, not a complete study method. First contact with material — a lecture, a chapter, a practice passage — still happens outside it. The workflow that holds up across every exam type: learn something, immediately convert the parts worth keeping into cards, and let the scheduler own everything after that. Your planning energy then goes to choosing what to learn next, not to deciding what to re-review.

Timing matters enormously. Intervals need room to grow, so the system pays compound interest: cards made three months out cost a fraction of what the same cards cost when made three weeks out. If your exam is close, spaced repetition still helps, but it behaves more like organized cramming — set expectations accordingly and lean harder on practice questions.

Protect a daily slot, and keep it honest with a floor rather than a ceiling: reviews first, every day, even if that is all you do. On a good day, ten minutes of due cards leaves the rest of your block free for new learning and practice tests. Skipped days do not vanish — they stack tomorrow's queue, which is how systems die.

Nearly every abandoned system fails one of the same five ways. Knowing them in advance is most of the cure.

You do not need to design a perfect system before starting; you need seven honest days. Day one: pick a single topic you studied recently and write ten atomic cards. Days two through seven: clear your due queue every day — it will be tiny — and add five to fifteen new cards from whatever you studied that day. By the end of the week you will have felt the core loop: material you converted into cards stays retrievable, and material you merely read starts slipping. That contrast, felt once, is more persuasive than any article.

From there, scale the same loop. The system's demands grow slowly and predictably, and the daily review count becomes a surprisingly honest dashboard of how much you actually know.