Spaced Repetition: The One Learning Technique Worth the Hype

Most study advice is contested. Spaced repetition isn't. It has been replicated for over a century, across ages, subjects, and formats, and it consistently produces large improvements in long-term retention over the alternatives.

It's also widely misused, mostly by people who turn it into an elaborate hobby that produces less learning than the fifteen-minute version would have.

The finding

Reviewing material at increasing intervals produces far better long-term retention than reviewing the same material the same number of times in a block.

Same total minutes. Different distribution. Substantially different outcome.

Study something today, review it tomorrow, then in three days, then in a week, then in three weeks. Each successful recall pushes the next review further out, because the memory has become more durable.

The mechanism is straightforward once stated: memory strengthens most when retrieval is difficult but successful. Review something five minutes after learning it and recall is effortless — nothing is being strengthened, because nothing was at risk. Review it three days later, when you have to genuinely reach for it, and the act of reaching is what consolidates it.

This is why cramming works for Friday and fails for December. Massed practice produces high performance immediately and steep forgetting after.

Why it feels wrong

Spaced practice feels worse than blocked practice while you're doing it. You forget things between sessions. You get answers wrong. It's uncomfortable and slow.

Blocked practice — twenty problems of the same type, back to back — feels smooth and competent. You're getting them right. You feel like you're learning.

The research is unambiguous that the second feeling is misleading. Performance during practice and retention afterward are close to inversely related. Anything that makes practice feel easier usually makes learning worse.

Most people optimize for the feeling, which is the single most common way study time gets wasted.

What it's actually for

Spaced repetition handles retrieval of things you already understand. That's a narrower claim than the enthusiasm around it suggests, and getting the scope right matters:

Excellent for: vocabulary, anatomy, legal citations, drug interactions, formulas, keyboard shortcuts, names, dates, syntax, technical definitions — any large body of discrete facts you need available instantly.

Poor for: understanding, judgment, reasoning, synthesis. A deck can make you fluent at recalling that a mutex prevents race conditions and give you no ability whatsoever to reason about a concurrency bug.

This is the most common failure mode: people make cards for things they don't yet understand, then grind reviews trying to memorize their way into comprehension. It doesn't work in that direction. Understand first, then retain with spacing. Cards preserve knowledge; they don't create it.

Doing it without building a second job

The failure pattern is predictable. Someone discovers spaced repetition, makes 900 cards in a week, faces a 300-card daily queue by week three, and quits.

Rules that prevent this:

One idea per card. If the answer has three parts, it's three cards. Multi-part cards fail unpredictably — you get two-thirds right, and there's no honest way to grade that.

Make cards for things you already understand. If you can't explain it without the card, you're not ready to memorize it.

Cap new cards. Ten or twenty a day. Every new card is a recurring obligation for months; the daily review load settles at roughly ten times your daily new-card rate. Twenty a day is a sustainable habit. A hundred a day is a future you'll abandon.

Delete aggressively. A card you've failed six times is a bad card, not a character flaw. Rewrite it or bin it.

Cloze over prose. "The capital of Australia is [...]" works better than a question-and-answer pair, because it forces a specific retrieval rather than a vague sense of the topic.

Don't import other people's decks. Making the card is part of the learning. A downloaded 5,000-card deck is someone else's understanding, and it mostly generates guilt.

Without software

You don't need an app. The Leitner box version works and takes no setup:

Five piles. New cards start in pile 1. Get one right, it moves up a pile. Get it wrong, it goes back to pile 1. Review pile 1 daily, pile 2 every other day, pile 3 weekly, pile 4 biweekly, pile 5 monthly.

Same principle, no software, no streak to break. For a few hundred facts this is genuinely as good as anything.

There's a lighter version that costs almost nothing and that most people would benefit from more than a card system:

At the end of any learning session, write down three questions the session should have made you able to answer. Tomorrow, answer them from memory before looking. In a week, again. Ten minutes total, and it captures most of the effect.

The essential part is from memory. Looking at your notes and nodding is re-reading with extra ceremony.

Combining it properly

Spaced repetition pairs well with the things it doesn't overlap with:

Understanding first. Get taught the material — by a person, a book, a course, an AI teacher — until you can explain the mechanism out loud. Then build cards for the facts worth keeping.

Application alongside. Cards keep knowledge available. Using it keeps it usable. Those are different, and you need both.

Explanation as review. Occasionally replace a review session with explaining the topic out loud to nobody. It's a harder retrieval than a card and it tests the connective tissue cards can't reach.

A workable overall shape: learn a topic properly, make ten cards for the facts, apply it to something real within a day, review the cards on a schedule, and once a month explain the whole area out loud from scratch.

The honest summary

Spaced repetition is the most reliable technique in learning science and it solves exactly one problem: forgetting facts you once knew.

That's a real problem and worth solving. It is not the same problem as understanding a subject, and treating it as such is how people end up with 2,000 cards and no ability to use the material.

Learn it properly first — Voca exists for that part, teaching a generated curriculum out loud and checking whether it landed. Then let spacing keep it. The two do different jobs, and the sequence matters.