Spaced repetition has a mystique problem. It gets described like a brain hack, sold in apps with algorithms named after scientists, and wrapped in graphs of the forgetting curve. Underneath all of it is one scheduling rule: review a fact right before you'd forget it.
That's the whole trick. Everything else is implementation detail — and the details are where prep programs quietly turn into card maintenance jobs.
Forgetting is the feature
A review lands hardest when retrieval is difficult but succeeds. Review too soon and it's a rubber stamp — easy, fluent, and nearly worthless. Too late and you're relearning from zero. The productive moment is the edge of forgetting, where your brain has to reach. Effortful retrieval is what tells memory “this is worth keeping.”
That's why the intervals expand. Each successful reach extends how long the memory holds, so the next review can wait longer. Miss one, and the interval resets short — not as punishment, but because the miss revealed the memory was weaker than scheduled.
The intervals that matter
The exact numbers matter less than the shape: roughly doubling gaps, reset on failure. A perfectly serviceable schedule fits on an index card.
| Review | When |
|---|---|
| 1st | The next day |
| 2nd | Three days later |
| 3rd | One week later |
| 4th | Two weeks later |
| 5th | A month later — and at this point, it's probably yours |
Keeping the deck from eating your prep
The failure mode isn't the algorithm — it's the deck. Cards multiply, reviews balloon, and suddenly you're spending ninety minutes a day servicing a queue instead of studying. Three rules keep it honest.
- Cards come from misses, not chapters. A card earns its place by representing a question you got wrong. Carding entire chapters preemptively is how 2,000-card decks happen.
- Cap new cards at about twenty a day. Every new card is a recurring future obligation; the cap is what keeps week four's review load survivable.
- Suspend leeches. A card you've failed eight times isn't a memory problem, it's a understanding problem — the card goes on pause and the topic goes back into real study.