How reviewing works

The four buttons, the intervals, and why you can't break it.

Rate by recall speed, not by difficulty

When a card is revealed, ask yourself: how fast did the answer come? Pick the button that matches. The system uses your answer to schedule the next time you'll see this card — automatically.

You don't need to think about intervals — the system has already done the math.

What each button actually does

ButtonMeansEffect on next interval
AgainBlanked outReset — see it again in ~1 minute
HardRecalled but it hurtShorter than expected — system thinks you need more practice
GoodRecalled normallyStandard growth (roughly 2× the previous interval)
EasyInstant, no effortLarger growth (roughly 3.5×) — stop bothering you with this one

Why intervals are dynamic

Each card carries two hidden numbers — stability (how long you'll likely remember it) and difficulty (how hard it is for you). Stability grows when you recall successfully and resets when you blank. The next interval is calculated to catch the card just before you'd otherwise forget it — that's the point at which review is most efficient.

This is the FSRS algorithm (a successor to Anki's SM-2). The four-button design is the industry standard for spaced repetition; we use the ts-fsrs library directly.

You can't break this

  • Tapping the wrong button doesn't ruin progress — the affected card just relearns. Other cards are untouched.
  • If you really meant something else, press U to undo the last rating.
  • Don't optimize. Pick the button that matches how you actually felt; the math handles the rest.
Back to vocab study →