The memory algorithm behind Anki, Tomadora, and every flashcard app worth using — explained without the jargon.
Long-term memory is not a function of how long you study. It is a function of how often you nearly forget and successfully recall. Each successful "near-forget recall" strengthens the memory more than reading the material five times in a row. This is why cramming works for the test on Friday but loses 90% of what you learned by Monday. Spaced repetition algorithms time your reviews to hit that near-forget moment as often as possible.
In 1885 Hermann Ebbinghaus measured how much of a list of nonsense syllables he forgot over time. The result is a steep exponential decay: after 20 minutes you remember about 60%, after 1 hour 50%, after 1 day 30%, after 6 days 25%. Reviewing right at the moment of near-forgetting flattens that curve dramatically. Each successful review pushes the next one further out — the gap can grow from 1 day to 6 days to 16 days to 35 days, etc.
SM-2 was developed by Piotr Wozniak in 1985-1990 for SuperMemo and is still the dominant practical algorithm. After every card, you grade your recall on a 1-5 scale: 1 (Again) means total fail, 2 (Hard), 3 (OK), 4 (Good), 5 (Easy). The algorithm uses your grade to update three numbers per card: an "ease factor" (how easy the card is for you), an "interval" (how many days until the next review), and a "repetition count" (how many times you have correctly recalled it). A grade of 1 resets the card to day 1. A grade of 3+ multiplies the interval by the ease factor. Over time, easy cards drift to 6-month intervals; hard cards stay daily.
Myth 1: "More cards is better." False. Spaced repetition tax scales with deck size. Most people fail because they make 2,000-card decks they can never sustain. Aim for the smallest possible deck that covers the material. Myth 2: "I should always re-read the textbook before reviewing." False. Active recall (try first, look up second) is far stronger than passive re-reading. Myth 3: "Spaced repetition makes you a genius." False. It only protects what you already understood. If your initial encoding was bad, no algorithm rescues it.
Three rules. (1) Do reviews every day, even if briefly. The algorithm assumes consistency. Skip a day and you owe double tomorrow. (2) Be honest with the grade. Hitting "Easy" because you do not feel like reviewing later breaks the algorithm. (3) Add cards slowly. 10-20 new cards a day is sustainable. 100 new cards a day will collapse your deck within 2 weeks. Tools that fit this naturally: Anki (the original, requires a separate ritual), Tomadora (uses the same SM-2 but interleaves reviews into your existing 5-minute Pomodoro breaks so there is no separate ritual to maintain).
Spaced repetition is the closest thing to a free lunch in learning. The hard part is sustaining a daily review habit. Most people quit Anki by week 6 because the standalone ritual fails. Bind your reviews to a habit you already have — like the Pomodoro break — and the algorithm does the rest.
Download Tomadora — free →