The forgetting curve
In 1885 Hermann Ebbinghaus tested his own memory on lists of nonsense syllables and found something that has held up ever since: memory fades on a predictable curve. You forget most of what you just learned within days, unless something interrupts the fade. That interruption is a review.
The insight is about timing. Review a word the minute after you learn it and you gain almost nothing; it was still fresh. Review it a month later and it is gone; you are relearning from scratch. Review it just before it slips, and you get the biggest strengthening for the least effort. The next safe gap then gets longer.
Study the same thing at the right growing intervals, and each review buys you a longer memory than the last.
Why spacing beats cramming
Cramming packs all your reviews into one session. It feels efficient, and it works for about a day. Spacing the same number of reviews across days or weeks produces dramatically longer retention for the identical study time. This is the spacing effect, and it is not a small edge.
A synthesis of 254 experiments found spaced practice reliably outperforms massed study.Cepeda, Pashler, Vul, Wixted & Rohrer (2006)
The practical upshot: the same hour of study, sliced into short daily sessions, teaches you far more than the hour spent all at once.
Recalling beats re-reading
There is a second effect stacked on top. Retrieving a word from memory, actually trying to recall it before you see the answer, strengthens it much more than simply re-reading it. Re-reading feels like learning because the word looks familiar, but familiarity is not recall. A good system always asks you to remember first, then reveals the answer.
Recalling a word strengthens it far more than re-reading it.Roediger & Karpicke (2006)
What FSRS is
You cannot track by hand when 400 separate words are each about to fade. That is a scheduler's job. Early systems used fixed rules of thumb (the classic SM-2: multiply the interval by a set factor each time you pass). The modern successor is FSRS, the Free Spaced Repetition Scheduler. It builds a little model of how likely you are to recall each card and picks the next review date for the moment your memory is predicted to drop to a target level. Because it learns from your actual hits and misses, it is more accurate than fixed-interval schemes, which means fewer reviews for the same retention.
Smisli uses FSRS under the hood. You just rate how a word felt (again, hard, good, easy) and the scheduler does the timing.
How to use it for a language
Vocabulary is the perfect target for spaced repetition: thousands of small items, each either recalled or not. The method fits it exactly. In practice:
- Save words as you meet them, ideally inside a real sentence so the card carries context.
- Do your due reviews daily, briefly. The scheduler decides which cards; you decide only how each felt.
- Trust the intervals. A card you know well won't reappear for weeks. That is the system saving your time, not neglecting the word.
That is the whole engine behind Smisli: you read Swedish you enjoy, tap the words you want, and FSRS makes sure they stay. If you want the vocabulary side of the story, see how many words it takes to be fluent.