What is FSRS (Free Spaced Repetition Scheduler)?
The Free Spaced Repetition Scheduler (FSRS) is an open-source algorithm that decides when each review card should be shown again. It models memory for every card with three values, stability, difficulty and retrievability, and schedules the next review for when the predicted chance of recall falls to a target level, adjusting after every answer.
How Personal Academy handles it
Personal Academy schedules daily review cards with FSRS-5, keeping stability and difficulty for each card, on the learner review page on the Advanced plan.
A worked example
Suppose a 300-person airline trains cabin crew on emergency procedures. Two crew members review the same card about door operation. One answers easily each time, so the scheduler raises the card stability for them and waits weeks before showing it again. The other hesitates and sometimes misses it, so the card stays difficult for them and returns within days until their recall improves.
For decades the most widely used scheduling rule in flashcard software was a variant of SuperMemo 2, published in 1987. It gives each card an ease factor and multiplies the interval by it after each correct answer. It is simple, but it does not model forgetting directly and can schedule reviews too early or too late.
FSRS was developed in the early 2020s by Jarrett Ye and contributors as an open-source alternative, based on the DSR model of memory: difficulty, stability and retrievability. Stability is the time it takes for the probability of recall to fall to a set level. Difficulty describes how hard the card is for this learner. Retrievability is the current probability of recall, which falls as time passes since the last review.
After each review, FSRS updates stability and difficulty based on the answer grade (again, hard, good or easy) and on how long it had been since the previous review. A successful recall when retrievability was low increases stability more than one when it was high, reflecting the finding that harder retrieval strengthens memory more.
The scheduler then chooses the next interval so that retrievability will reach the desired retention target, often around 90 percent, when the card is next shown. A higher target means more frequent reviews; a lower one means fewer reviews but more forgetting.
FSRS has gone through several versions, with FSRS-5 refining how same-day reviews and initial difficulty are handled. Its parameters can be fitted to a learner review history, though default parameters work reasonably for new users.