SM-2 vs FSRS-5: Which Spaced-Repetition Scheduler Belongs in a Corporate LMS

Why Corporate Training Needs a Scheduler at All

Hermann Ebbinghaus measured his own forgetting in 1885 and found that it is steep at first and then flattens. Murre and Dros repeated the experiment in 2015 and found the same shape (PLOS ONE, https://doi.org/10.1371/journal.pone.0120644). The practical consequence for training is simple: a single course followed by nothing leaves most of the work to the steep part of the curve.

Spaced retrieval answers this by asking the learner to recall a fact again just as it starts to fade. The scheduler is the part that decides when "just as it starts to fade" is, for each card and each learner.

How SM-2 Schedules a Card

SM-2 was published by Piotr Woźniak in 1987 for the SuperMemo program, and many flashcard tools still use it or a variant. After each review the learner grades recall from 0 to 5. Each card keeps an ease factor, starting at 2.5 and never below 1.3, which rises after easy answers and falls after hard ones.

The first interval is 1 day and the second is 6 days; after that, each interval is the previous one multiplied by the ease factor. A grade below 3 sends the card back to the start. SM-2 is easy to explain and to implement, but its first steps are fixed and it has no notion of a target recall rate.

How FSRS-5 Schedules a Card, and What Personal Academy Runs

FSRS (Free Spaced Repetition Scheduler) is an open-source scheduler that models memory directly. Each card has a stability, the number of days after which the chance of recall falls to 90%, and a difficulty from 1 to 10. After each review, rated again, hard, good or easy, both numbers are updated with 19 parameters, and the next review is set for the day the predicted recall reaches the target.

Personal Academy schedules its daily review cards with FSRS-5, computed in the database (supabase/migrations/20261007801600_aop_amb07_fsrs.sql; lib/engage/nuggets.ts). It uses the 19 published FSRS-5 default parameters, and the target recall is an organisation setting (learning.fsrs_desired_retention) that defaults to 90% and can be set between 70% and 97%. A higher target means more frequent reviews; a lower one means fewer.

Which Should a Corporate LMS Use?

For a small card deck reviewed by motivated learners, SM-2 works and is easy to audit by hand. For a workforce where review time is scarce and topics differ in difficulty, a scheduler that aims at an explicit recall target lets L&D trade review time against recall in one setting, which is the reason Personal Academy runs FSRS-5.