SM-2 vs FSRS
SM-2 is the original spaced-repetition algorithm from the 1980s; FSRS (the Free Spaced Repetition Scheduler) is its modern, data-driven successor. FSRS predicts memory more accurately, so it schedules fewer reviews for the same retention — and it is the algorithm EverFlip uses.
SM-2, designed by Piotr Woźniak for the original SuperMemo, is a fixed formula: each card has an "ease factor" that grows or shrinks based on your ratings, and intervals are computed from it. It was a breakthrough and still powers many apps, but it treats every learner and every card with the same hard-coded curve.
FSRS instead models memory with three quantities — difficulty, stability, and retrievability — fitted from a very large dataset of real reviews. Because it predicts the forgetting curve more accurately, it can schedule each review closer to the true point of forgetting: fewer reviews for the same retention, or higher retention for the same effort. The open-source benchmark behind it spans hundreds of millions of reviews, and Anki adopted FSRS as a built-in option in 2023.
In practice, the difference learners feel is workload: a more accurate scheduler stops showing you cards you clearly still know, so the daily review pile is smaller without sacrificing what you remember.
How EverFlip does this
EverFlip uses FSRS — the modern algorithm — out of the box, with no configuration. You get the smaller, better-timed review schedule automatically, the same engine Anki offers but with nothing to install.