RecallHaven · Learn

FSRS vs SM-2, explained without the math anxiety

Both algorithms answer the same question: when should you see this material again so you remember it with the least effort? They answer it differently.

SM-2 in one paragraph

SM-2 was created by Piotr Woźniak and first implemented in his SuperMemo program in December 1987 (he later described it in his 1990 master's thesis). Each item has an ease factor. After a review you rate how well you recalled it; the first intervals are 1 day and 6 days, then each interval is multiplied by the ease factor (starting at 2.5, so about 15 days next). A failed recall restarts the intervals. Hard recalls lower the ease factor, which can't go below 1.3. It's simple, predictable and has been used for decades, which is why Anki's default ("legacy") scheduler is based on it.

Strengths: easy to understand and predict. Works fine with little data. Weaknesses: the multipliers are fixed rules, not fitted to you. Items that are hard early can get stuck in low-ease "ease hell" with too many reviews.

FSRS in one paragraph

FSRS (Free Spaced Repetition Scheduler) is an open-source algorithm developed by Jarrett Ye and the open-spaced-repetition community, building on the DSR memory model used at MaiMemo. It models each item's memory with difficulty, stability (how slowly you forget it) and retrievability (the chance you'd recall it right now). It schedules the next review for when retrievability drops to a desired retention you choose (Anki's default is 90%). Its parameters can be fitted to a person's review history. Anki added FSRS as an option in version 23.10 (AnkiMobile 23.10, AnkiWeb, and AnkiDroid 2.17+).

Strengths: adapts to how you actually forget, and lets you trade workload against retention directly. Weaknesses: less intuitive, and it benefits from review history to tune well.

Side by side

Which should a STEM student pick?

• Short course or exam in a few weeks, few reviews logged: either works. SM-2 is the predictable choice.
• A semester or longer, reviewing consistently: FSRS is worth trying, because it adapts as your history grows.
• Don't switch back and forth every week. Whichever you pick, consistent review matters more than the algorithm.

Independent benchmarks published by the FSRS project compare FSRS with SM-2 on large public review datasets. If you want the numbers, read the open-spaced-repetition benchmark on GitHub rather than trusting any summary, including this one.

How RecallHaven uses them

In RecallHaven you choose FSRS adaptive or SM-2 adaptive in settings. You review a note's quiz section by section, and when a section is done the app schedules its next review with the algorithm you picked. Due sections show up in Today, the review queue and the calendar.

FAQ

Is FSRS better than SM-2? It's designed to be more personalized and is often more efficient with enough history, but "better" depends on your data and consistency. See the open-spaced-repetition benchmark for published comparisons.

What's a good target retention for FSRS? 90% is Anki's default. Higher means more reviews, and the Anki manual recommends staying below 97%.

Do I need to understand the algorithm to use it? No. Answer honestly and review when things are due. In RecallHaven your section score feeds the scheduler.

Can I change algorithms later? In RecallHaven it's a setting, so yes.

Want to try FSRS or SM-2 on your own course notes? RecallHaven's free plan includes automatic spaced repetition. Try it free →