Spaced Repetition Explained: Why FSRS Beats SM-2

Have you ever crammed for an exam, only to forget everything a week later? You're not alone. Hermann Ebbinghaus documented this phenomenon in 1885 — his forgetting curve showed that we lose roughly 56% of newly learned material within one hour, and up to 75% within a week, without deliberate review (Ebbinghaus, 1885; replicated by Murre & Dros, 2015).

Spaced repetition is the solution. And the algorithm behind it matters more than most learners realize.

What Is Spaced Repetition?

Spaced repetition is a learning technique that schedules review sessions at increasing intervals. Instead of reviewing everything daily, you review each piece of information just before you're likely to forget it.

The core insight comes from Cepeda et al. (2006), who analyzed 254 studies involving over 14,000 participants and found that distributing practice across time consistently outperformed massed practice (cramming) — often by 100% or more on long-term retention tests.

Here's a simplified example of how intervals grow:

| Review # | Interval | Your recall | | -------- | ---------- | ---------------- | | 1st | 1 minute | You just saw it | | 2nd | 10 minutes | Starting to fade | | 3rd | 1 day | Needs effort | | 4th | 3 days | Getting easier | | 5th | 1 week | More confident | | 6th | 3 weeks | Nearly automatic |

Each successful recall strengthens the memory trace, allowing longer intervals between reviews.

The SM-2 Algorithm: A Pioneer With Limitations

For decades, most flashcard apps used SM-2 (SuperMemo Algorithm 2), created by Piotr Wozniak in 1987. SM-2 was revolutionary — it brought spaced repetition to software for the first time.

SM-2 works by assigning each card an "easiness factor" that adjusts based on your ratings. But it has significant limitations:

• One-size-fits-all: SM-2 uses the same starting intervals for every learner, regardless of their memory capacity
• Fixed model: It assumes everyone forgets at the same rate
• No personalization: Your success rate on 1,000 cards doesn't inform scheduling for card 1,001
• Brittle easiness factor: A few missed reviews can trap cards in a cycle of overly short intervals

Wozniak himself acknowledged these issues and went on to develop SM-17 and SM-18, but Anki — the most popular flashcard app — remained on SM-2 for over 15 years.

FSRS: The Next Generation

FSRS (Free Spaced Repetition Scheduler) was developed by Jarrett Ye and the open-source community as a modern replacement. In 2023, Anki officially adopted FSRS as an alternative to SM-2.

FSRS is built on the DSR (Difficulty, Stability, Retrievability) memory model, which tracks three parameters for each card:

• Difficulty: How inherently hard this card is for you
• Stability: How long your memory of this card will last before dropping below a threshold
• Retrievability: The probability that you can recall this card right now

What Makes FSRS Better

FSRS uses machine learning to optimize its parameters based on your actual review history. The results are measurable:

Fewer unnecessary reviews. A benchmark study by Ye et al. (2024) compared FSRS against SM-2 across millions of review records from Anki users. FSRS achieved the same retention rate with fewer total reviews — meaning you spend less time on cards you already know.

Personalized scheduling. FSRS learns your individual forgetting curves. If you tend to forget vocabulary faster than grammar, FSRS adapts. If you're stronger in the morning, the intervals reflect that over time.

Better handling of lapses. When you forget a card, SM-2 resets to a short interval. FSRS takes a more nuanced approach, considering how long you held the card before forgetting and your overall pattern of lapses.

How Spaced Repetition Works in Practice

When you review a flashcard on iwill.study, you rate your recall on a four-point scale:

• Again — you forgot it completely
• Hard — you struggled but eventually recalled it
• Good — you remembered with moderate effort
• Easy — it was effortless

FSRS uses this rating, combined with the card's full review history and your personalized parameters, to calculate the optimal time to show you that card again.

The process looks like this:

1. See a new card → FSRS sets an initial interval based on your learning rate
2. Rate your recall → the algorithm updates the card's stability and difficulty
3. Card is scheduled → it appears again at the calculated optimal moment
4. Repeat → intervals grow as the memory strengthens

Over weeks and months, well-learned cards might appear only once every few months, while challenging ones are reviewed more frequently.

Spaced Repetition for Language Learning

Spaced repetition is particularly effective for language learning because vocabulary acquisition requires exactly the kind of repeated, distributed retrieval practice that the technique provides.

Research by Nakata (2015) in The Modern Language Journal found that spaced retrieval practice produced significantly better long-term vocabulary retention than massed practice for second-language learners — even when total study time was held constant.

You can also explore how many words you need to speak a language, learn about how your brain stores new words, and discover why active recall doubles learning speed.

Getting Started With Spaced Repetition

The best time to start spaced repetition is now. The algorithm gets smarter the more you use it, so even 10 minutes a day builds a powerful learning habit.

1. Pick a deck — start with a language pack or create your own cards
2. Study daily — consistency matters more than session length
3. Trust the algorithm — if FSRS says a card is due, review it
4. Be honest with ratings — accurate self-assessment helps FSRS personalize your schedule

References

• Cepeda, N. J., Pashler, H., Vul, E., Wixted, J. T., & Rohrer, D. (2006). Distributed practice in verbal recall tasks. Review of General Psychology, 10(4), 354-380.
• Ebbinghaus, H. (1885). Memory: A Contribution to Experimental Psychology.
• Murre, J. M. J., & Dros, J. (2015). Replication and analysis of Ebbinghaus' forgetting curve. PLOS ONE, 10(7).
• Nakata, T. (2015). Effects of feedback timing on second language vocabulary learning. The Modern Language Journal, 99(4), 631-649.
• Ye, J. et al. (2024). FSRS: A modern spaced repetition algorithm. Open-source benchmark, available at github.com/open-spaced-repetition.