At the heart of Scholium is FSRS (Free Spaced Repetition Scheduler), an open-source algorithm that predicts when you're about to forget something. Here's how it works.
The Forgetting Curve
Hermann Ebbinghaus discovered in 1885 that memory decays predictably over time. This "forgetting curve" follows an exponential pattern—you forget rapidly at first, then more slowly over time.
The key insight: if you review something just before you forget it, you strengthen the memory and extend the curve.
Why FSRS?
Traditional spaced repetition algorithms like SM-2 (used by Anki) use fixed multipliers. If you get a card right, the interval doubles. This is simple but not optimal.
FSRS uses machine learning to model how memory actually works. It tracks four parameters for each card:
- Stability (S): How long until you have a 90% chance of forgetting
- Difficulty (D): How hard this card is for you personally
- Retrievability (R): Your current probability of recall
- State: Whether the card is new, learning, or mature
Personalization
FSRS adapts to your learning patterns. If you consistently remember cards longer than predicted, the algorithm adjusts. If certain types of material are harder for you, it notices and schedules more reviews.
The Results
Studies comparing FSRS to SM-2 show significant improvements:
- 30% reduction in total review time
- More accurate prediction of forgetting
- Better long-term retention
FSRS is what makes Scholium's "study less, remember more" promise possible.