Scheduling built on FSRS, not fixed intervals

Older spaced-repetition algorithms (like the classic SM-2) use a fixed interval ladder. FSRS models your actual memory instead.

Every review updates two numbers for that card: stability (roughly, how many days until you'd have a 90% chance of still recalling it) and difficulty (how hard that specific fact tends to be for you). Your next review date is calculated from those, not from a fixed sequence like "1 day, then 3, then 7."

A card you find easy gets spaced out further; one you keep missing comes back sooner and its difficulty rises. Same-day reviews (a card you just relearned minutes ago) use a different, short-term update than a review a week later — the two situations aren't the same memory event, and treating them identically is a common gap in simpler implementations.

This is the same algorithm family used by current-generation Anki as an option, applied here as the default. See how cards get generated in the first place, or decay radar for what happens with that stability number beyond just scheduling the next review.