Our Spaced Repetition Algorithm

Flashka uses SM-2 as its scheduling backbone. SM-2 was introduced with the first computerised SuperMemo systems and remains the clearest, well-documented baseline for spaced repetition.

🏁 Core idea

Each concept is stored as a small “item.” Every item has its own ease value and review schedule. You study an item, rate how well you recalled it, and the algorithm computes the next interval. This per-item approach is the hallmark of SM-2.

How SM-2 schedules reviews

1. Initial ease (E-Factor) and early intervals

• Every item starts with an E-Factor of 2.5.

• The first two intervals are 1 day and 6 days.

• After that, the interval multiplies by the item’s E-Factor:
  I(1)=1, I(2)=6, I(n)=I(n-1)×EF for n≥3.
  These values come directly from the original SM-2 description.

2. Quality ratings (0–5) and ease updates
   After revealing the answer, you grade your recall on a 0 to 5 scale (0 = blackout, 5 = perfect). SM-2 updates the item’s E-Factor with a simple formula that increases ease after good recalls and lowers it after poor ones. The E-Factor is never allowed to drop below 1.3, which prevents intervals from collapsing.

3. Handling lapses
   If your quality rating is below 3, the item is considered forgotten. SM-2 restarts its learning steps (using the early intervals) while keeping the same ease value.

Why this works

SM-2 operationalised the spacing effect by adapting intervals to item difficulty (via E-Factor) and your current performance (via ratings). In SuperMemo’s own history, SM-2 marks the shift from fixed tables to data-driven, item-level optimisation