SM-2 Is Not Enough: Where Classic Spaced Repetition Breaks Down

SM-2 Is Not Enough: Where Classic Spaced Repetition Breaks Down

SM-2, the algorithm behind Anki since 1987, is still widely used. But it has a one-way ratchet that pins difficulty to 1.3 after lapses, causing permanent over-scheduling.

Why This Matters

SM-2 assumes a fixed exponential forgetting curve, ignores response latency, and treats cards as independent. For small decks (<200 cards) these flaws are invisible, but at 5000+ cards they cause ‘ease hell’ and permanent over-scheduling. Modern approaches like FSRS model difficulty, stability, and retrievability from user history, but require cold-start data before fitting.

Key Insights

• SM-2 algorithm published in 1987 (source: dev.to, 2026) uses a binary success/failure grade with no response latency signal.
• Ease hell concept: difficulty ratchet that never recovers; a card failed early stays near 1.3 floor even after learned.
• FSRS modern approach models three variables: difficulty, stability, retrievability, and fits to user history (SmartRecall).
• SM-2 treats cards independently, ignoring interference; confusable pairs can be scheduled adjacently by chance.

Working Examples

SM-2 algorithm implementation in Python for scheduling reviews.

if quality >= 3:
    interval = 1 if reps == 0 else 6 if reps == 1 else round(interval * ease)
    reps += 1
else:
    reps, interval = 0, 1
    ease = max(1.3, ease + (0.1 - (5-quality) * (0.08 + (5-quality) * 0.02)))

Practical Applications

• Use case: Anki with SM-2 for large decks (5000+ cards) suffers ease hell and suboptimal scheduling. Pitfall: Using SM-2 without adjustment leads to over-scheduling of previously difficult cards.
• Use case: FSRS for personalized retention targets explicit rate like 0.9. Pitfall: Requires hundreds of reviews before fitting, so cold-start is worse than SM-2.
• Use case: SmartRecall uses AI-generated cards with modern scheduling that avoids ease floor. Pitfall: Complexity may not be worth it for small decks (<200 cards).

References:

• https://dev.to/alexrchen/sm-2-is-not-enough-where-classic-spaced-repetition-breaks-down-4hdd