FSRS vs SM-2: Why We Use the Next Generation Algorithm

Learning That Grows With You

Language Games is built on a simple idea: learning should be personalised and social. We believe the best way to grow your skills is with a system that understands exactly where you are—and friends who keep you accountable.

That's why we track your progress across all four skills—reading, writing, listening, and speaking—separately. Your vocabulary adapts to your level, conversations use words you've actually studied, and study groups help you stay motivated.

But none of this works without the right foundation. At the heart of our system is FSRS—a modern algorithm that uses game design principles and cognitive science to show you exactly what you need, exactly when you need it.

Game Design Meets Science

Great games keep you in the "flow state"—challenged enough to stay engaged, but never so hard you give up. That's exactly what FSRS does for learning.

Unlike old-school flashcard systems that treat every learner the same, FSRS predicts your personal forgetting curve for each word. Easy words get spaced out quickly. Tricky words get more practice. You're never bored, never overwhelmed.

The result? You spend less time reviewing and more time actually using your new language—whether that's in conversations, games, or real life.

The History of Spaced Repetition

For over 30 years, SM-2 (SuperMemo 2) has been the gold standard for spaced repetition algorithms. Developed by Piotr Wozniak in 1987, it was revolutionary for its time. Every major flashcard app (Anki, Mnemosyne, and countless others) has used SM-2 or a variant of it.

But science doesn't stand still. In 2022, researcher Jarrett Ye introduced FSRS (Free Spaced Repetition Scheduler), an algorithm built on modern machine learning research and millions of real review data points. The results speak for themselves.

Retention Over Time

The most important metric: how well do you actually remember what you learn? FSRS maintains significantly higher retention rates over long periods.

Fewer Reviews, Same Results

FSRS requires 30-50% fewer reviews to achieve the same retention. This means less time studying and more time living.

Smarter Interval Scheduling

SM-2 uses a fixed formula that doesn't adapt to individual learning patterns. FSRS uses 17 parameters trained on real data to predict the optimal review interval for each card.

Key Technical Differences