Downpat uses the SM-2 (SuperMemo 2) algorithm — one of the most well-known and thoroughly tested spaced repetition algorithms in the world. It was developed by Polish researcher Piotr Wozniak in 1987 and has since become the foundation for dozens of learning programs worldwide.
Scientific foundation
SM-2 is based on two fundamental discoveries in memory psychology:
1. Ebbinghaus Forgetting Curve (1885)
German psychologist Hermann Ebbinghaus experimentally proved that without repetition we forget up to 70% of new information within 24 hours. But each subsequent review slows the forgetting process — information transfers from short-term to long-term memory.
2. The Spacing Effect
Repetition with increasing intervals is significantly more effective than cramming. This has been confirmed by hundreds of scientific studies. The brain consolidates information better when there is a pause between reviews — this is when memory consolidation occurs.
How SM-2 works
The algorithm tracks three parameters for each card:
Interval — how many days until the next review. Starts at 1 day and grows with each successful review.
Repetition count — how many times you answered correctly in a row. If you make a mistake, the counter resets.
Easiness factor (EF) — a number from 1.3 to 2.5 that shows how easy this card is for you. The higher it is, the faster the intervals grow.
Algorithm in action
Imagine you are learning the English word "reluctant":
Day 1: You see the word for the first time and answer correctly. Interval = 1 day. It will appear again tomorrow.
Day 2: Review. You answer correctly. Interval = 6 days.
Day 8: Review. Correct again. EF = 2.5, so interval = 6 × 2.5 = 15 days.
Day 23: Review. Correct. Interval = 15 × 2.5 = 38 days.
Day 61: Review. Correct. Interval = 38 × 2.5 = 95 days (over 3 months!).
See? Just 5 reviews over 2 months — and the word is locked into long-term memory for years. No cramming, no stress.
What happens when you make a mistake
If you answer incorrectly or press "Don't know", the algorithm reacts like this:
• Repetition counter resets to 0
• Interval returns to 1 day
• Easiness factor decreases (but not below 1.3)
This means difficult cards appear more frequently, while easy ones appear less often. The algorithm adapts to your knowledge and focuses attention on exactly what needs improvement.
Answer quality rating
SM-2 rates each answer on a scale from 0 to 5:
5 — Perfect answer, without hesitation
4 — Correct, but with a slight delay
3 — Correct, but with difficulty
2 — Incorrect, but the answer seemed familiar
1 — Incorrect, but recalled upon seeing the correct answer
0 — Complete blackout, answer unknown
In Downpat, this rating is assigned automatically based on your answer: correct = 5, incorrect = 1, "Don't know" = 0. This simplifies the process — you don't need to self-assess your knowledge.
The SM-2 formula
For those interested in the details, here is the core interval calculation formula:
First review: interval = 1 day
Second review: interval = 6 days
Subsequent: interval = previous interval × EF
EF update after each answer:
EF' = EF + (0.1 − (5 − q) × (0.08 + (5 − q) × 0.02))
where q — quality rating (0–5)
If EF drops below 1.3, it is capped at 1.3 — this guarantees that intervals always grow, even for the most difficult cards.
Why SM-2 and not something newer?
Newer algorithms exist (SM-15, SM-18, FSRS), but SM-2 remains the gold standard for several reasons:
• Simplicity and transparency — easy to understand how it works and why it suggests a particular interval
• Decades of research — the algorithm has been tested by millions of users since 1987
• Effectiveness — the difference in results between SM-2 and more complex algorithms is minimal for most tasks
• Predictability — the user always understands why a card appeared at this moment
SM-2 is a time-tested tool that helps memorize information for the long term. That is why we chose it for Downpat.