Pillar guide
Spaced Repetition, Explained Without the Hype
The memory science behind Anki, WaniKani, and Inku. Why it works, where it breaks, and how to use it without wasting your time.
The forgetting curve
In 1885 Hermann Ebbinghaus ran one of the most boring experiments in psychology history. He memorized lists of nonsense syllables, then tested himself at intervals ranging from 20 minutes to a month. He plotted the results.
What he found is now called the forgetting curve. If you learn something once and never see it again, your retention drops fast. Within 20 minutes you have lost roughly 40 percent. Within a day, about 60 percent. Within a week, over 75 percent. The curve is steep, it is consistent, and it applies to almost every kind of declarative knowledge.
Ebbinghaus ran a second experiment. He memorized the same material multiple times at different intervals, and measured how much time he saved each subsequent session. This is the relearning curve, and it is the foundation of every spaced repetition system ever built.
The core insight
If you review a fact right before you would forget it, the review is cheap (you still remember it) and the next review can be scheduled further out. If you review too late (after you have forgotten it), the review is expensive (you have to relearn from scratch). Spaced repetition is the practice of finding the sweet spot: reviews that are timed so the material is still in your head, but barely.
How SRS actually works
A spaced repetition system is, at heart, a queue of cards and a scheduler. You answer each card with a confidence rating (roughly "I forgot" / "I remembered, barely" / "I knew it easily"). The scheduler uses that rating to compute when to show you the card next. Easy cards move out further. Cards you struggled with come back sooner.
The maths vary, but they all share three moving parts:
- A stability estimate: how long this particular card can sit before you forget it. Stability grows with each successful review.
- A difficulty estimate: how hard this card is for you specifically. Cards you fail often are marked harder and reviewed more aggressively.
- A target retention: the probability you want to maintain that you will remember each card at review time. Anki's classic default is 90 percent. Modern FSRS lets you choose.
SM-2, FSRS, and what Inku uses
SM-2: the classic algorithm
Anki runs on SM-2, an algorithm from 1987 by Piotr Woźniak. It is simple and proven, but it has two main problems. First, the interval calculation is a rigid formula that doesn't adapt to how individual learners actually forget. Second, it tends to over-review easy material and under-review material in the "interference zone" where you know the shape of a fact but not the details.
FSRS: what replaced it
FSRS (Free Spaced Repetition Scheduler) was developed by Jarrett Ye and became Anki's default in 2024. It uses a three-variable memory model (stability, difficulty, retrievability) and fits the parameters to each learner's actual history. In practice, this produces 10 to 20 percent fewer reviews for the same retention, sometimes more.
FSRS also exposes a single tunable knob (desired retention) so you can trade off time against rigour. If you want 95 percent retention (expensive), you can set it. If you want 80 percent retention (cheap, lots of forgetting), you can set that too.
What Inku uses
Inku uses FSRS under the hood with a target retention around 90 percent tuned for adult Japanese learners. Cards the app flags as frequent interferers (the confusable kana pairs, for example) get small scheduler bumps so they surface together. You do not see any of this, and there is no knob to tune, on purpose.
Why Inku hides the scheduler
Anki's UI exposes every dial, which is great for tinkerers and terrible for beginners. Most learners who burn out on Anki do so because they felt the machinery. Inku's first design rule was that the user should feel a short stack of cards and nothing else. The scheduler should vanish.
Doing SRS right
Keep new cards low
A new card today becomes one review tomorrow, one three days later, one after ten days, and so on. Adding 20 new cards today adds roughly 60 future reviews over the next month. Most beginners burn out because they front-loaded 40 new cards a day for two weeks, then faced a 400-card review queue on day 15.
For adult beginners in Japanese, 10 new cards a day is a sustainable pace. You can push it to 20 if you are a student with time. Twenty-five is the rare ceiling.
Rate yourself honestly
When you answer a card, the scheduler trusts your self-rating. If you lie ("I knew that" when you did not) you poison the scheduler's model of your memory. Results: bad intervals, more reviews, worse retention. Answer honestly, especially the first week.
Make it daily, not weekly
SRS rewards consistency more than intensity. Ten minutes every day beats 70 minutes on Saturday. The scheduler assumes daily touch; if you deviate, it overcompensates.
Short sessions are fine
A 5-minute session is still a session. If you are scared of the queue, do five cards. The scheduler will be happier than if you skip entirely.
The four common mistakes
Mistake 1: too many new cards
The single most common way to burn out. Feelgood early (new cards are easy) becomes panic at week three. Cap yourself.
Mistake 2: low-quality cards
A card with a vague cue or an ambiguous answer wastes the scheduler. For vocabulary, each card should have one clear prompt and one clear answer. For kanji, be specific about what you are testing (reading? meaning? writing?). Using a pre-built deck like Inku's spares you from this altogether.
Mistake 3: skipping for a week
When you skip, the queue grows geometrically. When you come back, the queue feels punishing, so you skip more. The cycle is real. During busy weeks, set new cards to zero and just clear the daily review; don't stop entirely.
Mistake 4: gaming the ratings
“I just want this to go away” rating. If you mark everything easy, the intervals stretch past what you can hold, and in two months you have forgotten everything. The ratings are an input to a machine that works for you. Feed it truthfully.
When SRS is the wrong tool
SRS is great for discrete facts: vocabulary, kanji readings, grammar pattern endings. It is bad for:
- Fluency in a skill (speaking, listening, writing sentences). These need practice, not review.
- Conceptual understanding.You can't SRS your way to understanding when to use は vs が. Understand first, then use SRS to drill example sentences.
- Motivation. SRS is a memory tool, not a motivation system. If you hate it, you will quit. Use it with content you care about.
From a learner
“I spent two years doing Anki for 45 minutes a day. I passed N3. I also never spoke Japanese to anyone. SRS is necessary, not sufficient.”
Common questions
What is spaced repetition in one sentence?+
Spaced repetition is a study technique where you review information at expanding intervals (for example, one day, three days, ten days) so you rehearse each fact right before you would forget it.
Does spaced repetition actually work?+
Yes. It is the most studied memory technique in cognitive psychology. Fifty-plus years of research show that spaced reviews produce retention 2 to 3 times higher than massed practice (cramming) for the same total time.
What is FSRS and why does it matter?+
FSRS (Free Spaced Repetition Scheduler) is a modern scheduling algorithm that replaces the older SM-2 used in Anki for decades. It predicts when you'll forget each card based on your own history, instead of using a one-size-fits-all curve. Most people see 10 to 20 percent fewer reviews for the same retention.
How many cards a day is too many?+
For most adult learners, 20 to 30 new cards a day is the ceiling before the review burden explodes. Ten new cards a day is sustainable indefinitely. New cards multiply your future review load more than they add to today's load.
Should I leave spaced repetition alone if I'm busy?+
You can skip a day or three without breaking the system, but going quiet for weeks will cause a flood when you return. The least painful strategy during a busy period is to reduce new cards to zero (not stop reviewing) until the busy stretch ends.
Related: how to learn Japanese from scratch, Inku vs Anki, or the specific SRS mistake that wastes 90 percent of learners' time. If you want the SRS machinery done for you, download Inku and let the scheduler stay out of your way.