Spaced repetition, explained

Spaced repetition rests on a handful of findings from memory research that have held up for a long time. You don’t need them to use Askesis, but they explain why it asks you to study the way it does.

#The forgetting curve

In the 1880s Hermann Ebbinghaus learned lists of nonsense syllables and tested himself after different delays. He forgot fast at first and then more slowly: most of the loss came in the first day or two. Later studies found the same shape for all kinds of material.

The useful part is what a review does. Each successful recall makes the memory more durable, so the next decline is slower. A card you remembered after 3 days may hold for 8 the next time, then 20, then a couple of months. That is the pattern in the intervals Askesis gives (an example).

#Active recall

Pulling an answer out of memory strengthens it more than reading it again. This is the testing effect (or retrieval practice), one of the most consistently replicated results in learning research. Rereading feels productive because the material looks familiar, but recognizing something isn’t the same as recalling it.

That’s why a card shows only its front at first. Produce the answer, in your head or out loud, before you reveal it. Flipping the card and nodding along loses most of the benefit.

#Spacing

The same number of reviews works better spread over days than packed into one sitting. This spacing effect is why cramming can get you through tomorrow’s test but fades soon after.

Spaced repetition adds one idea: let the gaps grow as the memory gets stronger. A card you know well comes up a few times a year; a shaky one comes back tomorrow. Your time goes to the cards that need it, which is what keeps a large collection manageable.

#Desirable difficulty

Recall that takes some effort does more for memory than recall that’s effortless. The psychologist Robert Bjork called this a desirable difficulty. A review shortly before you would forget is harder than one an hour after learning, and it is worth more.

So the scheduler doesn’t try to make every review easy. Aiming for about 90% recall means roughly one review in ten ends in Again. That is by design: a schedule where you never forget would have you reviewing far more often than you need to.

#What your answers mean

Your grade is all the scheduler learns from, so each button has a specific meaning:

Hard is a pass, not a fail: if you recalled it, even slowly, choose Hard or Good. Hard, Good and Easy earn the same XP (XP), so there is nothing to gain by grading up. The exact rules are in How scheduling works.

#Why the due day matters

A card’s due date is the scheduler’s estimate of when your chance of recalling it drops to about the target. Reviewing around that day gets you the most memory for your time:

• Too early: the memory is still strong, so the review is easy and does little. It costs time and earns less.
• Too late: the chance that you’ve forgotten keeps rising. A forgotten review card goes back to relearning with a short interval, so a lapse costs more than one extra review.
• On time: hard enough to count, likely enough to succeed.

A few days late isn’t a disaster. When you still remember an overdue card, Askesis counts the extra time in your favor: with the default scheduler, Good adds half the delay to the next interval and Easy adds all of it. The real cost of skipped days is the pile that builds up, which is why a short daily session beats a long weekly one. See Handling a backlog.

#Retention and workload

The share of cards you aim to remember sets how much you review. Aim higher and cards come back sooner, so you do more reviews to forget less, and the returns shrink near the top. With FSRS’s default parameters, a card that comes back in 30 days at 90% comes back in about 12 days at 95%: well over twice the reviews to halve the forgetting.

The default scheduler has no target to set; you steer it with its ease and interval options. The retention charts in Askesis compare against 90%, a common goal. With FSRS you choose the target yourself, as desired retention. How to pick one: FSRS and desired retention.

#What it won’t do

• It won’t make sense of things for you. Understand a topic first, then make cards for what you want to keep. Memorizing what you don’t understand is slow and fragile.
• It can’t fix a bad card. A vague or overloaded card stays hard however it’s scheduled. See Writing good cards.
• It needs regular, short sessions. A few minutes most days works far better than an hour once a week.