The spacing effect, in plain English
A hundred and forty years of research says the same thing: study it, forget a bit, study it again. Here's how far apart the gaps should be.
In 1885 Hermann Ebbinghaus spent months memorising nonsense syllables and testing himself on the results. He found two things that have survived every attempt to knock them down since.
The first is that forgetting is fast and curved: you lose a lot in the first day, then progressively less. The second is more useful — the same total study time produces far more retention when it is spread out than when it is crammed together.
What "spaced" actually means
Not "study more often." Study at increasing intervals, timed so each review lands while the memory is still there but has started to fade.
Review too early and it's free — you still remember, nothing is strengthened, you've spent the time for very little. Review too late and it's gone, and you're relearning from scratch. The useful window is the awkward one in between, where you almost have it and have to dig.
A conventional ladder for something you want to keep for a term:
| Review | When | | --- | --- | | 1st | Later the same day | | 2nd | Next day | | 3rd | Three days later | | 4th | A week later | | 5th | Two to three weeks later | | 6th | A month or more |
Those numbers are a starting point, not a law. The right gap depends on how well you did last time, which is why fixed schedules underperform adaptive ones.
Why cramming still works, briefly
Cramming isn't magic-free — it genuinely works for about 48 hours, which is why it survives as a strategy. Students who cram often do fine on Friday's exam. The problem shows up in the follow-up: on a delayed test weeks later, spaced learners retain dramatically more from the same hours of work.
If your only goal is passing Friday and never touching the subject again, cramming is a defensible use of your time. For anything cumulative — a qualification, a language, a subject where term two depends on term one — it quietly costs you the whole thing twice.
The bit nobody mentions
Spaced repetition is not hard to understand. It's hard to run.
To do it properly you have to know, for every individual fact, how well you did last time and therefore when it's next due. With twenty facts that's a notebook. With two thousand it's a database, and the moment you miss a few days the whole schedule needs recomputing.
This is precisely the kind of bookkeeping that computers are good at and people are not. It's also why the standard advice — "use spaced repetition!" — has such a low conversion rate into people actually doing it. The technique is free; the admin is what costs you.
What we do with it
DuckEars schedules from your answers rather than from a fixed table. Miss something and it comes back tomorrow. Get it three times running and it disappears for a month. Skip four days and the plan recompresses around your deadline instead of dumping the backlog on you at once — because the failure mode we saw most in testing wasn't people forgetting the material, it was people opening the app to 60 overdue cards and never opening it again.
Further reading: Cepeda et al. (2006) is the meta-analysis worth reading on optimal gaps; Ebbinghaus's original 1885 monograph is short and surprisingly readable.