What is spaced repetition?

The direct answer is that spaced repetition — reviewing material at increasing intervals over time, rather than all at once — is genuinely, robustly supported by learning science, with roots going back to Hermann Ebbinghaus's 1885 studies of the forgetting curve and strong modern confirmation: a large meta-analysis of 317 experiments (Cepeda et al.) found spaced practice reliably outperforms massed (crammed) practice, and a 2025 meta-analysis found a substantial effect size (d=0.54) favoring distributed over massed practice. This is one of the more solidly evidenced claims in learning science, not a trendy productivity idea with thin backing. It's worth contrasting this with several other popular learning claims discussed elsewhere on this site, some of which — like micro-learning specifically — rest on considerably thinner and less consistent evidence than spacing does.

The mechanism: why spacing beats cramming

Ebbinghaus's original 1885 work established the "forgetting curve" — the observation that newly learned information decays predictably over time unless reinforced — and the core insight behind spaced repetition is that reviewing material just as it's about to be forgotten, rather than while it's still fresh, forces a small amount of effortful retrieval that strengthens the memory more than an easy review of material you already clearly remember. Reviewing something too soon, while it's still fully accessible, provides little of this retrieval effort and correspondingly little benefit; reviewing it too late, after it's fully forgotten, means starting over rather than reinforcing anything. Spaced repetition systems are built around estimating that specific, individual-varying sweet spot and scheduling the next review right around it. Ebbinghaus's original methodology was famously rigorous for its era — he tested himself repeatedly over years using nonsense syllables specifically designed to have no prior associations, which controlled for a confound (prior knowledge) that would otherwise have made his forgetting-curve results harder to interpret cleanly. It's also worth knowing Ebbinghaus published his findings well over a century before modern statistical methods existed, yet the core shape of his forgetting curve has held up remarkably well against later, more rigorous replications using modern methodology — a rare case of a finding this old still being considered foundational rather than superseded.

How the modern algorithm actually works

The most widely used practical implementation traces to Piotr Wozniak's 1985 SuperMemo algorithm, which calculates an expanding interval for each individual piece of material based on how easily it was recalled at the last review — get something right easily, the next interval expands further; get it wrong or hesitate, the interval shrinks back — producing a personalized review schedule rather than a fixed one-size-fits-all timetable. Most modern spaced-repetition software (Anki and similar tools) uses a descendant of this same underlying approach. The strength of the evidence here comes specifically from the spacing itself, not from any particular app or algorithm brand — the mechanism was well-supported before software existed to automate it, and the software is a practical implementation of an already-established finding, not the source of the finding. It's also worth knowing that the specific interval-lengthening algorithm varies somewhat between different spaced-repetition tools, and no single formula has been shown to be definitively superior to close variants — the core mechanism (expanding intervals based on recall difficulty) is what the evidence supports, not any one specific mathematical implementation of it. It's also worth noting that most spaced-repetition tools let you adjust the underlying algorithm's aggressiveness manually, which is worth experimenting with if a default schedule consistently feels either too easy or too demanding for a particular subject.

How strong the evidence actually is, compared to other learning claims

It's worth being specific about how well-evidenced this particular claim is relative to others discussed on this site: Cepeda et al.'s review synthesized 317 individual experiments, which is a substantial evidence base by the standards of learning science, and the more recent 2025 meta-analysis's effect size (d=0.54) is a moderate-to-large effect in the context of educational research, where much smaller effects are common. This puts spaced repetition on considerably firmer ground than several more loosely evidenced learning trends — worth contrasting with our guide on micro-learning, where the underlying research is meaningfully thinner despite superficially similar marketing. This level of evidence is also part of why spaced repetition shows up so consistently in language-learning apps, medical education, and other fields with a heavy factual-recall component — it's one of the rare learning techniques where the underlying research is strong enough that adoption isn't really a matter of ongoing scientific debate. It's also genuinely one of the few learning techniques where following the evidence and following common practice among high-performing students and professionals in fields with heavy memorization requirements point in exactly the same direction.

Where Mr. Grummel fits

Mr. Grummel's own quiz mechanism draws real, if informal, benefit from this same underlying principle — repeated exposure to questions across sessions functions as a form of spaced retrieval practice, though it's a byproduct of the blocking mechanism, not designed as a dedicated study tool in the way a purpose-built spaced-repetition app is. None of this makes Mr. Grummel a substitute for a dedicated spaced-repetition tool — the quiz bank isn't built or scheduled around individual recall difficulty the way a purpose-built system like Anki is — but the underlying principle is a genuine, if secondary, benefit of encountering the same kind of question repeatedly over separated sessions. For anyone specifically wanting the full, deliberate benefit of spaced repetition for actual studying, a dedicated tool built around it remains the better choice — Mr. Grummel's version is a pleasant side effect of the blocking mechanism, not a replacement for one.