Study Smarter,
Not Longer.

In the 1880s the German psychologist Hermann Ebbinghaus did something unusual: he used himself as the experimental subject. He memorised lists of meaningless syllables, then measured how many he could relearn after various delays. His results, published in 1885, described a pattern that has been replicated in many forms since. Memory loss is steep at first and then levels off. A large share of what you learn can slip away within days if you do nothing with it.

The finding is often drawn as a falling curve, and it is often over-simplified. The exact shape depends on how meaningful the material is, how well it was encoded, how rested the learner was, and what they already knew. Meaningful material connected to existing knowledge decays more slowly than arbitrary syllables. The curve is best read as a pattern about unreviewed, weakly encoded information, not a fixed law with fixed percentages.

The practical lesson is the one that matters. Each time you successfully retrieve a memory, the trace is strengthened and the next period of forgetting tends to be slower. This is why the timing of review matters as much as the amount: reviewing just as a memory begins to fade is more productive than reviewing something you still know cold, and far more productive than rereading something fresh in short-term memory.

Ebbinghaus also described the savings method, measuring how much less effort was needed to relearn something than to learn it the first time. Savings are a useful reminder that even “forgotten” material leaves a trace. Relearning is faster than starting from zero, which is a good reason not to be discouraged by gaps in memory when you return to old material.

How the site uses it: the retention curve in the analytics section lets you compare a no-review path with a spaced-review path. It is a teaching illustration, not your personal data.

Working memory is the mental workspace where you hold and manipulate information in the moment, and it is severely limited. Early estimates suggested that people can hold around seven items; later research proposed a smaller figure, closer to four chunks, when rehearsal is blocked. Whatever the exact number, the central point stands: the workspace is small and a crowded one makes learning slow and error-prone.

John Sweller’s cognitive load theory divides the demand on that workspace into three kinds. Intrinsic load comes from the inherent complexity of the material and how many elements must be considered at once. Extraneous load comes from the way material is presented, such as confusing diagrams, split attention between a picture and distant text, or cluttered notes. Germane load is the effort devoted to building lasting mental structures, called schemas, which is the productive kind of work.

The study implications are concrete. Break difficult topics into smaller steps and master each before combining them. Place labels next to the parts of a diagram they describe. Work through examples with the solution partly shown before attempting problems from scratch, because worked examples reduce load for beginners. As expertise grows, the same support can become a hindrance, a phenomenon called the expertise reversal effect, so fade the scaffolding as you improve.

Chunking is the learner’s main tool here. Experts do not have bigger working memories; they have richer schemas, so what looks like twelve separate items to a novice is one familiar pattern to them. Retrieval practice, the work of recalling and applying what you have learned, is how those schemas get built and automated, which frees working memory for harder questions.

How the site uses it: short focus blocks keep sessions inside your attention budget, and flashcards break content into single, testable chunks.

The German science journalist Sebastian Leitner popularised a wonderfully low-tech approach to spaced repetition in his 1972 book on learning. You need only a set of cards and a few boxes. Every new card begins in the first box. Each time you review a card and answer correctly, it moves one box further along. Each time you answer incorrectly, it goes back to the first box.

The boxes are reviewed at different frequencies. Box one is visited often, perhaps daily; later boxes are visited progressively less often. Cards that are easy for you travel quickly to the infrequent boxes and almost leave your attention, while cards that keep failing stay in the high-frequency box until they stick. This is spaced repetition with an automatic prioritisation system built in, achieved with nothing more than sorting.

Its strength is simplicity and the discipline it imposes. The rule for promotion and demotion makes the decision for you, which prevents the common mistake of reviewing the cards you enjoy rather than the ones you need. Its weaknesses are real as well: a fixed schedule ignores differences in card difficulty, and a single mistake on a mature card returns it to the very start, which can feel harsh. Modern algorithms adjust intervals continuously per card for this reason.

Pairing the boxes with good cards is vital. Make each card test one idea, phrase the front as a question, and write answers in your own words. Cards that merely ask you to recognise a term teach less than cards that ask you to explain why or how. Interleaving topics within a review session adds a further benefit by forcing you to identify which kind of problem you face.

How the site uses it: the flashcard module follows a Leitner-style philosophy: honest verdicts that decide how soon the card should return.