Why the app makes you wait.

What you actually want.

To still have the material months from now, without study becoming your main occupation. Both halves matter. Retention with unlimited time is not a hard problem; retention on the forty minutes a week you actually have is the whole problem.

What you run into.

The obvious approach is to review often, and to review what you have just learned while it is still fresh. Keep it warm, do not let it slip. It is the intuition behind cramming, and behind most revision timetables.

It fails twice, for two unrelated reasons.

Put together: the intuitive schedule costs the most and returns the least. And like rereading, it feels productive throughout, because reviewing something you already know produces a long run of correct answers.

What we do instead.

Ignia models forgetting for each learning point separately, and schedules a review for the point where a review is currently worth the most — roughly, where you are still able to retrieve it but no longer able to do so effortlessly. The scheduler is FSRS, an open spaced-repetition algorithm in the same family as the one behind Anki, and it works on the individual point rather than on the lesson or the course.

The quantity the app tracks is retention: the estimated probability that you could recall a given point right now. It falls on its own between sessions and rises when you successfully review. The queue is ordered by it (forecast_retention, mastery/mod.rs), so what comes back is decided by decay, not by where you are in the course or which topic you last opened.

Consequences you will notice

• Intervals grow. Each successful review pushes the next one further out. A point you have held several times will not be seen for weeks, and that is the system working rather than losing track of it.
• Reviews are interleaved with new material. A session mixes what is due with what is next, rather than finishing one and then starting the other.
• A miss is not a reset. Missing a point shortens its interval and moves it forward in the queue. It does not erase its history or send it back to the start.
• Reviews outlive the course. A finished course still has a review queue. Finishing the material and keeping it are different events, and the app does not treat the first as the second.

What this does not buy.

This is the part most easily left out, so it gets its own section.

• It is an estimate, not a reading of your memory. Retention is a model's output, computed from your answer history. It is not a measurement of your brain, and it should be read as a good guess about a population of points rather than a fact about any single one.
• It is weakest exactly when you meet it first. The model has little to go on for a point you have answered once or twice, so early estimates are the least reliable ones — and they are also the ones a new user sees. It improves with history. We would rather say that than present a confident number from day one.
• It cannot see anything you did elsewhere. If you used the material at work, read about it, or talked it through with someone, none of that reaches the schedule. The app knows what you answered in the app.
• It makes sessions feel harder, not easier. Being asked for something at the edge of recall is uncomfortable, and it is meant to be — the discomfort is the same thing as the benefit. If you want the pleasant version, review everything the day after you learn it; it will feel much better and work considerably less well.
• One session a day is a convention. It is a planning decision that keeps the calendar honest about the workload, not a demonstrated optimum. There is no evidence we can point to that says one a day is the right number for you, and nothing in the app stops you doing more.

Questions.

Sources.

What each one is doing here, so you can tell which claim rests on which work. Every one of these is a published paper you can look up.

• Cepeda, N. J., Pashler, H., Vul, E., Wixted, J. T., & Rohrer, D. (2006). Distributed practice in verbal recall tasks: A review and quantitative synthesis. Psychological Bulletin, 132(3), 354–380.

  The spacing effect, synthesised over several hundred studies. The claim that the same study time produces more retention when spread out comes from here.

• Cepeda, N. J., Vul, E., Rohrer, D., Wixted, J. T., & Pashler, H. (2008). Spacing effects in learning: A temporal ridgeline of optimal retention. Psychological Science, 19(11), 1095–1102.

  The optimal gap is not fixed — it scales with how long you want to keep the material. This is why the app schedules per point rather than on a fixed weekly rhythm.

• Free Spaced Repetition Scheduler (FSRS) — open-source scheduler built on the difficulty/stability/retrievability memory model, a descendant of SM-2 (SuperMemo, Woźniak).

  The algorithm the app actually runs. Note what kind of claim this is: it is validated by engineering benchmarks on large review datasets, not by a psychology meta-analysis, which is why this page calls its output an estimate.

You will find effect sizes in these papers. We deliberately do not print them on this page: an average across studies does not predict what any one person will get, and a number quoted without its conditions reads as more precise than it is. In the papers they come with those conditions.