The spaced repetition algorithm

EON does not show you cards at random or in a fixed order. It decides when each one comes back based on how long it has been since you last reviewed it and how you rated it. The goal is to have you review right before you forget: not too soon (that would be wasted effort) and not too late (you would have already forgotten).

Desired retention

Desired retention (%) is the percentage of cards you want to recall correctly, on average, when they appear. If you raise it, the algorithm schedules reviews more often to make sure you do not forget. If you lower it, reviews are spaced further apart so you study less, but you will forget slightly more. A value around 90% is a common balance: you remember most of your material well without overloading your sessions.

When to change it: if you consistently feel well-prepared and want to study less, you can lower it slightly. If you are studying for an exam where every detail matters, raise it.

Maximum review interval

Cards you know very well could theoretically be scheduled months or even years out. Maximum review interval (days) sets a ceiling: no card will wait longer than that many days between reviews, even if the algorithm would space it further.

When to change it: if you use EON for languages or long-term knowledge you want to keep forever, you can leave it high. If you are studying for an exam in the coming months, lowering it ensures cards come back before the date arrives.

Interval fuzz

When many cards were added at the same time, they tend to come due on the same day, creating workload spikes. Enable interval fuzz adds a small random offset to each due date to break up those clusters. The result is a more even workload spread across the week.

When to change it: it is usually best to leave it on. Turn it off only if you need precise control over the exact day each card appears.

Short-term learning steps

New cards and cards you forgot first go through a state called Learning before entering the long review cycle. While in that state, they come back several times the same day or over the next few days until they are consolidated. Enable short-term learning steps controls whether this intermediate state is used.

When to change it: for most decks, keeping it on is the right call. If your cards are very simple or the deck is small, you can turn it off.

Problem cards (leeches)

If you forget the same card many times, something is wrong: maybe the wording is confusing, or the concept is too complex for a single card. Leech threshold (lapses) defines how many consecutive lapses a card must accumulate before EON flags it as a problem.

When a card reaches that threshold, EON can do one of two things depending on the Leech action:

• Tag only — the card keeps appearing in sessions, but it is marked so you can review and improve it.
• Tag and suspend — the card is removed from the study queue until you manually unsuspend it.

When to change it: if many cards are suspended and you feel you are losing them, raise the threshold. If you want to catch problems earlier, lower it.

Study order

When several cards are due, EON decides the order in which to show them. There are three modes:

• Priority (most overdue first) — cards that have been waiting the longest appear first. This is the most efficient order when you have a backlog.
• Interleaved (review + new) — mixes review cards with new cards throughout the session.
• Random — fully shuffled order.

When to change it: if you have many new cards piling up, Interleaved helps you introduce them without stopping the review of older ones. If you just need to catch up, Priority works well.

Changing settings does not erase your review history. The algorithm adapts upcoming dates using the existing history together with the new parameters.

Frequently asked questions

What happens if I raise desired retention to the maximum? Cards come back much more often. Your daily workload rises noticeably, but so does the percentage you get right when they appear.

Can I use different settings for each Deck Pack? Yes. Review Preferences can be set account-wide or overridden per individual Deck Pack.

Do algorithm settings affect cards I already know well? Yes, but the change happens gradually: the algorithm recalculates upcoming intervals the next time each card comes up for review.

Next step

• The optimizer — automatically tunes algorithm parameters to your real review history.
• Review sessions — how to rate cards well so the algorithm works correctly.
• Calendar and planning — how to see and manage your upcoming workload.