Main idea

A spaced review system — one tray every three days, one every two weeks, one every two months, a card changing tray when it seems known — organizes the occasions for rereading. It says nothing about what you do during the rereading, and that is where everything is decided.

Checking that you can recite a card teaches one thing only: that it is memorized. The useful operation is different. It consists in putting the wording up against what has happened since the last exposure: has an experience confirmed it, has a discussion displaced it, has a new source qualified it, has a concrete situation weakened it?

Time doesn't test ideas through its duration alone. It tests them because it brings experiences, contradictions and new sources — and it only brings them to the card if the rereading organizes the encounter. A card that holds is strengthened, a card that shifts is refined, a card that no longer holds goes out.

Layer added on 2026-09-20. The encounter doesn't always come from the calendar. It often comes from an outside proposition — a tweet, a reading, a conversation from the week — that you put up against the cards instead of asking them for confirmation. The operation is then explicitly a confrontation: do I agree, where does my disagreement begin, what has changed since the card was written, what nuance is missing. It can strengthen the card, displace it, weaken it or reveal that it no longer holds — and that is the only result worth having, since immediate agreement teaches nothing.

Why it matters

It changes what a successful review is. Success isn't knowing the card, it's having something to set against it — and a review that never produces an adjustment signals a ritual rather than a test.

It also explains why the tray mechanism is secondary: it sets the frequency of the encounters, not their content.

Nuances and limits

Not every period brings material. Between two closely spaced exposures, sometimes nothing has happened that concerns the card, and forcing the confrontation would manufacture an artificial verdict.

And rereading judges with the eyes of the moment: a conviction can be weakened by a bad week as much as by a fact.

Open questions

• Should you keep a trace of a card's successive displacements, or is the latest wording enough?