Scheduled maintenance: planning reviews across weeks

Three weeks after a lecture, most of it is gone. That is not a sign that you studied badly, it is what memory does with material nobody has asked for. The useful response is not to work harder on the first day, but to decide in advance when you come back to it.

Forgetting is the normal state

Something you understood clearly on Monday will be thin by Friday and largely missing in a month unless something interrupts the process. This happens to everyone, and to material that was genuinely understood. It is not evidence of a weak memory or of insufficient effort.

Because it is predictable, it can be planned around. What you cover today has a rough shelf life, and a review schedule exists to touch it again shortly before that runs out, not long after. Most students do the opposite. They meet the material once and then again in the week before the exam, by which point the second meeting is not a review but relearning from close to zero.

Scheduled maintenance

Aircraft components are not serviced when someone notices a problem. They are serviced at fixed intervals set in advance, after so many hours or cycles, whether or not anything looks wrong. The point is that the check happens before the failure, and nobody has to make a judgment call in the moment.

A review schedule works the same way. You do not revisit a chapter because you feel shaky about it, since that feeling arrives late and is unreliable. You revisit it because the date says so.

Growing intervals as the default

A workable starting pattern: come back about a day after you first covered the material, then three days later, then after a week, then two weeks, then a month. Each review that goes well buys a longer gap, because the material is more stable than it was.

The exact numbers matter less than the shape. If a chapter keeps collapsing at the one week mark, shorten its intervals. If it comes back effortlessly every time, stretch them. Adjusting per topic is normal, and a schedule that is never adjusted is usually one nobody is really running.

What does matter is that the gaps grow instead of staying flat. Reviewing everything every week is expensive and increasingly wasteful, because you keep paying full price for material that has already settled.

A review is a test, not a rereading

This is where the method usually fails in practice. The date arrives, you open the chapter, you read it through, you feel a pleasant sense of recognition, and you tick it off. Nothing has been maintained. You have confirmed that the text is still legible.

A review means closing the source and producing the content: answering your questions on the chapter, writing the main argument from memory, reworking one problem, saying a derivation out loud. Ten minutes of that is worth more than thirty of rereading, and it tells you something rereading cannot: whether the interval was right.

Write the outcome down in one word. Solid, shaky, gone. That word sets the next date.

Running the plan without an app

You do not need software for this. A dated list per chapter is enough: the chapter name, and under it the five dates you intend to come back, written the day you first cover it. When a review is done, tick that date and confirm the next. A single page in a notebook holds a semester of this for several subjects.

What an app genuinely takes off your hands is the arithmetic and the sorting. It sets the next date from how the last review went, and it shows you today’s items without you scanning a list. That is real convenience, especially for large card decks where the number of items makes manual scheduling impractical.

What no app does is the work. It cannot make a review a test rather than a read, and it cannot decide that one chapter matters more than another. People adopt a tool and conclude that the system is running, when what is running is the notification.

What it does to exam preparation

The visible effect shows up in the last two weeks before the exam. If you have been reviewing across the term, the material arrives at that point mostly intact, and the final phase can be what it is supposed to be: past papers, weak spots, timing. If you have not, those two weeks have to carry the relearning and the practice at once, and the practice is what gets dropped.

The total hours are not obviously smaller. They are distributed differently, and distributed time is easier to find. Twenty minutes on a Tuesday in week four is available. Six hours on the Saturday before an exam is available in theory and rarely in practice.

The backlog after a break

Miss a week and you come back to a list where everything is marked overdue. This is where most people abandon the system, because the backlog reads like a debt and the schedule looks like it is lying to you.

Do not try to clear it in one sitting. Sort the overdue items by importance rather than by date, take the five or six that matter most, and do those properly, as tests. For everything else, reset the dates as though you were starting today. The schedule is an instrument for planning, not a record of obligations, and it is allowed to be rewritten. A backlog cleared badly, by skimming forty items in an hour, is worse than one reset honestly.

What it costs, and what it cannot do

For one subject across a term this settles at roughly twenty to forty minutes a week, most of it in short pieces. It grows during weeks with heavy new content and shrinks later, because older material moves to monthly intervals and costs almost nothing.

The honest limit is that spaced review preserves what was understood once. It does not create understanding. Schedule a chapter you never really followed and the reviews will faithfully preserve your confusion at growing intervals. Anything that has not yet made sense belongs back with reading, worked examples and explanation, not on a maintenance list. Another article here covers the review you do the same day, right after a session; this one is about the weeks that follow.

If you run study sessions on a timer that logs each one as a flight, put the day’s due reviews at the start, before the new material. They are short, they are already decided, and the log will show that the maintenance actually happened.