Afisena
There is a student who genuinely understood it in August and does not in September. Not lazy, not pretending. They really did know it at the time.
Every teacher has seen this. What is rare is a way of finding out before the exam tells you.
A mastery score knows nothing about time
When a system records that a student has mastered a concept, that number moves only when the student answers something. There is no time term in it anywhere.
Which means a concept mastered in the first week of term reads exactly the same in December. The learning path was marked complete long ago, and nothing ever asks about it again.
A system like that can tell you what a student has learned, and has no idea what they are losing.
So retention is worked out separately
Two things are kept apart on purpose, because they are genuinely different: how well something was known, and how long it stays.
The first comes from how the student answered. The second comes from how many times they have come back to it.
Something met once and answered correctly loses about half its hold in three days. Each further practice stretches the interval — roughly doubling it — until at some point the concept is effectively kept and there is nothing to gain by asking again. The ceiling is six months.
Worth saying plainly: this is an estimate, not a measurement. Nobody can read the inside of a child's head. What is being counted is how well they once answered, how often, and how long ago — and a long-established curve turns those three into one guess you can check.
What it changes is the order
If a revision list is built from what has gone untouched longest, the order is wrong.
A concept practised five times three weeks ago is in far better shape than one answered once last Tuesday. What should come first is the most faded, not the oldest.
Two things it deliberately will not do:
It will not ask about something never learned. If the mastery was never really there, that is not forgetting — it is unfinished. Unfinished belongs to teaching, not to revision. Asking a student to "remember" something they never got is a way of making them feel stupid about somebody else's mistake.
It will not ask twice in a day. There is a minimum gap of twelve hours. A student who revises this morning and gets asked again this afternoon learns exactly one thing: that the prompt is not worth reading.
And the summary figure on a student's own screen counts only what they had learned. Count the unfinished too and a diligent student's number drops every time they begin something new, which is a penalty for progress.
Two weeks before mid-term exams
The advice most students get is to go over everything again. For most of them that means re-reading the chapter they already know, because that is the comfortable one to read.
The spacing effect is not new: short revisions spread apart hold far better than one long evening. What has always been hard is not accepting that, it is answering the question that follows it — revise which parts.
That list now exists, ordered by what has faded most, and it is usually much shorter than the student expects.