A failed midterm hands you the most valuable study asset you'll get all term: a marked exam, with every miss identified and every point lost traceable. The instinct is to waste it — reread the chapters as punishment, drop the course mentally for two weeks, or redo the same midterm until the score looks better. All three ignore what the exam actually gave you: a diagnostic.

The repair starts with a taxonomy. Sort every miss into a bucket: cue (didn't know what the question was asking), method (knew the topic, picked the wrong approach), arithmetic (right method, wrong computation), wording (knew it, wrote it badly). The largest bucket is the repair target — if 6 of 10 misses were wrong-method-on-lookalikes, that's a discrimination problem, not a coverage problem, and rereading the method chapters is the wrong fix.

Why the shame reread fails

Rereading the chapter after a failed exam is penance, not repair — it feels like doing something, and it produces nothing. The reread covers the material again; the exam showed the material wasn't the problem — the retrieval or the application was. A chapter tour re-exposes you to content you mostly knew; it doesn't fix the specific operation that failed.

The tag-and-variant approach matches the repair to the failure. Feedback works when it's specific — "I got it wrong" is vague; "I used the integration method on a differentiation problem" is a tag. And the repair is another retrieval — a new variant that forces the failed move again, not a reread that lets you nod along. Specific feedback plus another retrieval is the fix; a shame reread is fluency dressed as penance.

The taxonomy workflow

Tag every miss by type. Cue (didn't parse the question), method (wrong approach), arithmetic (right approach, wrong computation), wording (right answer, poorly expressed). The tag tells you what broke; the count tells you what's worth fixing. Six method misses out of ten is a different exam problem than six arithmetic misses — and the repairs don't overlap.

Find the largest bucket. Not the most painful miss — the most frequent one. One dramatic error is a story; six similar errors is a pattern. The largest bucket is where the points are: fix the method-selection misses and the exam score moves; fix a one-off wording issue and it doesn't.

Write variants of the missed moves. New problems that force the same operation — the method selection, the computation, the interpretation — with new surface details. The variant is the re-test: can you do the move when the story changes? If yes, the miss was a one-off; if no, the tag was right and the repair continues.

Repair with retrieval, not rereading. The variant problems are the fix — attempts that force the failed move, not exposure to the material behind it. Rereading is what the exam just proved doesn't work; the variant is the retrieval that actually repairs.

Office hours with tagged misses. Bring two tagged stems — "I picked method A on these lookalikes, what's the cue that separates them" — not "I don't get it." The tag is the question; the instructor can answer a specific confusion, not a vague one.

A 40-minute session on the clock

The weak plan is the shame reread — cover everything again, redo the same exam until the score improves, which trains the specific answers, not the failed operation. The strong plan matches the tag: 6 method-selection misses means the repair is discrimination among lookalikes — mixed problems, method first, then compute. The chapter tour treats it as a coverage problem; the tag says it was a discrimination problem, and the repair follows the tag.

Why the failed exam is the asset

A marked midterm is a diagnostic you already paid for — every miss identified, every point traceable, the exact operations that failed now visible. No practice test is this good: the practice test is simulated; this was the real conditions, the real pressure, the real gaps. The failed exam is the most honest assessment you'll get.

The waste is ignoring the diagnostic — treating the exam as a verdict instead of a map. "I failed" is a verdict; "6 of 10 misses were method-selection on lookalikes" is a map to the repair. The verdict says you're behind; the map says exactly where — and a student with the map has an advantage over a student who passed and never learned where their gaps were.