You Copied a Whole Error Log — So Why Are You Still Stuck?

A lot of error logs look like this: copy out the question, copy out the correct solution, close the book. Half an hour spent, hand tired, and the next question of the same type still goes wrong. Because copying is passive — your hand moves, your judgement does not. A useful record of a mistake is only a few lines, but every line has to be something you worked out: which step did I get wrong? Why did it go wrong that way? What will remind me next time?

For each wrong question, classify it first — the category itself tells you what to fix: • Concept error: the method itself was wrong, or you misremembered a theorem's conditions. Go back and re-read that topic. • Calculation error: right method, wrong arithmetic. Drill speed and a checking habit, not the theory. • Reading error: missed a condition, misread the requirement (asked for area, computed perimeter). Practise circling keywords as you read. • Time error: you can actually do it, you just ran out of time. Practise the solving speed for that question type. Tally these over a few weeks and you will find your mistakes cluster heavily in one or two categories. Those are where your time should go; you can afford to do fewer questions in the others.

You do not need a long write-up. Three lines per wrong question: 1. Which step did I first go wrong on? (Not "I couldn't do this one" — pin it to a specific step.) 2. What was I thinking at that step, and why was that thinking wrong? 3. Next time I meet this type, what do I remind myself first? (Write it as one actionable sentence, e.g. "composite function — confirm inner and outer layer before differentiating.") Line 3 is the point — it is the thing you can actually use next time. Do enough and these reminder sentences build into a checklist that is yours.

Writing it down is only step one — without review you forget it anyway. A reference schedule: log it the day you get it wrong, redo it after 3 days, again after a week, again after two weeks. Each time, cover the solution and redo it yourself — not a read-through of the worked answer. Three correct in a row and the question counts as genuinely mastered; cross it off the review list. Running that schedule by hand is fiddly and easy to let slip. DuckMath's mistake notebook automates it: wrong questions come back to test you on a forgetting-curve schedule and "graduate" automatically after three correct in a row. You just do the questions.

Now go practise

Understanding the theory is not enough — practice is what makes it stick. DuckMath has a topic-organised, human-reviewed question bank waiting.