What blocked practice actually teaches
When every problem on the page is the same type, you only have to solve it once. After the first, you already know which method to reach for — the page told you. So you spend nineteen repetitions practising the *execution* of a method and zero repetitions practising the thing the exam actually asks, which is choosing the method.
This is why students walk out of exams saying "I knew how to do all of it, I just didn't know what it wanted." They are describing precisely this gap. They drilled execution. The exam tested selection.
What interleaving does instead
Mix the problem types and every single question forces a decision before it forces a calculation. Is this a substitution problem or a parts problem? That discrimination — noticing which features of a problem signal which method — is a skill, it is separately learnable, and blocked practice never trains it at all.
The catch, and the reason nobody does this voluntarily: it feels awful. Your accuracy during practice drops. You make more mistakes. Every instinct says you are studying badly. You are studying well; the difficulty is the mechanism, exactly as in active recall.
Where it helps most
- Maths and physics — anywhere a problem could be attacked several ways and picking wrong costs you the question.
- Anything with confusable categories — drug classes, cell types, case law doctrines, grammatical constructions.
- Diagnosis-shaped subjects, where the whole task is 'given these features, which thing is it'.
Where it helps least: the first encounter. You cannot interleave methods you have not learned yet. Block while you are learning a method for the first time, then interleave the moment you can execute it at all.