Insight

Stop Practising the Part You Already Know

Oli·Saturday, October 17, 2026 Edition

WHEN SMOOTH MEANS STUCK

Fatou Mensah is at her kitchen table at half past nine, working through a stack of calculus problems. She has been at it for forty minutes. The problems have started to feel almost obvious. She works faster, checks her answers, gets them right. Something in her chest loosens. She feels, with some certainty, that she is getting it.

She isn't, quite. What she is getting is fluency with a format. When you drill the same problem type in sequence, your brain stops retrieving the method and starts recognising the shape. The mechanism that actually builds skill — deciding which approach fits this particular problem — gets bypassed. The ease you feel after an hour of blocked practice is real. It just isn't evidence of learning. Interleaving works because it forces retrieval every single time: you never know which tool you'll need, so you have to reach for it rather than coast toward it.

The underlying research here goes back to work by Robert Bjork on what he called desirable difficulties. The counterintuitive finding that making practice harder in specific ways improves long-term retention. The key result is simple and uncomfortable: blocked practice produces better performance during the session; interleaved practice produces better performance later, when it counts. Students who trained on mixed problem sets scored significantly higher on delayed tests, even though they reported feeling less confident during the training itself. That gap between felt fluency and actual retention is the trap.

HALFWAY THROUGH THE WORK

So: you are three problems into tonight's session, and you just got one wrong. You had it right yesterday. That moment. Mild irritation, the urge to back up and repeat the easier version until it clicks again. Is exactly where the principle bites. The wrong answer is doing something for you that the right answer wasn't. It is forcing your brain to retrieve, not recognise. Fatou, somewhere around problem six, probably started picking up speed again and mistook that for progress. You can let the difficulty stay. One problem from each type, out of order, even when it slows everything down.

Learning that feels like learning usually isn't. The sessions worth having are the ones that leave you slightly unsure whether you're improving at all.

🎯

Try This

Pull three different problem types from whatever you're currently studying and work them in rotation rather than finishing one set before starting the next. When you get one wrong, note whether you mixed up the method — that confusion is the session working.

How?

Key Facts

*Fluency is not the same as learning: the ease you feel after blocked practice measures recognition, not your ability to use the skill in a new context.

*Interleaving problem types during a single session forces your brain to retrieve the right method each time, which is the actual work that builds transferable skill.

*When a session feels frustratingly slow, treat that friction as signal rather than failure — difficulty during practice predicts retention far better than smoothness does.