"I keep practising the wrong version," said Fumiko Reyes, half to herself, in the car park outside her weekly conversation class. She had been studying Spanish for two years. She was brilliant in the classroom. At the restaurant, ordering, she froze completely.
Skills do not travel. That is the short and irritating truth. What you build in practice is a capability tied to the conditions of practice — the low stakes, the forgiving pause, the absence of a real face waiting. Move those conditions and you move the floor. Fluency in the approximation is its own trap, because it feels exactly like progress. The reps are going in. The confidence is building. But confidence in the simulation is not the same thing as capability in the room.
George Miller's foundational work on memory and chunking showed this pattern at the cognitive level: skills encode in context. But the clearest practical demonstration came from research into surgical training, where simulator proficiency repeatedly failed to predict real operating-room performance until training was restructured around actual procedures under actual time pressure. The gap was not knowledge. It was the conditions under which knowledge had to fire.
You are three paragraphs into the draft, and the cursor has slowed. You have written this kind of thing in your notes, in outline, in your head at midnight. It felt ready. Now the blank document has an audience behind it — real publication, real names attached. And the words are not coming the way they did in private. That is not a confidence problem. The practice you did was real writing; this is a different act entirely. The only way through is to have done more of this, in exactly this form, before. Fumiko found that out at the restaurant. You are finding it now.
Practising hard and practising the right thing are two separate commitments, and most of us have only made one of them. The discomfort of the real attempt is not a sign the skill is broken. It is the skill, finally, being asked to exist.