The Groove: why good clinicians stop improving

Two clinicians assess the same client. One finished their degree eighteen months ago and still rehearses the session in the car. The other has fifteen years behind them and a name that comes up when people ask around. Everything we assume about experience says the gap between their observed performance should be wide, and researchers have spent decades going looking for it, across medicine, nursing and beyond, with real measures of real performance rather than reputation. The gap keeps not being there.

That is not a claim that the years teach nothing. Performance climbs steeply early, then flattens, and the flattening is so consistent that expertise researchers treat the plateau as the standard trajectory of a profession rather than the exception. The years are doing something. They are just not doing what we assume.

The reason is a mechanism nobody resents once it is explained. When you get good at something, the skill automates, and automation is the entire point of getting good: it frees your head for the complex parts, the family dynamics, the thing the referral did not mention. It is the stage colleagues compliment. You could do that in your sleep, they say, and the sentence is more accurate than it intends to be. Automated performance is smooth, reliable, and generates no learning at all, because the skill has stopped asking anything of you. You did not plateau despite getting good. You plateaued because you did.

Which is how a clinician arrives in the Groove: the stretch of a career where the work runs beautifully and teaches nothing. A groove is comfortable, hard-earned, and exactly the same shape as a rut. The only difference between them is how long you stay in it.

The researcher who mapped the way out was Anders Ericsson, and he did it by taking expertise apart to see what it was made of. He ran memory experiments until ordinary volunteers could recall strings of eighty digits, and he collected the practice diaries of violinists at a Berlin conservatorium to find out what the best ones actually did with their hours. What he found, over and over, was that the people who keep improving are not doing more practice. They are doing a different kind, which he called deliberate: a defined objective, work at the edge of current ability, feedback on the result, refinement, repeat.

Ordinary practice maintains a skill. Deliberate practice improves it. The tenth fluent repetition of a familiar assessment holds your skills exactly where they are, and there is nothing wrong with that, maintenance is most of any working life. But it is not development, because development requires the one thing fluency has engineered out: the possibility of being wrong about something.

So where is feedback supposed to come from, once nobody watches you work? The cheapest source available is prediction. Before your next assessment, or the next intervention with an uncertain outcome, write down what you expect: what you will find, what will change, how much, by when. Then check. Fluent clinicians predict constantly and write none of it down, which means being wrong slides past unexamined, wearing fluency's uniform. Deliberate practice grew up in chess halls and conservatoriums, where feedback arrives fast and clean, and its transfer into slower, noisier clinical territory is argued rather than proven. But prediction against outcome is the closest thing our work offers to a clean signal, and it costs nothing beyond honesty.

None of this needs a course, and it does not need more hours, which is fortunate because nobody has any. It needs a small share of the week to carry work with an edge on it, which is partly a question of whether you have any real say over what fills your week in the first place. And it needs the Groove seen clearly for what it is. On a record, the groove is where all the music lives. Drop the needle anywhere in it and it plays, perfectly, every time. It just never plays anything new.