Don't Change Direction

Every hour you have ever put into getting good at something — all of them, added up, across every field anyone has studied properly — explains about twelve percent of how good people end up.

In the studies that measured practice properly instead of asking people to remember it, five percent.

This piece is about the sailor’s word this show is named after, the one thing it gets right about all that, and the two things it gets badly wrong.

The word

Fetch is the uninterrupted distance the wind blows across open water. That’s NOAA’s definition, and the two qualifiers are the whole thing: uninterrupted, and without a significant change in direction.

Wind crosses water, hands energy to the surface, ripples become waves, waves become bigger waves. How big depends on exactly three things: how hard the wind blows, how long it blows, and how far it blows across. The third is the fetch.

It’s a good name for a show because you never see it. In 1963 Walter Munk put six recording stations on a line from the bottom of the North Island to a beach in Alaska and tracked swell sixteen thousand kilometres, finding it arrived having lost almost nothing. Nobody standing on that beach could have told you. They’d have seen a wave.

That part is true, and worth naming a show after. It’s the sentence underneath that has the problem:

Nothing becomes a wave immediately. A wave isn’t made by force — it’s made by a long, uninterrupted run of small pressure over distance. Longer fetch, bigger wave. The variable is distance, not effort.

Unlike most metaphors, that one is specific enough to be wrong. It makes two checkable predictions about people: distance determines how big you get, and an interruption costs you the distance. People have been tested on both.

Prediction one: distance makes you big

This is the ten-thousand-hours story. It traces back to a 1993 study of forty violinists at a Berlin music academy, and then went round the world in a book.

In 2014, Macnamara, Hambrick and Oswald pooled the whole literature: 88 studies, 111 independent samples, 157 effect sizes, some 11,000 people. Across everything anyone had measured, deliberate practice explained 12% of the variance in performance (95% CI 9–15%).

The split by domain matters more than the headline:

Domain	Variance explained
Games (chess etc.)	26%
Music	21%
Sports	18%
Education	4%
Professions	<1%, not significant

The gradient isn’t random. The more a domain is a closed system with fixed rules and a clean scoreboard, the more the hours buy you. The more it resembles an actual job — messy, social, changing underneath you — the less.

And then the finding almost nobody quotes: how you measure the practice changes the answer. Retrospective interviews gave 20%. Questionnaires, 12%. Contemporaneous logs — the most objective method — 5%. The better the measurement, the smaller the effect. That’s the direction you never want a result to move in.

The authors’ own conclusion is admirably restrained: “deliberate practice is important, but not as important as has been argued.”

What the 12% does not say. It doesn’t say practice is pointless. It says practice doesn’t explain the differences between people who are all already doing it — every participant in that literature was a practising musician, a competitive player, an athlete. Nobody was on the couch. Measuring the spread among people who all showed up is a different question from whether showing up matters.

And the other 88%? The paper doesn’t say. It notes the variance is “potentially explainable by other factors” and lists candidates — starting age, working memory, general ability, personality — without claiming to have shown which. Most people will hear 12% and fill the rest in with “talent.” The study establishes an absence, not an answer.

So: the honest version for someone already doing the thing is keep going, it may well not make you great, and you probably want a reason that isn’t greatness.

The ocean doesn’t support “distance, not effort” either. Waves don’t grow forever: a sea becomes fully developed when the wind’s energy input equals what’s lost in breaking, and past that, more distance does nothing. The height of that ceiling scales with the square of the wind speed. Double the wind, quadruple the wave. Distance is a threshold; the wind is the multiplier. The line got its own metaphor backwards.

Prediction two: an interruption costs you

This is the streak. The chain you don’t break, the little flame in the app, the whole architecture of how people are told to build habits — and it’s what the metaphor most obviously implies.

The paper everyone half-knows is Lally and colleagues, 2010: 96 people, 84 days, each choosing one simple daily behaviour in a consistent context and rating how automatic it felt. It’s where “66 days to form a habit” comes from — a number quoted far more confidently than it deserves, given the range was 18 to 254 days and the curve fitted only 39 participants.

But the number that matters is one almost nobody quotes. They looked at what happened when someone missed a day.

Missing a single opportunity reduced automaticity by less than half a point, and did not meaningfully affect the trajectory.

It’s noise. You cannot see it in the curve. So the metaphor’s second prediction is not what the data shows — and the belief that it is does real damage. People don’t quit because they missed Tuesday. They quit because they believe missing Tuesday means it’s ruined, and if it’s ruined you may as well stop. The metaphor supplies that belief.

The distinction the metaphor actually makes

Except the metaphor doesn’t quite say that, and this is where it earns its keep.

Read the definition again: uninterrupted distance, without a significant change in direction. Two separate conditions, and only one of them is about continuity in time. If the wind drops for an hour and returns on the same bearing, the fetch isn’t reset — you have a slightly smaller sea and the same run-up. What genuinely destroys fetch is the wind swinging thirty degrees, because now it pushes across the waves it just built instead of along them, and the organised thing collapses into chop. That is when you start again from ripples.

So the question was never did you miss a day. It’s are you still pointed the same way.

The trap is that turning feels like effort. Rebuilding your system — new programme, new coach, new framework, new app, each sensible on its own — feels like the most productive week you’ve had in months, and nothing about it feels like quitting. The wind never dropped. What was never there was three hundred kilometres in one direction.

This last part is a bet, not a finding. Nobody has measured it. The missed-day result is Lally’s; the direction claim is a reading of the ocean extended one step past where any data goes. It’s the useful-sounding part, which is exactly why it needs the label. This show’s whole complaint about science coverage is people taking a true mechanism and shaving off the qualifier — so: this bit is measured, this bit is a bet. Both allowed. Label them.

What to do on the day you miss

Nothing. Don’t make it up, don’t double tomorrow, and above all don’t treat it as information about whether you’re the kind of person who does this. It cost less than half a point on a scale you can’t feel.

The dangerous part isn’t the missed day — it’s the response to it. That’s when the turning happens: you miss Tuesday, and by Thursday you’re redesigning the whole thing into a better plan that accounts for Tuesdays. That redesign is the thirty-degree swing.

So: you’re allowed to miss. You’re not allowed to respond to missing by changing direction. Sit in the gap and carry on the same bearing.

---

Sources: Macnamara, Hambrick & Oswald, “Deliberate Practice and Performance…,” Psychological Science 25(8), 2014 · Lally, van Jaarsveld, Potts & Wardle, “How are habits formed,” European Journal of Social Psychology 40(6), 2010 · Ericsson, Krampe & Tesch-Römer, Psychological Review 100(3), 1993 · NOAA JetStream, Ocean Waves · University of Hawaiʻi, Sea States · Pierson & Moskowitz (1964) via WikiWaves · Snodgrass et al., Phil. Trans. R. Soc. A 259, 1966.