Part II of Brian Sommer’s three-part series on Moneyball, golf’s numbers gane, and what statistics truly show
Take the “mental game.” A golfer gets angry. Then hits another poor shot. Observation. Now: “He carried the anger into the next shot.” Explanation. Then: “Anger interferes with performance.” General causal claim. Then: “Golfers need to regulate anger.” Prescription. Then: “Elite players have trained reset routines because emotional regulation is necessary under pressure.” Necessity.
Look how far we have traveled. Now introduce the golfer who gets furious after a poor drive, throws down another ball, and stripes it. What happens to the theory? Often, nothing. We simply explain the counterexample away. “He overrode the anger.” “He subconsciously reset.” “The second ball didn’t count.” “The pressure was different.”
Perhaps. But notice what has happened. Evidence that could challenge the explanation is instead absorbed by it. The theory survives every outcome. That is not an especially impressive theory. It is an insulated one.
Predictors Are Not Causes
Modern performance science introduces another danger. Prediction sounds impressively close to explanation. It is not. Suppose a study finds that golfers with characteristic X tend to perform better. Excellent. X might have predictive value. But several possibilities remain. X might contribute causally to performance. X might be a consequence of expertise. X and performance each might arise from something else. X might simply accompany successful performers. The relationship might depend heavily on context. Training X might not reproduce the outcome at all.
This is why the movement from: “X predicts performance” to “X causes performance” requires additional evidence. And the movement from: “X causes performance” to “athletes need to train X” requires more again.
Golf often behaves as though these are synonyms. They are not. “Predicts.” “Correlates with.” “Contributes to.” “Causes.” “Is necessary for.” Those phrases represent very different claims. The language should become stronger only as the evidence becomes stronger. Instead, coaching often does the reverse. The evidence remains modest. The language becomes heroic.
Miller Barber Walks Into a Modern Lab
Now imagine Miller Barber walking into a modern golf performance center as a teenager. Nobody knows his future. Nobody knows about the professional victories. Nobody knows how extraordinarily effective he will become. We simply see the swing. It is unusual. So, we put markers on him. Measure him. Film him. Compare him with a database. Perhaps we find several deviations from contemporary elite norms. Would we recognize an extraordinarily effective individual solution? Or would we diagnose deficiencies? That question bothers me.
Because the distance between someone’s movement and our preferred model is not necessarily the distance between that person and great golf. Now reverse the experiment. Scottie Scheffler walks into the lab after becoming the best player in the world. We measure him. Suddenly every unusual characteristic becomes fascinating. His footwork is no longer a defect. It is part of his genius. His impact conditions become clues. Hismovements become instructive. One player receives: “You don’t match the model.” The other: “Perhaps everyone should study your model.” But what if both movements are simply solutions created by the golfers who own them? With Barber, we risk treating difference as deficiency. With Scheffler, we risk treating excellence as prescription. Different mistakes. Same assumption: The model has authority over the player.
We Might Be Measuring the Solution
There is a phrase I keep returning to: We might be measuring the solution and mistaking it for the command that produced the solution.Scheffler’s clubface does what it does. His feet do what they do. His pelvis does what it does. His pressure shifts. His body organizes. Those movements are real. But perhaps they are not instructions being executed by some internal foreman. Perhaps much of what we measure is the visible result of how an exceptionally skilled performer has learned to respond to a club, ball, target, environment, intention, and task. That possibility matters. Because if the measured movement is partly the expression of learning, prescribing the measurement back to the learner may reverse the causal direction. We see what expertise produces. Then teach the product as though it produces expertise. That is a very different proposition.
Data Are Great. Reality Is Better.
None of this is an argument against technology. Quite the opposite. Data can reveal things the eye cannot. It can challenge memory. Expose bias. Reveal patterns. Identify distinctions.
Correct bad assumptions. Show us that a player we dismissed is far more effective than we thought. That is exactly what made sabermetrics powerful. But the proper relationship should remain:
Data answer questions about reality. They do not replace reality.
When the numbers contradict my explanation, I should question the explanation. When the numbers reveal something unexpected, I should become interested. When the player’s experience contradicts my model, I should not automatically correct the player. When two elite performers solve the same problem differently, perhaps the variation is information rather than noise. The danger is not measurement. The danger is authority. The scout once said: “I know a baseball player when I see one.” The modern analyst can say: “The model says this is optimal.” Different vocabulary. Same temptation.
Talent Is Not Anti-Science
Some people become uncomfortable with the word “talent’’ because it can sound like surrender. If we call Scheffler “unusually talented,’’ are we simply saying we cannot explain him? Not necessarily. Talent is not an argument against inquiry. Study him more. Measure more. Ask better questions. Investigate perception. Learning. Movement. Development. Practice history. Decision-making. Equipment. Environment. Biology. Experience. But remain willing to say: We still do not completely know. There is dignity in that answer. Ted Williams was Ted Williams. Scottie Scheffler is Scottie Scheffler.
We can understand more and more about what they do without pretending that the remaining mysteryis merely waiting for the next sensor. Perhaps elite performance emerges from relationships among so many changing variables that there is no single explanatory lever worthy of the sentence: “That’s why.” That possibility should not frustrate us. It should make the inquiry better.