Start hereTiger Woods first picked up a golf club at seven months old. Yet adult world-class athletes, on average, started their main sport later than national-class ones. So when does a head start pay off?
About the book
Range: Why Generalists Triumph in a Specialized World is by David Epstein, a former Sports Illustrated writer and the author of The Sports Gene. Riverhead Books published it on 28 May 2019. It is popular science for general readers, built from stories of athletes, scientists, artists and inventors and the studies behind them.
Its argument, in the publisher’s words, is that “in most fields—especially those that are complex and unpredictable—generalists, not specialists, are primed to excel.”
Head starts work best in “kind” worlds
The book opens with two famous head starts. One is Tiger Woods, who first picked up a golf club at seven months old. The other is the Polgár sisters. Their father set out to make his children into geniuses and chose chess. He took his eldest, aged four, to a Budapest chess club, where she beat the grown man she played.
None became overall world champion, but, as the book puts it, all were outstanding. Epstein’s question is whether these stories generalise.
The book borrows the terms “kind” and “wicked” learning environments from the psychologist Robin Hogarth. Chess and golf, he argues, are kind learning environments: patterns repeat, and feedback is accurate and usually rapid. Much of life, he argues, is wicked: the rules are often unclear or incomplete, and feedback is often delayed, inaccurate, or both.
Does it hold up? The idea of kind and wicked settings does. In a 2015 paper Hogarth and colleagues describe kind environments, where what you learn from closely matches the situations you later face, as “a necessary condition for accurate inferences”.
Separately, in 2009 Daniel Kahneman and Gary Klein concluded that judging an intuition means asking two things. Is the environment predictable, and has the person had the chance to learn its regularities? They add that “subjective experience is not a reliable indicator of judgment accuracy”. How much of life is wicked is Epstein’s argument, not something these papers measure.
That a head start pays off most in narrow, rule-bound activities is consistent with the practice research, but not directly tested. A 2014 meta-analysis, as corrected by its authors in 2018, measured how much of the differences in performance deliberate practice explained. It was 24% in games, 23% in music and 20% in sports, but 5% in education and 1% in professions.
It measures the amount of practice, though, not a head start. Ericsson, whose violin study began the debate, replied that the meta-analysis defined deliberate practice much more broadly than he had.
Try it: before trusting a gut feeling, ask whether the setting is predictable and whether you have had the chance to learn its patterns.
The Roger path is more common than the Tiger path
Against Woods he sets Roger Federer, who tried many sports, from swimming and skiing to skateboarding, before settling on tennis. Epstein argues that research on top performers shows a sampling period of broad, varied activity followed by later specialisation. He argues this pattern is the usual one.
The challenge, the book says, is how to keep “the benefits of breadth, diverse experience, interdisciplinary thinking, and delayed concentration” in a world that rewards hyperspecialisation (p. 13).
Does it hold up? Partly: for adult world-class athletes, not for young ones. A 2022 meta-analysis of 51 study reports covering 6,096 athletes found that adult world-class athletes had done more multisport practice as children and adolescents. They had also started their main sport later and progressed more slowly at first than national-class athletes.
Among youth athletes it found the opposite: the better performers started earlier and did more main-sport practice. A 2025 review in Science, by the same group, reports the same split in science, music and chess.
A 2026 follow-up of 627 top-ranked US high-school basketball players shares the same lead author. In that study the 40 who reached the NBA had more often played other organised sports until age 14 (90.0% vs 52.5% of matched peers who did not). Two of its authors work for the NBA.
These are associations, mostly from athletes’ own recall of their childhood. They compare group averages; they do not count how many champions specialised early, which is what “the exception, not the rule” would need.
Learning that feels slow tends to last
A chapter on learning argues that the methods that work well in the long run often do not feel like progress at the time. They include spacing study out, struggling to produce an answer before being shown it, and mixing up types of problem. Researchers call these “desirable difficulties”, which, as the book puts it, “intentionally sacrifice current performance for future benefit” (p. 85).
Epstein’s example is a study at the US Air Force Academy. Calculus professors whose students shone in their own course turned out to leave them worse off in later courses.
Does it hold up? Mostly, yes. A 2013 review of ten study techniques gave practice testing and spaced practice its highest rating, because they benefit learners of different ages and abilities. It rated highlighting and rereading of low utility.
Robert and Elizabeth Bjork add a limit: a difficulty only helps if the learner has the background to respond to it successfully. A 2025 meta-analysis of maths learning found a small-to-medium benefit for spacing (g = 0.28), possibly smaller than in other subjects. Across only seven studies, it found no reliable benefit yet for testing over restudying.
Mixing problem types works for some material, not all. In a randomised trial in 54 seventh-grade maths classes (787 students), classes given mostly interleaved practice scored 61% against 38% on an unannounced test a month after a review.
A 2019 meta-analysis of 59 studies found a moderate overall benefit (g = 0.42). It was larger for paintings and other visual material, and smaller for maths tasks (g = 0.34). It was unclear for expository texts. For word lists it reversed: practising one category at a time did better (g = −0.39).
At the US Air Force Academy, students are randomly assigned to professors. Students of professors who as a group did well in the first maths course did significantly worse in follow-on maths, science and engineering courses. Students’ evaluations predicted results in the current course but were poor predictors of later ones.
Try it: space study sessions out and test yourself instead of rereading; mix problem types where the material allows it.
Knowledge that travels: far transfer
The learning chapter ends on what researchers call far transfer. In the book’s words, it is when a knowledge structure is “so flexible that it can be applied effectively even in new domains or extremely novel situations”. Epstein links it to the interleaving studies: the students had to spot deep similarities between problems and match a strategy to problems they had never seen.
Does it hold up? It is still argued, and the evidence leans against broad far transfer. In 2017 Giovanni Sala and Fernand Gobet reviewed studies of chess, music and working-memory training. The better the experimental design, the smaller the effect. They “conclude that far transfer of learning rarely occurs”.
The main evidence on the other side is about action video games. A 2018 meta-analysis of intervention studies found an average effect of a third of a standard deviation (g = 0.34). Its authors also estimated that published effects were 30% larger than in the full literature.
Music is argued over too. In 2022 Emmanuel Bigand and Barbara Tillmann reanalysed the data from Sala and Gobet’s 2020 meta-analysis of music training and found a small but significant effect (g = .234).
The weight of the evidence we read leans towards far transfer being rare, with action games and music training the main exceptions claimed.
Try things, then commit: match quality
Economists use match quality for how well a person’s work fits their abilities and interests. Epstein draws on a study comparing English students, who once had to choose a narrow subject before university, with Scottish students, who studied several subjects first.
In his summary, the English and Welsh graduates “were consistently more likely to leap entirely out of their career fields than their later-specializing Scottish peers.” His conclusion is that time spent sampling is not wasted: it is how people find out what suits them.
Does it hold up? Partly: the evidence that later specialisers find better-fitting careers comes mainly from one British natural experiment. In that study, Ofer Malamud found that those who specialised early were more likely to switch to an occupation unrelated to their degree. He read this as evidence that the gains from a better match are large enough to outweigh the extra skills early specialists give up when they switch.
His data cover British graduates from the 1970s to the early 1990s. In US data the raw pattern runs the other way: students who chose their major later were more likely to change fields. The authors of that US study say this has to be adjusted for who chooses late.
Their model found that delaying specialisation tells students something useful, though the signal is noisy. It estimated that forcing everyone to specialise at entry would lower expected earnings by 1.5%.
Try it: treat trying things out as part of choosing, not as time lost.
Knowledge from outside the field
Several chapters praise people who bring ideas across from other fields. One example is Gunpei Yokoi, a Nintendo engineer who was a keen hobbyist rather than a star student. His philosophy was “lateral thinking with withered technology”: finding new uses for simple, older technology, as in the Game Boy.
Breadth helps judgement
The later chapters turn to experts. Drawing on Philip Tetlock’s forecasting studies, Epstein contrasts narrow hedgehogs, who view every problem through one theory, with foxes, who borrow from many. He also argues that “highly credentialed experts can become so narrow-minded that they actually get worse with experience, even while becoming more confident”.
Does it hold up? The forecasting finding does. Tetlock collected forecasts from experts in different fields. He sorted them, after the philosopher Isaiah Berlin, into foxes, who know many things and draw on many traditions, and hedgehogs, who know one big thing.
As his publisher summarises the result, the fox “is more successful in predicting the future than the hedgehog”. A 2010 review of his work states the same finding.
What holds up, and what the book overstates
What the sources show for each idea
The parts of the book that rest on learning research hold up best: spaced practice and self-testing are well supported. So is the idea that experience only builds reliable instincts in predictable settings.
The sport evidence fits the book for adult champions but runs the other way for young athletes, and it shows association, not cause. The career evidence is promising but thin, and far transfer is still argued.
Who it’s for, and who can skip it
It suits readers who have changed direction, or are wondering whether to, and parents and coaches weighing early specialisation for a child. It is also a readable tour of research on learning and judgement.
Some of its learning evidence, such as the work on spacing and testing, is about how anyone learns, not about generalists in particular.
Quick answers
What is Range by David Epstein about?
It argues that in most fields, especially complex and unpredictable ones, people who sample widely and specialise later do well, and that an early head start pays off most in narrow, rule-bound activities such as chess and golf.
What are kind and wicked learning environments?
Terms the book takes from the psychologist Robin Hogarth. In kind environments, patterns repeat and feedback is accurate and usually rapid; in wicked ones the rules are often unclear or incomplete, and feedback is often delayed or inaccurate.
Is early specialisation bad?
The evidence is mixed. Adult world-class athletes had, on average, done more multisport practice as children and started their main sport later, but among youth athletes the better performers started earlier. These are associations, not cause.
Does learning one skill make you better at others?
It is still argued. Sala and Gobet concluded that far transfer of learning rarely occurs; action video games and music training are the main exceptions claimed.
Use this
Space and mix your practice, and treat trying things as part of choosing.
Is this a kind setting with quick, accurate feedback, or a wicked one?
Quick quiz
Nothing is scored or saved. Recalling it once now is what makes it last.
1. According to Kahneman and Klein, what does a trustworthy intuition need?
Answer: A predictable setting and a chance to learn its patterns.Read it again ↑
2. What did the 2022 sport meta-analysis find about youth athletes?
Answer: The better ones started their main sport earlier.Read it again ↑
3. At the US Air Force Academy, how did students of professors whose classes did well in the first maths course fare later?
Answer: Significantly worse in follow-on courses.Read it again ↑
How we checked this · Last checked October 2026
- What we read of the book: the publisher's description and its online excerpt of the opening of chapter 1, the author's book page, a 2019 interview with the author in Outside, and six reviews that quote the book, some with page numbers.
- For the research: the abstracts of the studies listed, the methods section of the 2022 sport meta-analysis, the full texts of the 2020 practice review and the 2021 Ericsson reply, the Bjorks' chapter, the NBER and St Andrews working papers, and the US Institute of Education Sciences page on the interleaving trial.
- We did not read the full 2020 interleaving paper, or Tetlock's book, which we know through its publisher's summary and a review.
- Far transfer is a genuine disagreement: the page gives both sides and says which way the evidence we read leans.
Sources
- D. Epstein, Range: Why Generalists Triumph in a Specialized World (Riverhead Books, 2019): the publisher’s product page and online excerpt of chapter 1; the author’s book page.Primary
- Reviews quoting the book: K. W. Lin, “Range: Why Generalists Triumph in a Specialized World”, Family Medicine 52 (2020); O. Smilansky, “Review: Range”, Columbia Magazine (Fall 2019); J. Kelly, “What I’m reading: Range”, Notre Dame Magazine (June 2019); “The problem with specialization”, Christian Century; C. Chin, “Book summary: Range”, Commoncog; M. Demirbas, “Book review: Range” (2019).Popular
- Interview: B. Stulberg, “David Epstein makes the case for being a generalist”, Outside, 22 May 2019.Popular
- A. Güllich, B. N. Macnamara and D. Z. Hambrick, “What makes a champion?”, Perspectives on Psychological Science 17 (2022); A. Güllich, M. Barth, D. Z. Hambrick and B. N. Macnamara, “Recent discoveries on the acquisition of the highest levels of human performance”, Science 390 (2025); A. Güllich et al., “From youth basketball to the NBA”, Sports Health (2026).Scholarly
- R. M. Hogarth, T. Lejarraga and E. Soyer, “The two settings of kind and wicked learning environments”, Current Directions in Psychological Science 24 (2015); D. Kahneman and G. Klein, “Conditions for intuitive expertise: a failure to disagree”, American Psychologist 64 (2009).Scholarly
- D. Z. Hambrick, B. N. Macnamara and F. L. Oswald, “Is the deliberate practice view defensible?”, Frontiers in Psychology 11 (2020), summarising Macnamara, Hambrick and Oswald (2014) and its 2018 corrigendum; K. A. Ericsson, “Given that the detailed original criteria for deliberate practice have not changed…”, Psychological Research 85 (2021).Scholarly
- J. Dunlosky, K. A. Rawson, E. J. Marsh, M. J. Nathan and D. T. Willingham, “Improving students’ learning with effective learning techniques”, Psychological Science in the Public Interest 14 (2013); E. L. Bjork and R. A. Bjork, “Making things hard on yourself, but in a good way”, in Psychology and the Real World (2011); E. Murray, A. J. Horner and S. M. Göbel, “A meta-analytic review of the effectiveness of spacing and retrieval practice for mathematics learning”, Educational Psychology Review 37 (2025).Scholarly
- D. Rohrer, R. F. Dedrick, M. K. Hartwig and C.-N. Cheung, “A randomized controlled trial of interleaved mathematics practice”, Journal of Educational Psychology 112 (2020), as summarised on the IES project page; M. Brunmair and T. Richter, “Similarity matters: a meta-analysis of interleaved learning and its moderators”, Psychological Bulletin 145 (2019).Scholarly
- G. Sala and F. Gobet, “Does far transfer exist? Negative evidence from chess, music, and working memory training”, Current Directions in Psychological Science 26 (2017); B. Bediou et al., “Meta-analysis of action video game impact on perceptual, attentional, and cognitive skills”, Psychological Bulletin 144 (2018); E. Bigand and B. Tillmann, “Near and far transfer: is music special?”, Memory & Cognition 50 (2022).Scholarly
- S. E. Carrell and J. E. West, “Does professor quality matter?”, Journal of Political Economy 118 (2010); M. Braga, M. Paccagnella and M. Pellizzari, “Evaluating students’ evaluations of professors”, Economics of Education Review 41 (2014).Scholarly
- O. Malamud, “Discovering one’s talent: learning from academic specialization”, ILR Review 64 (2011), and “Breadth versus depth: the timing of specialization in higher education”, Labour 24 (2010); L. Bridet and M. Leighton, “The major decision”, working paper, University of St Andrews (2015).Scholarly
- P. E. Tetlock, Expert Political Judgment, new edition (Princeton University Press, 2017); F. C. Keil, “When and why do hedgehogs and foxes differ?”, Critical Review (2010).Scholarly