Note: this is an update of a blog I wrote a couple of years ago on another site. Other than some grammatical fixes, none of the content has changed. Reading this book was a major inspiration for starting this blog on how to live the rest of your life.
The theme of “Range” is simply that generalists often do better than specialists, even in niche topics like sports, musical performance, or research and development. It resuscitates old ideas that original thinking comes from inspiration, not just perspiration; that too much (deep) education can be a bad thing, while just reading broadly is enormously valuable even when not directed.
I found it highly relevant to my own life, and to working at Microsoft, and in the cloud industry. Many people I know have read and talked about this book: from colleagues at work to one of my online guitar instructors. Many of them have said something like, “this is a flawed book that could use editing, but it spoke to me and made me feel better about my chaotic, disorganized approach to life and learning”. Or at least, that’s how I feel about it, and I’ve interpreted the feedback of others based on that.
I almost didn’t finish reading the first time through because I thought it bogged down and became repetitive after chapter 7. But I plowed through to Chapter 11, which is about how data-driven organizations full of deep technical expertise and using rigorous processes can fail spectacularly. Suddenly I found myself questioning the entire tech industry and how these ideas can be applied. It was enough that I read the book again and started recommending it widely. The second time through, I noticed even more flaws: the anecdotes often can be interpreted the opposite of the author’s intent completely. But I also more deeply understood its message about the complexity of the real world and how context is good, even when it increases ambiguity. I also think the book itself may reflect the theme of Chapter 4, that we learn better when we struggle with the material.
If you can’t get through all of it, read to 7, and then read chapter 11 if you work at any large organization.
I also highly recommend Chapters 6 and 7 to anyone thinking about changing jobs or starting a new part of their education. I wish I had read this when I was about 16 years old. The premise of these chapters is that many successful people try many different things before they settle into their career, and often even then, they change into completely different fields late in life.
A shorter review is here: https://www.npr.org/2019/05/28/725755061/range-argues-that-specialization-should-not-be-the-goal-for-most
Murat’s view here: http://muratbuffalo.blogspot.com/2019/06/book-review-range-why-generalists.html
Introduction: Roger vs Tiger
Tiger Woods started golfing before he could speak, and his father kept him entirely focused on that sport growing up. Roger Federer played a wide variety of sports growing up and only started focusing on Tennis fairly late. Contrary to intuition or common lore, it turns out that Roger’s path is the more common. Studies show that most elite athletes start by playing many different sports and only specialize later.
Comment: the book ignores the obvious conclusion that truly elite athletes are simply more naturally athletic and thus find it easier to play many different sports than the rest of us. I kept coming back to this thought as I read the rest of the book.
Epstein’s path to “Range” started with looking at athletes for his earlier book “The Sports Gene” and learning about “Roger vs Tiger” and the advantages of late specialization and then found out the same was true in many areas. It seems that people do best who sample many different specializations, genres, college majors, or careers and then pick one where they have a particular affinity, skill, or just like. Going further, Epstein started to find out about the dangers of specialization, including the “if all you have is a hammer…” problem and local optimizations.
Chapter 1: The Cult of the Head Start
This section talks about Laszlo Polgar’s successful experiment raising three daughters to be chess prodigies, including his daughter Judit, the strongest female chess player in history and the only woman ever in the top 10 among all players. This story and that of Tiger Woods seem to demonstrate that supreme talent can simply be trained from an early age and suggests that any other path will tend to mediocrity. The study of savants also indicates that single-minded focus can have spectacular results. AI research shows that general intelligence is not necessary in games like chess, go, and even RTSes.
Not everything in the real world is a closed game with carefully defined rules and deterministic outcomes. Real-life (the wicked world) often involves unclear, changing rules, and in these cases, expertise is of less use. Studies have shown that experts do worse than amateurs in games where the rules are suddenly changed: they are misled by their own expertise. No savant, no matter how impressive their skills in arithmetic or music, has ever been a Big-C creator: a great composer or award-winning mathematician. In fact, Nobel Prize winners have been shown to be at least 20 times more likely to spend considerable time at other hobbies, skills, or fields outside their award specialty than the general scientific population.
Research shows that top performers use analogies and draw inspiration from a wide range of areas.
Chapter 2: How the Wicked World was made
The Flynn Effect (Flynn is a New Zealand Political Scientist) is that average IQ has been rising for over a century. IQ scores always average to 100 because, over time, researchers have made the test harder. Given the same test from 100 years earlier, an average IQ 100 woman today would score in the 98th percentile. Controlling for nutrition and education explains some of this, but not all. Tests of general knowledge have hardly budged in this time, but abstract tests about “similarities” and tests designed to be culturally neutral about problem solving showed the biggest improvement.
In the 20s and 30s, the Soviet Union reached into illiterate peasant communities and introduced education, collective farms, and industrial development. Psychologist Alexander Luria decided to test how these changes also changed the minds of the people. He found that their ability to organize things (colors, thread, animals) in abstract groupings increased (aside: reading is an abstraction; does reading itself change the mind’s ability to handle abstraction?). In turn, pre-modern people seem to be less susceptible to optical illusions.
The modern world is full of abstractions and requires conceptual thinking that can jump from one particular to another. But Flynn observes that modern education is increasingly specialized: pushing people early into studying facts and specifics rather than transferable abstract thinking. And his studies have shown that students in deep majors often are unable to apply scientific thinking outside of their own domain.
To counter this, some universities are introducing broad interdisciplinary majors. For example, the University of Washington has a course called “Calling Bullshit” focused on critical thinking and broadly applying philosophy and scientific processes.
Chapter 3: When Less of the Same is More
17th century Venice saw a revolution in music performance, and composers like Vivaldi took advantage of this to create new, more complex concertos and solos. These performers were foundling women abandoned at the Hospital of Mercy in Venice and trained their whole lives to play a wide range of musical instruments (as well as reading, writing, other vocational talents).
Like with sports, in the modern world. It is often assumed musical talent must be developed early, and kids must choose their instrument as young as possible. But studies have shown that the best musicians come from less musical families and spend less time practicing and less on a single instrument at first. Likewise, students who play more instruments do better playing their primary. However, classical musicians indeed seem to do better with focus and early start, while Jazz musicians start much later and are less likely to specialize in one instrument.
It turns out to be rare for musicians to be world-class in both jazz and classical music. Moreover, classical musicians, no matter how talented, often find it difficult or impossible to improvise, while many jazz musicians have no musical training and often cannot read music. As a result, for many (most?) learning music seems to be like learning a language: where children do it simply by imitation.
“Breadth of training leads to breadth of transfer.” The more contexts in which things are learned, the easier it is to abstract the principles and apply them elsewhere.
Chapter 4: Learning, Fast and Slow
Studies of education around the world show that students do better when they struggle to learn. Teachers who guide the students may help them complete their homework and score well on tests, but it has been shown the students retain less later. Standard measures of teacher quality, including test scores and student assessments, seem to contraindicate long-term learning (I was reminded that I learned the most from the college professor who flunked almost everyone out of numerical methods).
“Spacing” is the technique of leaving time between bouts of deliberate practice. Students who study straight through do worse than students who study less, with breaks where they think about something else: sometimes these differences are spectacular. But spacing, testing without hints and guidance, and using “making-connections” questions are shown to improve long-term results but make short-term learning harder. Blocked practice is another example, where focus on a single subject does better over the short term but worse over the long term than Mixed practice where different types of problems are studied together.
Head Start and other early education programs show great short-term results. But students tested later in life do no better and sometimes worse. So it may be that early education programs are teaching things kids will learn anyway naturally later.
Chapter 5: Thinking outside experience
Kepler struggled for years to understand the motions of the planets, why further planets moved more slowly than nearer, or why they even moved at all. Stuck with very little data, he used analogies to think about theories, eventually coming up with the idea of gravitation force based on thinking about magnetism. When faced with new situations, we reach out to analogous examples, and this hugely, other consciously and unconsciously influences our thinking. Furthermore, it has been shown experimentally that thinking too close inside the subject matter can be misleading: we tend to focus too much on deep experience when faced with a problem that may be much different. Kahneman and Tversky call this “the inside view.” When experts are asked to estimate how long a project will take or the expected outcome of an investment, they break it down into well-known sub-problems and chain these estimates together: and usually vastly underestimate the difficulties therein. Whereas when they are asked to step back and think generally about similar situations, ignoring the details, they make better estimates. Instead, algorithms like Netflix’s movie preference rating (what do people who liked similar movies like) work by general similarity or analogy.
Studies show groups with different experiences, working together and not focused too much on details do better than groups of experts at solving many tasks.
Chapter 6; The Trouble with Too Much Grit
Van Gogh tried many different careers and artistic pursuits for many years, failing at each, before trying painting. Then, when he found painting easier than expected, he still gyrated from one technique and subject to another, often spending only a few hours dashing out a painting that later turned out to be priceless. JK Rowling and Gaugin are other examples of people who failed spectacularly and repeatedly before finding a match of talent, interest, and work.
The US Army studied West Point students to guess which would pass through the incoming “Beast Barracks” of basic training and orientation. The Whole Candidate Score based on academic results and physical fitness measures failed to predict success, but a Grit test that asked students about their work ethic and singular focus did. However, there are two flaws with this: the Whole Candidate Score is only used against the relatively small group that has already chosen and been selected to attend West Point; and it isn’t clear whether the people who stuck through Beast out of pure grit really should have…some of them persevered at something that may not have been the right choice.
West Point graduates were dropping out of the military as soon as their mandatory service allowed. Attempts to entice young officers to stay in the military with higher pay backfired: the best officers left anyway, pocketing the extra income. While the knowledge economy in private industry encouraged fail-fast and quick career changes, recruiting these sort of people into the rigid military just led to them leaving for other opportunities. This trend was only reversed when candidates were given more choice on their career path: the branch and type of service and education they were allowed to pursue, improving the chance of finding a match to the kind of work they like.
A young person benefits from trying multiple different kinds of experiences, even though these may look like risky career moves until they find a great match.
Chapter 7: Flirting with Your Possible Selves
Studies of highly successful people find that they are more likely to have gone through various different careers and non-traditional paths rather than sticking to one track, company, or role. Instead of making long-term plans, these very successful people say they think short-term or without any plan at all but are willing to try new things. Broader studies of personal preferences and tastes over 10+ years show that people change much more than they expect or remember later. Even personality traits like introversion tend to change over time. Thus choosing a career path early may lock one into a direction that makes no sense for the person as they change.
Chapter 8: The Outsider Advantage
Eli Lilly had a list of molecules they were trying to synthesize unsuccessfully. Finally, they posted the problems openly on a website and asked for suggestions. The VP of research had realized that most hard problems were solved by “clever insights” rather than hard work. Outsiders from various fields, such as lawyers, solved chemistry problems that had eluded highly trained experts, usually by drawing from analogous problems and parallel solutions. This process was spun out of Ely Lilly into a startup called InnoCentive, and NASA and others have used it. Specialists look too closely at the details or try to fix minor problems in incorrect approaches, whereas outsiders can see broad solutions.
This problem can be even worse in academic research, where journals accept papers focused on narrow subspecialties and incremental improvements. As a result, problem-solving gets divided up into specialized areas with no obvious ways of connecting them.
Chapter 9: Lateral Thinking with Withered Technology
This chapter is primarily the story of Gunpei Yokoi, who was hired by Nintendo to service card-making machines but was a tinkerer in his spare time. One of these projects turned into a toy that was a massive seller. After designing a complex racing game that failed, he focused on designing simple toys and games based on older technology. Doing this made manufacturing easier and less expensive, and quicker to market where it could be tested. The NES and even earlier “Game and Watch” were fun, cheap, easily fit in a pocket, durable, and had good user experiences.
3M studied inventors and categorized them into two buckets: specialists and generalists. They found both types equally likely to get patents and innovation awards. But they found a rarer group of deep specialists in one area who were also broadly capable in many other areas. These polymaths were the most likely to be the most innovative, drawing on their knowledge of adjacent realms and applying it to their specialty.
In a study of comic book authors, neither long experience nor availability of resources predicted hit comics. The most successful authors were instead those with the broadest range of different genres on their resumes. They also found that individual superstars were more likely to come up with hits than teams.
Serial innovators often do not fit into the specialized roles found in large companies. They do not match job descriptions and often have resumes full of gaps and many switches. Overall, deep specialization is required in well-defined, deep problem areas, but generalists do just as well or better in undefined areas.
Chapter 10: Fooled by Experience
In his 1968 book “The Population Bomb,” Paul Ehrlich predicted resource shortages would cause 100s of millions, even billions of deaths by the end of the 1980s. Instead, commodity prices dropped, and the worldwide food supply per person is higher than ever in history. Research shows that such big idea pundits are usually wrong, even as they amass more and more information and expertise backing their cases. However, other groups have proven to be surprisingly adept at making predictions.
Phillip Tetlock studied a wide range of expert forecasters over 20 years and found they did poorly at all time scales and domains, regardless of expertise. However, “integrators” who based their forecasts on examining many ideas from a range of experts and non-experts and synthesizing their predictions were surprisingly capable. These integrators tended to reject simple “big” ideas and embrace ambiguity. Based on this research, the Intelligence Advanced Research Projects Activity (IARPA) organized a prediction tournament, where teams competed to predict economic measures and global events. Tetlock called openly for volunteers and people with wide-ranging interests and no specific deep expertise. This team so thoroughly beat the expert teams that the study was reorganized in year 2 to remove the experts. Small teams of generalists also beat crowd-sourced predictions and prediction markets. Tetlock calls these teams “superforecasters.” They are characterized by being open-minded, switching their views, and integrating ideas and perspectives from many sources (aside: I a reminded of Nassim Taleb, who, when he was a trader, read the Economist and poetry instead of watching his Bloomberg terminal).
While superforecasters remain open-minded, studies of the human population find that they are unwilling to read views contrary to their own, and scientifically literate adults are actually more likely to become dogmatic about scientific topics than the less educated, even when they are demonstrably wrong. Further studies found “scientific curiosity” to be more valuable than scientific knowledge. Research also shows that forecasters can get better by being trained to look for similar patterns in different domains, study failed predictions, and use other “foxy thinking” techniques.
Chapter 11: Learning to drop your familiar tools
This is a difficult chapter. It epitomizes the strengths and failures of the book: engaging, readable stories and anecdotes that don’t clearly support the thesis, inadequate citation, cherry-picking details, and contradictions in conclusions. Arguably, the book’s message is that the real world is messy and full of ambiguity, and this chapter reflects that. Or maybe it just needed an editor. However, the chapter makes valuable organizational insights about the dangers of conformity and congruence and the value of open communications, encouraging dissent, and breaking down hierarchy.
The first story is about NASA and the Challenger disaster. Engineers had a hunch that launching the shuttle on a cold day could lead to O-ring failure but, faced with insufficient data, went ahead with the launch. NASA had a long history of being data-driven, but without enough data, engineers could only follow procedure. In the next story, Epstein talks about forest firefighters who could have run to safety but instead died with their heavy packs and tools, even when ordered to drop them. Epstein makes the analogy that NASA could not give up on its process even when the data was lacking: a “mistake of conformity.”
NASA developed its data-driven culture under von Braun, who had listened to engineers at all levels of the organization and instituted the practice of “Monday Notes.” In this practice, every engineer submitted weekly single-page summaries of their top issues; von Braun wrote comments on these notes and circulated them back to the entire organization. This process informally spread information up and down the hierarchy and across the organization, alerted the organization early about problems, and allowed distant engineers to draw on similar solutions in other areas. After von Braun, the notes became formal status reports and only flowed upwards. Management no longer welcomed bad news or recognized the value of spotting problems before they became critical. The Columbia disaster showed the same sort of culture of conforming to process (some engineers asked the DoD for pictures of the shuttle to see if it was damaged, and management blocked the request and apologized for the violation of protocol). In contrast, Rex Geveden, the manager of Gravity Probe B, listened to engineers concerned about one potential problem and delayed their launch, despite administrative pressure. More problems were discovered while fixing the probe, which would have almost certainly failed.
The chapter concludes that data-driven organizations, full of smart experts, fail if they are too conformant to rigorous process because they do not acknowledge hunches or intuition when there is not enough data. Reporting upward simplifies complex situations and makes the problem seem more black and white. Instead, Geveden says, “The chain of communication has to be informal and completely different from the chain of command.” I think anyone who works in a large tech company can probably see the value of this idea.
Chapter 12: Deliberate Amateurs
In this chapter, Epstein tells the stories of a series of scientists and researchers who were successful by experimenting on crazy ideas, often outside their areas of expertise. These scientists and labs encourage playful experiments and interdisciplinary cross-fertilization. At John Hopkins, one program teaches courses about types of evidence and actively studies scientific errors (sounds like the “bullshit” course at UW). Researchers studying very creative projects found boundaries between teams were porous, allowing communications and allowing people to move between organizations and specialties. Papers that cite papers across a wider range of sources, especially journals not commonly cited, started out slowly but were more likely to be more influential 15 years later.
Conclusion: Expanding your range
In the conclusion, Epstein confirms my suspicion that he wrote Range based on his experience researching “the Sports Gene.” He wanted to advise athletes and their parents not to specialize in one sport too early and found this same advice applied across domains. Epstein leaves us with the advice to not feel left behind because you didn’t start early, try different things, and treat work and life as a series of experiments.