In my book summary for Range: Why Generalists Triumph in a Specialized World, I explain how David Epstein pushes back against specialization by highlighting the benefits of range.
Buy Range at: Amazon | Kobo (affiliate links)
Key Takeaways from Range
- Specialisation is overrated:
- We face a lot of pressure to specialise, and to specialise early. There is no entrenched interest pushing for greater range.
- Specialisation is great at solving “kind problems”, where the goals are clear and you get accurate, fast feedback. In these cases, specialisation is efficient. But range is better at dealing with “wicked problems” (which most problems in the modern world are).
- Range makes us more creative and innovative:
- When we learn things in more contexts, it improves our abstract reasoning abilities.
- Abstract reasoning then helps us apply existing knowledge to new situations, such as by using analogies.
- Epstein refers to many studies and examples that show how range has helped people be innovative or make important discoveries.
- Teams and organisations with diverse backgrounds (greater range) also tend to be more innovative.
- The benefits of range also extend to teams and organisations:
- Diverse teams tend to be more successful and have a higher chance of making breakthroughs.
- Some degree of incongruence in an organisation is good.
- Sampling leads to better match quality (how well-matched a person is with their job, partner, etc):
- Narrow, deliberate practice should only come after sampling. Elite athletes and musicians often went through a early sampling period to find the sport/instrument that suited them. Only then do they ramp up the practice.
- Grit depends on context—it’s not a stable personality trait. When you get good fit, it looks like grit.
- People who switch careers or take circuitous paths can feel “behind” initially, but end up happier in the long run as they get better match quality.
- Things that look inefficient may be more effective in the long run:
- Unintuitively, learning is more effective when it’s slow and challenging.
- Early specialisation can give people a head start, but that lead vanishes or even reverses over the longer run as people with more range catch up.
Detailed Summary of Range
Specialisation is overrated
We are pushed to specialise early in many areas
Areas where specialisation is pushed include:
- Sports. Stories about child prodigies such as Tiger Woods starting golf at the age of 2 are memorable. Some people in sports may also have a strong financial interest in making early specialisation seem essential.
- Education. Even though most US college graduates end up working in careers unrelated to their majors, our education system tends to teach narrow, job-specific skills. (There are exceptions, such as the professors who wrote Calling Bullshit and taught a course by the same name.)
- Work. Large companies recruit for well-defined, specialised roles. In the book Serial Innovators, Abbie Griffin and her co-authors found that the most innovative employees repeatedly claimed that they would be screened out under their company’s current hiring practices.
Epstein clarifies that range isn’t always better than specialisation. He just thinks the current balance is too skewed towards specialisation, so his book is intended to serve as a counterpoint. Specialties are also become increasingly narrow. Oncologists don’t just specialise in cancer, but in cancer of a single organ.
… there is often no entrenched interest fighting on the side of range …
Specialists are better at kind problems, generalists are better at wicked problems
What are kind and wicked problems?
Kind problems have repeating patterns, with fast and accurate feedback. Examples include: chess, golf, tennis (but not as much as golf), firefighting, poker, surgery and playing classical music.
But most real world problems are not like golf and chess. Most problems are “wicked”. In wicked domains, the rules are unclear or incomplete; patterns may not repeat frequently or may not be obvious; and feedback is often delayed and/or inaccurate.
Generalists are better at wicked problems
The world is constantly changing and most problems we face today are complex, wicked problems.
We can’t assume specialist skills extend to “wicked” domains. [There’s a fun term for this—”ultracrepidarianism”.] Methods that are “tried-and-tested” in kind environments can lead us astray in new situations. For example, wilderness firefighters used to holding onto their tools cling to them to their death, even if they could’ve escaped by dropping the tools.
Study: Klein vs Kahneman
Psychologists Gary Klein and Daniel Kahneman both looked at whether experts made better decisions, and came to opposite conclusions. How could that be?
Kahneman and Klein ended up co-authoring a paper in 2009 to try and reconcile their differences. They agreed that whether or not experience always led to expertise depended on the domain. It did in narrow domains like chess, firefighting, and poker. Klein had found that firefighters and naval commanders made around 80-95% of their decisions instinctively, within mere seconds.
However, in broader domains like predicting financial or political trends or how employees would perform, “experts” fared no better than average (see further below).
Computers in particular tend to be better at kind problems. So, as humans, our comparative advantage is in wicked problems. This is Moravec’s paradox — the idea that machines and humans often have opposite strengths and weaknesses.
Example: Computers vs humans
There’s a difference between tactics and strategy. Tactics are short-term moves that give you an immediate advantage. Strategy is the big-picture planning—winning the war, rather than the battles.
Computers are far better than humans at tactics. For example:
- Chess is 99% tactics. Being able to remember different patterns is critical, so early specialisation is very useful. But computers have been able to beat chess grandmasters for a long time, thanks to being better at tactics. In tournaments where humans can team up with computers (“centaur” teams), amateur humans could crush grandmasters because the computers nullified the grandmasters’ advantage in tactics.
- StarCraft, by contrast, is a strategy game. Players have to manage battles, plan infrastructure, conduct spying, etc. Computers are much worse at StarCraft.
- IBM’s Watson was excellent at Jeopardy! but dismal at cancer care. The difference is we know the answers to Jeopardy! while we’re still trying to figure out the right questions with cancer.
But narrow specialisation can still be useful
Birds fly high in the air and survey broad vistas of mathematics out to the far horizon. They delight in concepts that unify our thinking and bring together diverse problems from different parts of the landscape. Frogs live in the mud below and see only the flowers that grow nearby. They delight in the details of particular objects, and they solve problems one at a time. …
It is stupid to claim that birds are better than frogs because they see farther, or that frogs are better than birds because they see deeper. … The world of mathematics is both broad and deep, and we need birds and frogs working together to explore it.”
Narrow specialisation can be very efficient when dealing with “kind” problems. Specialists are also useful for facts, but not opinions. They tend to have blind spots and struggle to get a good perspective.
Study: Superforecasting
In Philip Tetlock’s early work on forecasting, he found that highly educated “experts”, including those with relevant classified information, were no better at forecasting than average punters. But experts were more confident than average. Events they claimed were “impossible” or “nearly impossible” occurred about 15% of the time, and events they said would definitely happen failed to occur about 25% of the time. There seemed to be an inverse relationship between confidence and performance, as well as between fame and accuracy. [Makes sense, since fame tends to select for people who make “outrageous” predictions.]
Specialists performed particularly badly on predictions within their domain—and got worse with more credentials and experience in that domain. This is because they cherry-picked details that fit their existing theories.
Generalists performed better overall. The best ones—the “superforecasters”— were intelligent, curious people with broad interests who read widely, but had no relevant background. Superforecasters recognised that causality is complex and that causes are often probabilistic, rather than deterministic. They flip-flopped a lot, actively seeking ideas that contradicted their beliefs and updating their predictions as they got new information. Teams of superforecasters even beat the wisdom of much larger crowds and prediction markets.
Range makes us more creative and innovative
Specialists see just a small part of the overall puzzle. Generalists are better at “far transfer”, making connections between seemingly unconnected areas. An analysis of 18 million scientific papers found that the most widely-cited papers tended to combine knowledge from different fields.
Modern life requires abstract reasoning
Abstract reasoning is becoming increasingly important in the modern world and each generation seems to be getting better at it. The “Flynn effect” describes how each generation keeps doing better in IQ tests. The gains are almost entirely in abstract reasoning like Raven’s matrices or similarities tests, rather than in general knowledge, arithmetic or vocabulary.
Study: Remote villagers
Remote villagers struggle with abstract concepts like shapes, numbers and formal logic. For example:
- When asked to group shapes together, one villager insisted that filled and unfilled circles could not be grouped together because one was a coin and the other was a moon, and coins and moons had nothing in common.
- When given a group of 3 adults and 1 child and asked to identify the odd one out, they would say the child had to stay with the adults.
- If told that, all bears are white in the Far North and a specific town is in the Far North, villagers could not say what colours the bears were. They responded that only someone who has been there could know.
Despite researchers’ best explanations, examples and cajoling, the remote villagers simply could not use any reasoning that was different from that used in their daily lives. But people exposed to even a small bit of “modernity” could easily grasp such abstract concepts.
Interestingly, remote villagers tend to do better at optical illusions like the Ebbinghaus illusion. A possible explanation is that premodern people don’t see the holistic context, so their perception is not affected by the extra circles. Villagers miss the forest for the trees, while we miss the trees for the forest.
When we learn things in more contexts, it improves our abstract reasoning
When you learn something in different contexts (“interleaving” or “mixed practice”), you create abstract models instead of relying on a particular example. In contrast, “blocked practice” is when you practise problems of the same type right after each other.
While people think they learn better with blocked practice, the opposite is true. With blocked practice, you learn procedures; with mixed practice, you learn how to differentiate types of problems and work out which procedure to use. [The difference between learning procedures vs making connections is explained further below. See also A Mind for Numbers for more on learning and interleaving.]
Mixed practice therefore helps you match the right strategy to a problem. When you can do this, you don’t just learn from experience. You may figure out the right strategy even without experience. (Mixed practice can also apply to physical skills. Robert Bjork, who has studied mixed practice has suggested that your ability to shoot a basket from the free-throw line may improve by practising behind and in front of the line, rather than always at the line.)
Abstract reasoning then helps us apply existing knowledge to new situations
Analogies are a form of abstract reasoning. We can use analogies to understand unfamiliar problems, including things we can’t even see. Billiard ball collisions, for example, can help us understand molecular motion.
Surface analogies vs Deep analogies
There’s a difference between surface analogies and deep analogies. Although things that are similar on the surface will often also be conceptually similar (particularly in “kind” domains), surface similarities can lead us astray. This risk is higher if a similarity causes us to focus on a single example rather than an entire reference class.
Example: Surface similarities misleading
In one study, Stanford international students had to recommend the US’s response in a fictional situation where a large, totalitarian country threatened its smaller, democratic neighbour.
Students given irrelevant details that made the situation sound similar to WWII were far more likely to choose war than those given irrelevant details that made the situation sound like the Vietnam War.
In contrast, deep analogical thinking focuses on conceptual similarities from different domains that have little in common on the surface. Deep analogies let you transfer your knowledge to new and unfamiliar domains. For example, Johannes Kepler invented the field of astrophysics using analogies to explain the laws of planetary motion. Genetic algorithms are similarly based on the theory of natural selection.
Example: Duncker’s radiation problem
In this famous radiation problem, you have to figure out how to use rays to destroy a tumour in the body. To destroy the tumour, the ray has to be at a sufficiently high intensity. But at that intensity, it will also kill any healthy tissue on the way to the tumour.
A helpful analogy is that of a general trying to capture a fortress in the middle of the country. The general’s full combined forces are needed to capture the fortress, but each road leading to the fortress was such that only a small force could pass through it safely. The solution is to divide the forces into small groups, each approaching the fortress from a different road, and forming a concentrated, combined force when they reach the fortress.
Only about 10% of people solve this problem initially. When people are given an analogy, the success rate increases greatly.
Outside view vs Inside view
This distinction between surface analogies and deep analogies helps explain why predictions that take the “outside view” are so much better than those that take an “inside view”. The outside view is unintuitive because it relies on underlying structural similarities, rather than the unique surface features.
When taking the outside view, you should use multiple analogies—a “reference class”, ideally—instead of fixating on one. Research shows that when people are given more “internal” details about a problem, their decisions become more extreme (unless they’re prompted to take the outside view). [For more on the outside view vs the inside view, see my post on Decision Hygiene and my summary of How To Decide by Annie Duke.]
Studies and examples that show how range has helped creativity and innovation
Epstein provides many examples of creatives and innovators who had a lot of range. I only outline a few here.
Creativity
Example: Van Gogh
Vincent van Gogh had a long string of failures when he started out in art. He enrolled in art school when he was almost 33, but lasted mere weeks. When he entered the class drawing competition, the judges recommended he belonged in a drawing class for 10-year-olds. He experimented with lots of different art styles and methods, flip-flopping between realism, landscapes, colour, black and white, etc.
Van Gogh found success mere months before he died at the age of 37, after Claude Monet gave his work a glowing review. If he’d died at 34 instead, he’d be nowhere near as famous as he is now.
Study: Comic book writers
One study by Alva Taylor and Henrich Greve looked at what made comic book writers successful. They had expected those who created more comics would make better ones on average, learning by repetition. But it turned out that a high-repetition workload negatively impacted performance as people became overworked. Instead, the key was how many genres a creator had worked in. An individual who had worked in 4 or more genres was more innovative than a team that had collectively worked across the same number of genres, because the individual was better at integrating those differences.
[Funnily enough, an example from Atomic Habits by James Clear comes to the opposite conclusion. In Atomic Habits, Clear writes about how Jerry Uelsmann, a photography professor, divides his class at the start of the semester into a “quantity” and a “quality” group. The quantity group would be judged solely on the amount of work they produced, while the quality group would be judged by their single best photo. At the end of the semester, Uelsmann found that all the best photos came from the “quantity” group. I didn’t include this story in my summary of Atomic Habits, because it was just an anecdote rather than a study. I also found the result unsurprising—you’d expect the quantity group to produce far more photos so, statistically speaking, you’d also expect them to produce the best photos.]
Scientific Innovation
Example: Claude Shannon, founder of information theory
Claude Shannon learned in a philosophy course that you can solve logic problems like math equations by using values of 1 and 0 to denote true/false statements. (This was based on George Boole’s work, hence the term Boolean values.) He later used this encode and transmit information more efficiently.
[I learned about Shannon in my Complexity Explorer course and did a bit more Googling on him after reading this. He studied mathematics and electrical engineering and was a true tinkerer:
I’ve always pursued my interests without much regard for final value or value to the world. I’ve spent lots of time on totally useless things.
He wrote his master’s thesis when he was just 22-years-old. By showing how algebra could be used to represent electronic circuits, this allowed people to test circuit designs mathematically without building them. Some have called his thesis possibly the most important master’s thesis of the century.]
Study: Patent inventors
Andy Ouderkirk used an algorithm to analyse 10 million patent by inventors at 3M. He found:
- Specialists and generalists both made contributions. Unsurprisingly, those who had neither depth nor breadth rarely did.
- Specialists’ contributions were high from around WWII to 1985, but have declined quite dramatically since. Ouderkirk thought this may be because specialists become less valuable as information becomes more broadly available.
- Polymaths, who had depth in one area (but not as deep as the specialists), as well as breadth (sometimes more than the generalists), turned out to be most likely to succeed at the company.
Research shows that more successful teams tend to be more diverse. Scientists who have worked abroad are also more likely to make a greater scientific impact.
Study: Diverse research labs make more breakthroughs
In the 1990s, Kevin Dunbar looked at how the world’s top research labs made new discoveries. He found the labs most likely to make breakthroughs made many analogies from various domains. Labs whose scientists had more diverse backgrounds made a greater variety of analogies and were more likely to make breakthroughs. One lab didn’t make any new findings during Dunbar’s project—everyone in that lab had similar and highly specialised backgrounds, and rarely used analogies.
Once, Dunbar saw two labs encounter the same problem. One lab just had E. coli experts while the other had scientists with a mix of backgrounds. The former lab got stuck on the problem for weeks, while the latter solved it quickly using an analogy from one person’s medical background.
Outside perspective
Sometimes outsiders can bring a different perspective and solve problems that specialists can’t. InnoCentive [now Wazoku] lets organisations in any field post “challenges” for people to solve. Around a third of posted challenges get solved, which is quite impressive because these are problems which have stumped specialists.
Example: Alaskan Oil Spill
In 1989, a massive Exxon oil spill occurred in Alaska. Because of the low temperatures, the oil–water mixture has the viscosity of peanut butter and is really hard to remove. Almost 20 years later, lots of oil still remained.
In 2007, the Alaska-based Oil Spill Recovery Institute posted a challenge on InnoCentive. John Davis, a chemist, came up with the winning idea. An analogy of a slushy came to mind, where you have to move the straw around to stir it and get the slushy out. Davis thought back to a day where he saw “concrete vibrators” being used to stop concrete from hardening out in the sun. He came up with the idea of using a similar idea to loosen the oil–water mixture. The solution was just three pages, including diagrams. And it worked.
Range in organisations and groups
Incongruence improves problem-solving
Epstein suggests that organisations should strive for incongruence.
Congruence is when the various parts of an institution—its values, goals, leadership styles, etc—all align. People like congruence and consistency, but researchers have found that cultural congruence does not increase chances of organisational success.
Incongruence, on the other hand, builds in cross-checks and makes for a more effective problem-solving culture. An organisation can become incongruent by identifying its dominant culture and deliberately diversifying away from it.
Chain of command vs chain of communication
There’s a difference between chain of command and chain of communication. Disagreement should be welcomed while decisions are being made, even if compliance is expected once decisions are finalised. It’s hard for decision-makers to sufficiently understand their organisation by just listening to voices at “the top”.
For example, Girl Scouts CEO Frances Hesselbein had a “circular management” model. Instead of a ladder, the organisation was made up of concentric circles, with Hesselbein at the middle. Anyone in one circle had multiple entry points to communicate with the next circle, rather than a single gatekeeper.
Example: The Challenger Launch
NASA had a very strong technical and quantitative culture. Their motto was, “In God We Trust, All Others Bring Data”, meaning that engineers had to produce data to back up any claims. The culture was a very congruent one.
Before the Challenger launch, engineers had concerns about the O-rings based on photographs, but they couldn’t quantify their concerns. The concerns turned out to be valid. Epstein suggests that NASA managers should have “dropped their tools” and improvised in that situation.
[This is easy to say with hindsight (though, to be fair to Epstein, he points this out himself). How do you know when to “drop your tools” and when to stick with the plan? In the book Noise, the authors point out that human judgement is very noisy, such that almost all models and rules outperform humans. It seems possible that NASA’s strong quantitative culture might serve them better in more cases than a culture which gives more weight to people’s “hunches”, even if it let them down in the Challenger example.]
Inefficiency can be good
The push for specialisation is often driven by a desire for efficiency. However, inefficiency can sometimes be good. For example, HIV is a retrovirus. When it emerged in the early 1980s, we already had knew a bit about retroviruses which, up to then, had just been regarded as a curiosity in animals. That knowledge helped us respond to HIV.
In recent years, Nobel Prize winners increasingly say that their breakthrough could not have happened today, because the push for efficiency has become too strong.
Truly original discoveries in science are often triggered by unpredictable and unforeseen small findings. … Scientists are increasingly required to provide evidence of immediate and tangible applications of their work.
Sampling leads to better match quality
“Match quality” describes how well you fit what you do, based on your abilities and preferences. Various studies have shown that people who specialise late, or switch careers, suffer an initial hit to skills and income because they have less specific skills. However, they catch up quickly on average and even overtake early specialisers, thanks to better match quality.
Study: College students picking majors
Economist Ofer Malamud compared students in Scotland with those in England and Wales. In England and Wales, college students had to pick their specialisation before even starting college, so they could apply to specific programs. In Scotland, however, students had to study different fields for their first two years of college, before landing on a specialty.
Malamud found that English and Welsh students were less likely to pick subjects like engineering that did not exist at their high school and were more likely to switch to a different field after graduating. Scottish students in contrast started out with lower incomes initially after graduating, but quickly caught up.
Narrow, deliberate practice should only come after sampling
Malcolm Gladwell’s book Outliers popularised the “10,000 hours” rule, which says that you need 10,000 hours of deliberate practice at something to master it. Many people have interpreted this to mean that kids should start early and practise a specific, narrow skill.
However, the evidence does not support this. Early on, elite athletes and musicians actually spend less time on deliberate practice than comparable (but not quite elite) performers. Instead, they usually go through a “sampling period” where they try different sports or instruments in a relatively unstructured way. Specialisation and deliberate practice comes later, once they have found a good match for their talents and interests.
Grit depends on the context
Angela Duckworth has written a lot about “grit” and the importance of perseverance. However, grit is not a stable trait. People can be gritty in one domain and not at all in another. So don’t force your kids to “grit” through things they don’t want to do, in the hope that they can become the next Tiger Woods. (Even Tiger says that his father never asked him to play golf; it was always his own desire to play.)
When you get fit, it will look like grit.
Knowing when to quit and when to grit is the tricky part. There’s a difference between quitting because something is difficult and quitting because the match quality is poor.
Example: US Army retention rates
The US Army cares a lot about retention rates, because they don’t want to spend a lot of time and money investing in someone to be an officer if they’re going to drop out early.
One way they tried to increase retention was by offering bonuses for staying extra years. It didn’t really work and turned out to be a massive waste because the only people who got the bonus were those who planned to stay anyway. Those who wanted to leave still left.
A much better option was to help officers find better match quality. The Army did this by allowing officers to choose a branch or a geographic post, in exchange for 3 additional years of service. Thousands of cadets signed up for this.
Switching and sampling in careers can make people feel behind, but they end up happier in the long run
Career switching and sampling builds range, which helps people find roles that suit them (good match quality) and to succeed in those roles. Fulfilled and successful people often took unusual and circuitous paths earlier on, as they looked for good match quality.
Study: the Dark Horse Project
Todd Rose and Ogi Ogas looked at people who had found career success and fulfilment despite taking circuitous career paths. They thought they’d have to interview around five successful people for every one that took a circuitous path, but it turned out almost every successful person did. And almost all of them thought they were unusual—hence the name, “dark horse”.
Rose and Ogas found that their “dark horses” all planned for the short-term, and only pursued long-term goals after going through a sampling period.
Good match quality can be hard to find as people change
Finding good match quality tends to be circuitous because people change. Career quizzes and “strengths-finders” are popular because people want answers about what they should do. They help us pigeonhole ourselves. But we consistently underestimate how much we will change in the future, even as we marvel at how much we’ve changed in the past. Dan Gilbert calls this the “end of history illusion”. Even things we think of as “core values” can change significantly, and people change the most in their late teens and twenties.
The best way to maximise match quality is by sampling, reflecting and adjusting. We learn who we are in practice, not in theory. Specialising early is therefore risky. Christopher Connolly has found that people who made successful transitions to a different career were those who had broad training early in their careers.
Instead of working back from a goal, work forward from promising situations. This is what most successful people actually do anyway.
Switching careers can be a gradual process. Herminia Ibarra studied mid-career professionals in their 30s and 40s who switched careers. Their new work identities did not magically appear overnight. Instead, people took smaller steps, such as trying a new role temporarily. Some career changers got richer while others got poorer. All felt they were “behind” for a while, but they all eventually felt happier with the change.
Range is useful at the management or executive level
In the industrial era, employees’ roles tended to be narrow and specialised and job switching was low. The growth of the knowledge economy by the 1980s increased demand for broad conceptual skills. LinkedIn has found that one of the best predictors of who will become an executive is the number of different job functions they have worked across. Executives with more varied careers are, in turn, more likely to have novel strategies.
Example: Frances Hesselbein, Girl Scout CEO
When Frances Hesselbein was 34, someone asked her to lead a local Girl Scout troop as a volunteer. At the age of 54, she started her first professional job as an executive director of the local Girl Scout council. Then six years later, she became CEO of the Girl Scouts—then a 3,000,000 member organisation.
At the time, the organisation was facing an existential crisis as girls were applying to colleges and careers in unprecedented numbers. Hesselbein modernised the Girl Scouts. She increased focus on activities involving math, science and technology and on diversity. The organisation grew under her leadership and minority membership tripled. Hesselbein led the Girl Scouts for 13 years. Six weeks after leaving that role, she became the CEO of what is today the Frances Hesselbein Leadership Foundation, a non-profit.
Hesselbein never had any “formal” training or long-term plan. She took a meandering path, doing whatever seemed interesting or helpful in the moment.
Things that look inefficient can be more effective in the long run
Learning is more effective when it is slow and challenging
I could show somebody in two minutes what would take them years of screwing around on the fingerboard like I did to find it. … It’s slow, but at the same time, there’s something to learning that way.
Many studies show that learning is more effective in the long run when it is slow, challenging and inefficient in the short run. Counterintuitively, when learning feels easy, it’s a sign you aren’t learning deeply.
Study: US Air Force math teachers
The US Air Force Academy randomly assigns cadets to Calculus I classes. Two economists looked at data of more than 10,000 cadets taught by nearly 100 professors over a decade.
They found students evaluated their teachers based on how they performed on tests immediately. The professors whose students performed best on Calculus I got excellent student evaluations. However, these professors were not great for them in the long run—on average, these top-rated professors’ students performed worse on more advanced classes.
Active recall and testing
For example, you learn better when you test yourself and try to generate an answer on your own, even when you’re wrong. Being wildly wrong seems to help even more.
When you struggle to recall something, it helps transfer the information from short- to long-term memory. That’s why spaced repetition is so helpful—it creates a level of “desirable difficulty” that improves learning. Counterintuitively, the greater the struggle, the better the learning.
Procedural vs Connections Questions
One study filmed actual classrooms around the world to observe what might account for countries’ different educational outcomes. In every country, teachers asked students two main types of questions:
- Procedural questions, which ask about something that has already been learned.
- Making connections questions, which require students to understand a broader concept. For example, a teacher may ask why a formula works or when the formula won’t work.
When teachers asked a hard connections question, students often tried to turn them into easier procedural questions. They did this by adopting a “trial-and-error” approach and eliciting hints from the teacher. Teachers from the higher-performing countries were more likely to let the kids slowly figure things out themselves, ensuring that connections questions remained that way. (Well-meaning parents can fall into the same trap, showing their kids an easier way to do their homework.)
Although procedural practice is important, it’s a problem if students see math (for example) as just a set of procedures instead of as a broader system. They won’t be able to understand how maths can be useful outside of answering questions in a classroom.
The head start from early specialisation vanishes or even reverses in the longer run
Epstein refers to studies in several domains that show a head start from early specialisation (or procedural practice) does not hold up in the longer run:
- Early childhood education. Greg Duncan, Drew Bailey and others studied early childhood education programs like Head Start, which claimed to boost academic achievement. Such programs did give a head start, but any academic advantage quickly diminished and often went away completely. The researchers thought this was because the programs just taught procedural skills like reading that everyone picks up eventually, anyway.
- Academic articles. Researchers looked at around half a million academic articles, and found that articles that combined knowledge in new ways were less likely to appear in top journals and more likely to be ignored on publication. However, after about 3 years, these combination articles outperformed more conventional articles. After 15 years, those articles were far more likely to be in the top 1% most-cited papers.
- Careers. A 2017 study which looked at data in 11 countries found that, straight after college, people who specialised early tended to find jobs quicker and had higher earnings initially. However, those advantages faded over time and they ended up with lower lifetime earnings. The researchers also thought that a general education tended to be more valuable in fast-growing countries.
Other Interesting Points
- The long version of the phrase “jack of all trades, master of none” is: “A jack of all trades is a master of none, but oftentimes better than a master of one.”
- Chess grandmasters do not have photographic memories.
- While they perform very well at recreating chessboards with a plausible game arrangement, they do no better than amateurs at chessboards with arrangements that would never happen in a game.
- Grandmasters are merely better at chunking chess patterns so that they only need to remember a few meaningful chunks instead of 28 separate pieces.
- Researchers have found similar results with musical savants, who are able to recall tonal music much more easily than “atonal” music that does not follow familiar harmonic structures.
- Nobel laureates are 22 times more likely to partake as an amateur actor, dancer or other performer than other scientists. Nationally recognised scientists are also much more likely than other scientists to be musicians, sculptors, painters, mechanics, poets, writers, etc (and Nobel laureates are far more likely still).
- The Flynn effect is very strong, amounting to gains of 3 points every 10 years. That means an average adult today would be in the 98th percentile compared to adults 100 years ago.
- Einstein spent the last 30 years of his life looking for a single “theory of everything”.
- Charles Darwin’s first four models of evolution were based on creationism. But Darwin was an actively open-minded person—he would make note of anything he encountered that conflicted with a theory he had, so he kept throwing away models that didn’t make sense.
- One story says that when Julius Caesar was young, he broke down in tears upon seeing a statue of Alexander the Great. At Caesar’s age, Alexander had already conquered many nations, while Caesar had not done anything remarkable at that time.
- Paul Gauguin was originally a stockbroker. He became a full-time artist at age 35, after the stockmarket crashed in 1882.
- Maryam Mirzakhani, the first woman to win the Fields Medal (a very prestigious maths prize), was not initially interested in maths and had wanted to become a novelist.
- Contrary to popular belief, the average age of a start-up founder is 45.
My Thoughts
I really liked Range. But Range told me things I wanted to hear, so I am a bit wary of confirmation bias on my part.
The book is well-researched, to the point that it’s practically drowning in examples and evidence. I’ve mentioned only a sample of them in my summary above. But the examples are relevant and compelling, and Epstein doesn’t rely on isolated examples of individuals like Van Gogh or Darwin at the expense of actual studies. (Funnily enough, he calls Malcolm Gladwell out on this in a debate at around the 11:45 mark: “When you restrict to only the handful of best performers in the world, all you’re left with is case studies. You’re just left with journalism, and not really science.”) Epstein also explains towards the end of the book that he wanted to give memorable examples specifically to counteract the “Tiger Woods” success stories that readily come to mind when people think about early specialisation.
But with so many examples, you need a really tight structure to keep it focused. In this respect, I found Range‘s structure a bit frustrating. It felt like there were particular examples (Lazlo Polgar, Tiger Woods, Van Gogh, Beast training) that Epstein wanted to highlight or rebut, and then he figured out the structure from there. A couple of times I got lost in a (very interesting) example but struggled to place that within the wider context. This made the book feel more repetitive than it actually was. Every example just seemed to support the broad point that “range is good for wicked problems”, whereas I think Epstein’s specific points were actually a bit more nuanced than that.
That said, I still enjoyed Range a lot. For reasons I’ll explain here, I suggest the main takeaway is more, “don’t worry if you naturally have wide-ranging interests or change careers later in life” and “don’t force your kids down a very specific path early on”, rather than “range is always better”. Despite being a book about breadth, Epstein is clear that there’s a place for depth and specialisation, too.
Buy Range at: Amazon | Kobo <– These are affiliate links, which means I may earn a small commission if you make a purchase. I’d be grateful if you considered supporting the site in this way! 🙂
What did you think of Range? Share your thoughts in the comments below!
If you enjoyed this summary of Range, you may also like:
- Range at an individual level vs a societal level
- Book Summary: Power: Why Some People Have It — And Others Don’t by Jeffrey Pfeffer (a book with lots of practical advice about gaining power in your career).
- Book Summary: So Good They Can’t Ignore You by Cal Newport (more career advice, provides something of a counterpoint to Range)
8 thoughts on “Book Summary: Range by David Epstein”
Another excellent summary, thank you very much!
I find it reassuring to remember that not all those who wander are lost”, and that we may connect the dots later.
I also think the takeaway about not forcing your kids to do one thing or another is a very good one.
I also appreciated the point about grit not being a stable trait and that it is a matter of finding the right match.
There was a lot of great info and detail here so I’ll be sure to return to this summary.
Well done.
Glad you liked it. “Reassuring” is a very good word to describe this book. I thought Epstein did a great job in highlighting the benefits of range without making it sound like range is the most important thing, ever!
I’ve been meaning to read this and I think I’ll read it next after reading this summary. Let’s discuss on Friday.
Awesome review! And I like how you’ve started to box off key examples and studies. It helps make the summary more digestible and structured, I think.
I find the breadth vs. depth of skill/experience debate very intriguing, so I’m glad to learn more about studies and research relevant to the topic. That said, I’ve always (and by always, I mean since 6 months ago when I really thought through this dichotomy) been of the opinion that it’s better to lean into skillset diversification rather than specialization, so it was nice to see some of my priors re-affirmed 🙂
Thanks! I’m glad to hear you like having the key examples and studies boxed off – it seemed especially important for this book as there were so many. I’ll try to keep it up for future summaries.
I too find the breadth vs depth debate intriguing. I do think there are benefits of specialisation/depth, which I’m planning to set out in a separate post and which Epstein did not really address. (Fair enough, since that was not the point of his book, and he repeatedly acknowledged there is a role for specialisation.)
Like many things, I think the answer ends up being “it’s a balance”, which feels very unsatisfactory. But I think Range acts as a valuable counterpoint to a lot of the pressures that push towards specialisation.
Yes!! Very salient point on the pressures to specialize, I completely agree. Maybe it’s just me speaking anecdotally as as a recent college grad, but I feel like I’ve all-too-often seen stories that place an emphasis on specialization (“Master your craft! Become an expert in this in-demand skill and you’ll be set for life!”). While I understand the value in holding a rare and sought-after skill, I always felt like those LinkedIn posts were missing the bigger picture. Anyways that’s enough rambling, thanks again for the summary!
Maybe to correct: The AI AlphaStar playing the Game Star Craft II is on Grandmaster level and beating the large majority of players.
Thanks for sharing that info. I see AlphaStar was only released to the public in 2019, the same year Range was published, so Epstein couldn’t have known. I also think his main point still stands in that chess is more to do with tactics than StarCraft is, and computers have been beating us for ages at chess.
But it’s good to know computers are beating us at StarCraft now, too!