Peak: Secrets from the New Science of Expertise by Anders Ericsson

Something like 50% of the books on the pop-psychology shelf of your local bookstore have a message like this: The greats in any field didn't get there mainly by being naturally gifted; they got there by practicing a lot. Therefore, if you want to get a lot better at what you do, you just need to work hard and practice too.

Something like 49% of pop-psychology books have that message and support it by citing Anders Ericsson’s research. Peak: Secrets from the New Science of Expertise is Ericsson’s attempt to tell the story himself, walking a thin line between being more academically rigorous than those authors who cited him but not so rigorous that normal people will lose interest.

Take the infamous “10,000 Hour Rule” that Malcolm Gladwell describes in Outliers, which draws heavily on Ericsson’s study of elite violin students. Ericsson says that Gladwell isn’t necessarily wrong in the big picture but he’s sloppy in the details; he joins a long line of people saying this, of course, but he at least gets to move to the front of that line as the guy who actually did the study.

First, he lays out some trivial-but-surprising mistakes:

The study actually found that the best students had logged 7,400 hours by age 18; Gladwell got to 10,000 by saying they were on pace to reach 10,000 hours by age 20. That’s true but meaningless; even by age 20 those students wouldn’t have yet been world-class performers.
Even that 7,400-prorated-to-10,000 figure was an average in Ericsson’s study, not a minimum. By definition, approximately half of top performers were actually below the supposed threshold.
And even if there is a threshold, it isn’t the same across all fields. Ericsson describes a participant in one of his other studies who needed only 100 hours of practice to set a world record by memorizing and recalling 40 random digits spoken to him in a row, because who else would devote even that much time to something so arbitrary? (That's a rhetorical question; now that it's a thing you can compete in, of course people are doing just that, and the world record is now 456 digits.)

Ericsson’s more fundamental objection to the 10,000 hour rule, however, is that just doing something for a long time doesn't necessarily lead to improvement. Other people sometimes make this point by pointing to driving: most of us will probably drive for 10,000 hours in our lifetimes, but we don’t get better at it over time (although I couldn’t actually find any evidence proving or disproving this latter claim). Ericsson, more provocatively, focuses instead on doctors, pointing to studies showing they don't improve at their jobs at all once they're working full-time -- if anything they get slightly worse despite clocking in more and more hours.

The key to improvement, he says, is not just doing lots of practice, but doing lots of deliberate practice. Deliberate practice has a few specific criteria:

It consistently pushes you outside your comfort zone to try harder and harder things.
It requires your full focus, not something you can do on autopilot.
It involves specific incremental goals for you to measure your progress.
It is structured so you get feedback on how well you're doing and what you're doing wrong so you can work on it.
It includes guidance from a coach or expert who knows more than you and can accelerate your learning process.

Now, Ericsson's violin study didn't actually measure any of these factors -- he just measured how many hours the students had practiced in their careers and compared the best players to mediocre ones. So how can he draw the sweeping conclusions above? In looking at classical music study (and other fields with a strong practice culture, such as chess, sports, and dance), he finds that the above principles are already applied consistently. All of the music students received one-on-one coaching from experts, spent lots of time on individual practice, set challenging goals, and knew when they were playing well or poorly. (Those of us who just replayed what was already comfortable and skipped practices had already plateaued and dropped out before that point.) So since everyone was practicing the right way, the ones who did more of it did better.

In contrast, most of the things that doctors do aren't structured in a way that fulfills these criteria. They may set aside time for broadening their knowledge (such as going to conferences or reading papers), but that’s not the same thing as actually practicing their skills. Ericsson outlines what a hypothetical "deliberate practice" program would look like for radiologists: they would go through hundreds of novel, challenging images and make diagnoses, then see what the actual outcomes were, learn from where they went wrong and try those cases over and over again until they consistently succeed.

What makes the role of deliberate practice so special? The key differentiating factor between experts and non-experts is that they have extremely advanced mental representations of what they do. For example, chess grandmasters have a tremendous ability to recall positions of pieces during games, allowing them to play out complicated scenarios in their heads and determine the best move (or to play dozens of games simultaneously while blindfolded). This isn't because they have a superhuman memory; it's because they compress the board into what really matters -- what sort of game it represents, what the advantages and disadvantages are -- and then fill in the details from there. We know this because, if you arrange pieces randomly on a board instead of a snapshot from an actual game, suddenly grandmasters are no better than mortals at remembering them. (According to psychology teachers, this was a surprising finding at the time, but we do the same thing every day: you'll struggle if I ask you to recall the exact character string "ajaj ilukkpx snwhzifn wiz ppqn bzodb"; but you'll easily recall the same-length string "your grandfather answered the door naked," because you can compress all the characters into one easily recognizable representation.) Similarly, elite athletes are better than others at recognizing what's happening in a game, allowing them to make better split-second decisions -- or, less usefully but more impressively, recall every second of a mid-game sequence.

Just as lifting weights makes us stronger by building muscle mass, cognitive exercise enables us to make more sophisticated representations by expanding or rewiring our brains: taxi drivers who passed London's famously difficult navigation test grew their hippocampi (the brain region responsible for spatial processing). For violinists, the corresponding skills might include recognizing music more quickly from a sheet, predicting what the next notes should sound like, and knowing more precisely where your fingers should move. By viewing expertise as an outcome of better mental representations, we can see why deliberate practice is necessary: you need rapid feedback to know if your representations are right or wrong; you need to challenge yourself to make them better; you need goals to guide where you should progress next; and an expert coach who already has more advanced representations can help you with all that.

What about "innate talent"?

The intent of this book may be to provide a more nuanced, rigorous understanding of a commonly discussed thesis, but that battle plan falls apart as soon as Ericsson meets the enemy of “innate talent.” Ericsson is so into the practice-beats-talent idea that he spends a lot of the book setting up and knocking down strawmen of naturally gifted performers. Usually these involve Mozart:

Prodigal performance at a young age is often chalked up to natural talent, such as when Mozart became a performance-caliber pianist at six years old. But he also practiced intensively from a very early age (his father apparently wrote one of the first books about teaching music to young children), and today you can find a hundred six-year-olds on YouTube who can play at a similar level after lots of practice, so it's unlikely that they all just have the elusive piano gene too.
Some experts are said to have been great from the very first try, such as Mozart writing his first symphony when he was eight, but these claims are usually overstated. Ericsson says those early symphonies weren't actually very good, and they were in the handwriting of his father (a trained musician himself), who claimed he was just "cleaning up" the younger Mozart's work, but come on.
Even seemingly "savant-like" abilities are usually developed by training rather than natural gifts. Mozart may have been the first person ever to demonstrate "perfect pitch" (it was written about as a novelty at the time, and I can't find anyone earlier cited as having it). But we now know that this skill can be trained, and that the people who have it coincidentally also have a lot of musical training and/or speak tonal languages where the pitch is critical for meaning.

If you don’t already buy into the practice-matters-more-than-talent theory, all this circumstantial evidence probably isn't going to convince you. I generally do buy into it, so whatever. But even if I’m convinced that innate talent is not sufficient to be great at something, however, I still wonder if it's necessary. Can anyone become great in most fields with enough practice and opportunity, or are there inherent limiting constraints?

You won’t get a satisfying answer here. Ericsson mostly just looks at the IQ scores of people who are already experts in cognitive fields, such as scientific research and chess, and shows that among those experts, there's no correlation between IQ and success among that group. A small problem with this is that IQ scores are a pretty rough proxy to stand in for every source of natural talent that might be relevant here. (For example, one study finds that, while "intelligence" has a minimal correlation with chess performance, all "inherited talents" combined may explain up to nearly half.)

The bigger problem is Berkson’s Paradox: by only looking at elite performers, you may just be selecting out everyone who doesn't have enough natural talent in the first place. For example, if you only look at NBA players, there's no correlation between height and basketball performance -- but being taller massively increases your odds of getting to the NBA in the first place.* So in basketball, it's clear that many (short) people have almost no chance of becoming great no matter how much they practice. How many other fields are like that in ways that are less visible? Ericsson makes a big deal out of how great scientists like Richard Fenman and William Shockley had IQs "only" in the 120s, so raw intelligence isn't that important. But if becoming a great scientist requires a 120 IQ (the top ~10% of the population), that sounds like a big deal!

*Some of the numbers at that source, such as the claim that one in seven 7-footers plays in the NBA, are probably overstated because basketball players inflate their heights, but the pattern stands.

Fine, Ericsson says, maybe there’s a threshold of raw talent required in some fields, but that could just be due to opportunity. Maybe IQ doesn't matter for actually doing research if you practice enough, but you need a high IQ to get a good GRE score, and you need a good GRE score to get into a Ph.D. program, and you need a Ph.D. to be a scientist.

Mostly, however, he wants to get away from that discussion as quickly as possible, because he thinks any belief in natural talent has bad effects: if you try piano once and you aren't Mozart right away, you might give up. Or perhaps more importantly, if you're a piano teacher with 20 students and 19 of them aren't Mozart right away, you shouldn't nudge them to go play Roblox instead and stop wasting your time. Maybe this is correct, and we're better off just believing that natural talent doesn't matter! Still, it's disappointing to not get any rael insight here.

What can the rest of us do?

Chess grandmasters, London taxicab drivers, perfect-pitch studies – if this sounds like a best-hits-of-cognitive-psych-101 course to you, you’re not alone. But even though I know all the big ideas and examples here, I still found reading this book in detail and writing about it valuable. (That's not unique to this topic; good life advice should be obvious, and you need to read it over and over to adopt it.) In particular, it was very motivating -- now I really want to go deliberate-practice my ass off and become great at something.

But ... what? The window for me to become a world-class violinist is probably closed, if not literally (there's no physical constraint for musical performance like there is in sports or dance, but learning new skills is slightly harder as we age), then practically (from a lifestyle and responsibility standpoint). I can still use the concepts to pick up a new hobby and get better, which is great, but what I'd really like to do is use them to get better at something useful to my career; let's say, "become an expert writer." But the key conditions of deliberate practice -- objective performance metrics, well-established training techniques, and experts who have done perfected them already and are available to coach you -- don't hold here.

One answer is to focus instead on skills that do fit those criteria: for example, improving my typing speed. That sounds trivial, but there's a good argument that typing is the most important skill for writers: the faster you can get words from your head onto the screen, the more time you have for thinking, experimenting, or finishing your work and moving on to what's next. This may even be the right answer, and if I'm not too lazy, I should actually do this.

But what about all the other fundamental-but-less-quantifiable skills I need? Some of them could be shoehorned into a similar box. Ericsson writes about how Ben Franklin devised a training program to become a better writer: he found a source he thought was very good, and then tried to replicate it after reading through it once, first by trying to replicate the exact words in each paragraph (improving his vocabulary and grammar) and then by trying to replicate the outline of the full piece (improving his argumentation). But it's a little bit hard to imagine extending the same logic to higher-level tasks -- making decisions about what to write about, running a team or organization, breaking into a new field that nobody has done before. And even if I could figure it out, coming up with these programs requires a lot of effort, and then doing them by definition requires a lot of effort, on top of the effort required to actually do my job in the first place.

Remember how doctors don’t get better at doctoring through repetition alone? There's at least one area that’s an exception: doctors who do a specific surgical procedure more frequently have better outcomes at it. How do we reconcile this with the general ineffectiveness of unfocused practice? Ericsson argues that doing surgery actually checks most of the boxes of deliberate practice: operators are (hopefully!) really focused on their work; they actually get immediate feedback (because when they mess up there tends to be a lot of blood); and although they (hopefully!) aren't seeking out brand-new challenges intentionally, some arise naturally when unexpected complications arise. So just doing their jobs under these conditions builds expertise.

Training programs could be designed to accelerate the skill development process: doctors could be given more practice opportunities outside of a live-patient setting before they go into the operating room; these opportunities could be made more difficult to expand doctors' mental representations faster; experienced coaches could be brought in to help guide them. Some of this, I understand, already happens in medical school, but it's not really continued throughout doctors' careers. A more vivid example is the Navy's fighter pilot training program popularized in Top Gun, which was instituted because "learning by doing" isn't so effective when one mistake can get you killed.

The stakes aren't quite so high in most of our jobs, making it hard for employers or employees to invest so much time in setting up and going through rigorous practice programs. The best advice I’ve seen comes from psychologists Peter Fadde and Gary Klein (drawing on other military training programs), who suggest focusing not on deliberate practice but deliberate performance. This involves doing your day-to-day responsibilities -- that's the "performance" part -- but approaching them in a way that gives them as many of the characteristics of deliberate practice as possible: repetition, immediate feedback, incremental challenge, and mentorship, all of which help you improve your mental models (in their language, "tacit knowledge"). Basically, try to make your job as much like Top Gun as possible.

They give four alliterative examples of what this looks like:

Estimation -- before starting in on any assignment, explicitly predict how long it will take (or, even better, break it into sub-tasks and predict each of those). This is perfect for learning because you figure out on your own, within the timeframe of the project, whether you were right or wrong. Not only is accurately estimating timelines a very useful skill on its own, but by noticing what you do get wrong and figuring out why, you'll improve your broader understanding of what you're doing. You can increase your opportunities even further by applying this practice to your coworkers' projects (although I recommend not talking too much about this after you correctly predict they'll miss their deadline).
Experimentation -- instead of always doing the same task the same way, try a different approach and see what happens. This one is for me the hardest one to stick to, but even if the experiment "fails" it can be worthwhile because you'll see how a certain cause has new effects.
Extrapolation -- talking about coworkers' successes and failures, so you can learn from what they have learned as well. This could also include playing out counterfactual scenarios like "what would have happened if we did X instead" or "how could this have gone even worse".
Explanation -- when you observe something happen, try to figure out why it happened the way it did. This works best with a little bit of coaching from someone more experienced: first try to come up with the reasoning yourself, and then share it with someone who has more context or has seen similar things happen more, and learn from where their explanation differs from yours.

If you're performing reasonably well to begin with, your responsibilities will change and grow over time, naturally increasing the level of challenge in these tasks (estimating the timeline of an individual assignment is easier than estimating the timeline of a long, multi-team initiative) and giving you new goals to shoot for in your deliberate performance. Your movie may not be quite as sexy as Top Gun, but you'll be getting better at what you do.