David Epstein Knows Something About Almost Everything - Freakonomics
""EPSTEIN: I mean, talk about Einstein — he talked about this specifically, he wrote about his concern that as physicists and scientists in general became more specialized, journals would become more remote from one another. They become like these little intellectual archipelagos and you wouldn’t have people making the kind of connections that he and some of his most impactful colleagues were making across disciplines. There was way less journal specialization when he was saying that than there is now. If you look at the history of science, so many of the breakthroughs have been from people connecting knowledge that was available already — but tying it together creates a new frame. One of the great examples is Claude Shannon, who was an electrical engineer at the University of Michigan who was forced to take a philosophy course to fulfill a requirement. And in it, he learned about an almost century-old system of logic by which true and false statements could be coded with ones and zeros and solved like math problems. This had accomplished nothing in the 80 years since its creator passed away, except for getting in philosophy courses. And then he did an internship at a phone company and realized he could use the relay circuits like ones and zeros and code information into circuits. And that gave rise to binary code upon which all of our digital computers rely today. And as Shannon said, it just so happened that nobody was familiar with those two areas at the same time. And I think we lose a lot if we’re forcing people to stay in their own trench since so much of the innovation has come from connecting information across trenches...
EPSTEIN: Well, to give an example of that short-term, long-term trade-off, a study that came out too recently for me to get it into my book randomized different types of math learning. Some got what’s called blocked practice, where you get problem type A-A-A-A-A, then B-B-B, and so on. And the students make quick progress. The other classrooms got randomized to what’s called interleaved practice, where it’s like, as if you took all the problem types and threw them in a hat and drew them out at random. The students are more frustrated. But instead of learning how to execute procedures, they’re learning how to match strategies to a type of problem or the structure of the problem. And then when all the classrooms got the same test, where now they were facing problems that they had not exactly seen before so they had to transfer some of their knowledge, the interleave group blew the block practice group away. Studying the same stuff, just an order that slowed them down and made it more difficult and more frustrating but forced them to learn some more general skills — it slows learning down, but makes it a lot more effective. Frustration is not a sign that you aren’t learning, but ease definitely is. And the fluency of learning experience is a really bad indicator of how well you’re learning. And I think that’s just an incredibly hard thing to sell to parents and students and teachers...
LEVITT: What I took from your book was that there are a lot of forces, parental and societal, that push people towards being quite specialized in what they do. And your argument was that both from a societal and from a personal perspective that fighting against that tide and trying to push yourself to be more general for many or most people would be a good idea. Is that a fair assessment or do you think I’m off base with that?
EPSTEIN: No, I think that’s a fair assessment. When I was at Northwestern spending some time with a woman named Dedre Gentner who’s one of the world’s experts in analogical problem solving — so using analogies to solve problems, especially novel problems, things that maybe nobody has seen before. So, drawing on analogies becomes a useful problem solving engine and long-story short, she and her colleagues created this test that tests whether students are good at solving novel problems. And they would structure these problems in different ways. And what she basically found was that most of the students were good at solving problems that looked like things that were very familiar to them from their major. But then not good at solving even problems that were basically the same problem, but just with different window dressing of a different domain. Their problem solving skills were very narrow. They weren’t recognizing similar problems in other areas. The exception to that was these students in this program where they had no major, they just had a bunch of different minors. Essentially, they would dip a toe into each of the sciences and see how those areas examine problems, how they approach problems, what different tools they use. Those were the students who did the best at solving novel problems. But when I went around and talked to her colleagues, they would say, “Oh, we don’t like that program because those kids are getting behind.”...
LEVITT: So, in some ways, your first book, The Sports Gene was a 300-page debunking of the 10,000 hours argument, or at least an argument that the 10,000 hours mantra is way too simple. Is that an accurate reading?
EPSTEIN: Yeah, definitely. I think the most popular conception is that there’s no such thing as talent. It takes 10,000 hours of so-called deliberate practice — this is effortful, cognitively-engaged, error-correction-focused practice to become an expert in anything and this is based initially on the work most prominently of a man named Anders Ericsson. The original work was at a world-class music academy and it studied 30 violinists. So a small study to start with, but the 10 best violinists, those who the instructors deemed capable of being international soloists, had spent on average 10,000 hours in deliberate practice by the age of 20. And the next two groups down, lesser. The first thing that raised my eyebrow about that paper was that there was no measure of variance included. So my question was, “What was the range of this?” Because I had read other expertise research, like in chess, where it took 11,053 hours on average for people to reach international master status, which is one down from Grandmaster. But some people made it in 3,000 hours and others were being tracked at 25,000 and they still hadn’t made it. So you could have an 11,053 hours rule, but it didn’t really tell you much about the reality of skill acquisition. And so that was the first thing that made me skeptical. When I asked after that data, the responses I got were basically like, “People were inconsistent on their retrospective accounts of how much practice they did and we’ll have better measures of variants when we have video diaries that people can keep.” So the answer is that your data aren’t very good? Like lots of people face that problem, but they still have to include measures of variance.
And then as I started to get into more research about skill acquisition, what I was realizing was that it was the rate of learning. So there was something underlying the 10,000 hours rule called the monotonic benefits assumption that basically meant that every person at the same level of skill should progress the exact same amount for the same hour of deliberate practice. And that turns out to be true, essentially, nowhere, except in very simple perceptual motor skills... rates of improvement differed gigantically between individuals, which made seeking out the place where you improve and learn more quickly, something that I think is important to encourage people to do. Whereas, in the 10,000 hours thinking, that makes no sense. You should just pick the first thing that comes and stick with it... in most studies, the athletes who went on to become elite, actually had less deliberate practice early on in the sport in which they had become elite. More sort of free-form activity, wider variety of physical activity... [Gladwell] says something like, “Yeah, I think I made an error where I conflated the fact that a lot of practice is important to become great, which I think is true with the idea that that implies you need early specialization, which I now think is false.”"
