Regarding that quote about Rhodes Scholars--since no one has mentioned it yet, I will mention the phrase: "regression to the mean". If, from a diverse populati

Regarding that quote about Rhodes Scholars--since no one has mentioned it yet, I will mention the phrase: "regression to the mean". If, from a diverse population, you select a subpopulation with high scores on trait X measured by some noisy test, and you then re-test them sometime later (so that the noise is generated by fresh randomness the second time), you will likely find that they score worse. This is because it is likely that the subpopulation you selected mostly benefited from the noise, and being lucky the first time is unrelated to being lucky the second time. (That's one way to characterize luck.) The regression-to-the-mean effect is obviously proportional to the ratio of "noise" to "intrinsic qualities". If the amount of noise is close to zero, then the regression will be small; if the noise is gigantic or the intrinsic quality is zero, then the regression will be almost complete. To illustrate the second case, you might imagine a population rolling six-sided dice, and a silly researcher selecting all those who rolled sixes and being astonished that their average had dropped to 3.5 on the second roll. If you then imagine that each person had a private number from 0 to 50 that they added to their die rolls, then the test suddenly becomes much more meaningful. With that in mind, we can hope to be more than simple-mindedly impressed or simple-mindedly cynical about Rhodes Scholars, child prodigies, Thiel Fellowship recipients, and so on. How much "noise" is there, and what can be done about it? For example, suppose that they're looking for people who do amazing things. Suppose that most people who do amazing things do them because they're interested in them, and that there's a bunch of pseudo-randomness in what you happen to be interested in. Then... they'll pick a bunch of people who just happened to be interested in doing things that the scholarship people found amazing. What will these people be interested in doing next? there's a good chance it'll be completely not-amazing. (Yes, some people have a deep, established interest in some field that is unlikely to change--but even within a field or subfield, you will find interesting things that are amazing and interesting things that are not.) In that case, either they will look disappointing, or they will decide (probably due to feeling pressure) to try to work on amazing things even if they're not interesting. I doubt if it's impossible to pull off the second option, but it is (by stipulation) a significant mode change for these people, and some received wisdom[1][2] says it's not likely to go well. I hope that, instead, they are looking for people who have intense interests and great natural ability, and will try to shelter them from external pressure to produce "amazing" things, and that this press release is just crap they put out to please the media. Meanwhile, I guess there might also be successful people who thrive on that kind of pressure... I just identify most strongly with the first group. Also, it probably is a good thing to celebrate people's achievements. It just carries some danger with it... Anyway, I've described one type of amazing youth for which their apparent amazingness depends significantly on a capricious factor (what they're interested in). I'm confident about my qualitative description, because I know myself and some other people that I think are of this type. How significant is the interest factor? how capricious is it? can interest be deliberately controlled to some extent? Are there other types of amazing youth? is their current amazingness also subject to significant amounts of randomness? This, at least, would enable us to judge how much disappointingness we'd expect simply from the regression-to-the-mean principle. Do people do serious scientific investigation about this sort of thing? Seems like the Thiel Foundation had better, given that they're working in this exact field. But charity-style things probably don't have systematic incentives to perform well... maybe we can hope that Peter Thiel will apply the kind of ruthless determination to succeed that I feel he has [maybe they're in fact selecting youths who exhibit this trait, which certainly seems it'd be more deterministic than "interest"], and figure this stuff out. Maybe. [1] http://www.physics.ohio-state.edu/~kilcup/262/feynman.html http://www.physics.ohio-state.edu/~kilcup/262/feynman.html [2] http://www.paulgraham.com/hamming.html http://www.paulgraham.com/hamming.html [see: "When you are famous it is hard to work on small problems. This is what did Shannon in."]