I think the tendency to "underspecialize" is universal and not limited to the extremely intelligent. Most people, left to their own devices, would prefer to be

I think the tendency to "underspecialize" is universal and not limited to the extremely intelligent. Most people, left to their own devices, would prefer to be a dabbler for a while, until choosing later on an area in which to specialize. (Then, the question is: in that field, will they specialize? There are general mathematicians who tend not to publish cutting-edge results but know the field very well, and then there are those who identify solely as algebraic topologists.) However, my suspicion is that people who hyperspecialize are (and always were) most successful in their time, while the more general-minded will be the ones who are remembered in 100 years. Eratosthenes had the derisive nickname "beta" because he was second-best in a lot of fields, but he's the first (as far as we know) to have proved the round earth.