A related question, concerning Hsu's point: How much can historical data be trusted to reflect current trends? During the 50's and 60's, physics (especially nuclear physics and the closely-related field of particle physics) were "hot topics" and everyone "wanted in"; it's not hard to imagine that the very best and brightest people entered the field during that time. But is that true anymore? It's possible that as the desirability of a physics career has decreased, the average level of intelligence in the physics community has also decreased, and that the reputation that physics has for being "the smartest of the smart" may just be a leftover from a previous era.