Range goes into this. Epstein talks about Kind and Unkind learning environments.

In Kind environments, the feedback is quick and ranking is easy to know. So the evidence says that the optimal strategy is drill ans kill.

In Unkind learning environments, the feedback is slow and ranking is difficult and untimely. So the optimal strategy is to learn as much as you can in as many very different disciplines as possible.

The paper that the Economist talks about extends this and (paraphrasing) says that the very top elite level, even Kind learning environments turn back into Unkind ones again as you try to push the field more.