[gall]

My best bursts of rapid learning are almost always project/puzzle driven. I didn't, for instance, set out to master FFT directly, but it seemed like something that could improve my abysmal performance on a Project Euler that I was working on, so I looked into it. I question (open-mindedly, not snarkily) the efficiency of ploughing through a course or series of courses. On the one hand there's the possibility of cross-pollination that having all sorts of cool bits of knowledge and techniques fraternizing in one's head for as long as possible promotes. On the other hand, there's the sense that the most efficient learning sequence is the one that matches the actual sequence of problems as they present themselves. Just-in-time learning of helicoptering, but if and only if you find yourself in a rooftop gunfight, as it were. Of course, then the issue becomes predicting forthcoming problems with enough lead time to learn the solution.