Stick to traditional agronomics [sic] and engineering degrees etc, and just build things that work. - that is why those people should do serious number crunching on things that look plausible, but thin their lists out first with the sort of approach I used.