A while ago I tried to make a case for solving domain-specific problems, rather than general ones. My basic view is that engineering is a limited resource, so providing an overly general solution takes away from making the useful specific one more awesome (or quicker/cheaper to develop).
