I had success with GP because, as a programmer, I found it straightforward to map the problem domain onto a set of operations. Mapping the problem domain, IMHO

I had success with GP because, as a programmer, I found it straightforward to map the problem domain onto a set of operations. Mapping the problem domain, IMHO, is the hardest part, and in all of my problems, it was hard to map it to differentiable functions. GP doesn't have that problem.