Wirth had (has?) a famous heuristic where he would reject any optimization of the compiler that didn't make it faster to compile itself. Since optimizations add

Wirth had (has?) a famous heuristic where he would reject any optimization of the compiler that didn't make it faster to compile itself. Since optimizations add code and thus cost something to compile, this is a clever way of testing cost/benefit.