1) regehr isn't saying that he doesn't like two's complement, he's saying that he doesn't like two's complement wraparound. He's saying that it would probably m

1) regehr isn't saying that he doesn't like two's complement, he's saying that he doesn't like two's complement wraparound. He's saying that it would probably make sense for integers to operate linearly or fail if they get to a point where they will become non-linear. 2) Arguing to keep a local optimum seems silly to me. Why not reach for the global optimum? 3) He's not arguing for a change in the order of evaluation, he's identifying a situation that would work with integers that operate linearly and showing that it doesn't work when integers behave non-linearly (ie: around INT_MAX and INT_MIN).