I actually had to fix a bug just last week in our production code that was caused by an arithmetic overflow, and I was only able to diagnose it because I knew 2s compliment. This stuff seems arcane and pointless. It's not. Over and over again in my career, I'm the only guy on the team who can solve some nasty little problem, and it's because my undergrad CS program taught us all the old-school theoretical goodness. Of course, they also completely left out useful things like OO, testing, or really anything related to software engineering. But these are all things you can learn on the job. You can't (or at least, won't) ever learn the mathematical underpinnings of this stuff at work. So you really have to do it in school.

touche