[_delirium]

One answer I've gotten from a mathematician was more or less: I like solving mathematics problems, and working in a company means I get a never-ending stream of mathematics problems to solve, where the question "why are these problems worth answering?" is someone else's job to solve. There's something liberating (for at least some kinds of jobs) about being able to just solve problems "for their own sake", because it's someone else's responsibility to invent and justify the problems, and you can treat them as magically justified problems that fall from heaven. This probably works better for people who like focusing on specific aspects of problems (e.g. treating something as a pure mathematics or engineering problem), and less well for people who like more integrated/cross-cutting problem solving (which working for yourself both allows and demands).

I personally find the bureaucracy/procedures a bit stifling, though. The paycheck is nice, but meetings and standard software and performance reviews and whatever is a drag, even in relatively freedom-loving academia.