That, I think, is the hardest thing about managing programmers, in which I include the increasingly important long-term job of managing one's own career. There are very few macroscopic indicators of what work is actually interesting (and more importantly, what projects fit which individuals).
If you asked me, at age 19 when I knew very little about software, whether it'd be more interesting to work on a game or work on an OS or a compiler, I'd have said the game based on the project's macroscopic attractiveness. Of course, most experienced programmers would prefer to work on the OS. If you're trying to hammer some C++ legacy garbage from another project into usefulness, the fact that it's "for a game" doesn't mean very much.
What I actually find is that there's only a very slight correlation between macroscopic and day-to-day "interestingness" in work (which is why it's so hard for managers to tell which projects are interesting). More important is the work environment-- whether you have the autonomy and creative control to solve problems in a decent way and make the work interesting.