"42 years ago today, Louis Armstrong and Buzz Lightyear opened the capsule door and went for a walk." ...And just stepping onto the Moon, Louis Armstrong started playing the saxophone and singing "Let my people go - that's one small step for a man":)

IMHO, the best you can do is to give this guy *really* hard, almost impossible tasks, and see if he copes with. Got some part of app that slows it down and everyone else failed to speed it up? Well, ask this guy to look at it! Got strange and mysterious bug? Well, the work for smart guy! There are two alternatives: if he doesn't succeed - then he's not that smart, and this will lower his self-esteem (and yes, he WILL write i18n stuff, or will be fired:) But if he does - then everyone will be happy: smart guy will have interesting things to dig into, others will have a person who can save your day in hopeless situations.

It was a kind of exaggeration about cleaning:) Really, I just mean that effective development supposes separation of duties. For instance, I'm a Sr.Dev/Architect for the project. I've got 2 Jr.Devs in my group. And I know that they can create a nice form in designer, write basic event-handlers/validators, test-cases, so on. And I have ENOUGH other work like creating business-logic, more complex parts of code, profiling/optimizations, etc. When some part is trivial, I may tell one of Jrs: "look here, could you finish this? If you have any questions or need help - just call me!" - and then switch to harder and not that obvious things.

Your smart guy is completely right about his position, but was completely wrong when deciding to get a job in your company. If you ask well-qualified programmer to clean a room or bring a coffee to his manager - first, you offend him. Second, you can't expect he'll clean the room as accurate as a cleaner:) Furthermore, his possible actions, in return, will include a kind of italian strike (in your case - he writes stupid cumbersome code) and looking for another occupation (that's smarter). Asking *really* good programmer to do things like localization is just inefficient. Very inefficient. So, the problem not in the smart guy, but in dumb project managers.