That's a... remarkably loose definition of corruption you've got going on there.

I'm not sure it's practical to make a political system completely free of incentives, as long as you're working with humans governing humans: the closest approximations I can think of would have to involve a leadership caste socially and economically isolated from the people they govern and without any means of improving their own welfare, and that's so far removed from anything historical I know about that I don't even want to try working out all its long-term implications. Imperial Japan's about the closest, and that degenerated at first into proxy governance by provincial warlords and later into a military-aristocratic dictatorship ruling in the imperial family's name but not in practice controlled by it.

Now, given anything resembling our existing politics, it seems naive to behave as if the default incentives surrounding political power are nonexistent or weak enough that they're drowned out by altruistic impulses among those inclined to seek power -- or even among random members of the populace, if you prefer direct democracy. This being the case, it makes far more sense to me to design systems to reward competent government -- however defined -- rather than to high-handedly dismiss any such attempts as unethical and rely wholly on the better angels of politicians' natures.

I'm not quite prepared to say that there can't exist any candidate systems where this wouldn't be necessary, but if you've got a proposal like that, we should really be talking about that proposal rather than speaking in generalities.

Doing X for specially crafted incentive Y for the intrinsic value of X is a form of corruption. It's not always possible that all decisions are made for the intrinsic value but if you have a political enviroment where there a lot pressure to do Y's.

Especially if you can't get any political power without Y, you won't have many people who persue political goals for their intrinsic value in your political system.

This seems like possibly quite a useful bit of abstraction and offer the potential of arguing the merits of a single principle that appears in many manifestations, in politics, corporations, volunteer organizations, etc. But I'm just having trouble getting it clearly in my head. Two things might help.

1) One or 2 concrete examples where you flesh out "X" and "Y". I spent 2 years in a math Ph.D. program, which is long enough to know that to move forward with an abstraction, it is best to start with at least a couple of examples.

2) Consider the "agency problem" (or "principle-agent problem"), which to me seems the most promising abstraction for reasoning about corruption. See http://en.wikipedia.org/wiki/Agency_problem, and maybe very close to what you're aiming at.