LINK: I read Chris Lawrence's suggestion for an American politics paper:
"Incidentally, there’s probably a good undergraduate or first-year graduate student paper in an analysis of the effects of various electoral college allocation systems (PR, bonus PR, congressional district, plurality), with particular focus on elections with relatively large third-party voting (1948, 1968, 1980, 1992, 1996, and 2000)."
And I had a bit all prepared about how this sort of thing is exactly the reason I'm not an Americanist, when it occurred to me that this would actually make for a rather interesting paper (even if the data collection would bore me to death). All that stuff my professors used to harp on (change the voting system and you can change the outcome! even with the exact same expressed preferences! it's just like Riker says!) apparently still piques my interest*.
Damn that Michigan education.
*All the quants at Duke are in International Relations (or so they tell me, and John Aldrich excepted), so this is a paper I'll probably never write, but you never know.