Michel Balinski and Rida Laraki have come up with a new voting scheme that they say will yield more accurate results in elections. It's called majority judgement. Instead of voting for one particular candidate, voters are asked to evaluate all the candidates on a scale from something like "excellent" to "reject". Here are the results of a trial that they ran in Orsay in France (scroll for English)...and you can see how the results differ from the official election results. To put this in a bit of US-centric context, imagine a voting system where honestly evaluating your feelings about the executive readiness of a third-party candidate like Ralph Nader doesn't necessarily harm a major party candidate's chances of getting elected. (*cough* Al Gore *cough* 2000 election.)
They're running an online majority judgement experiment using the 2008 US Presidential candidates...go sign up and vote. (thx, judy)