Interestingly, the electoral college actually should help us here. States decide how they select their electors[1], and so if California says, "OH, we're going to select our electors with IRV, and direct them to vote based on the situation in the electoral college prioritizing candidates based on votes[2]", then that works. If we scrap the electoral college, then any change has to happen at the national level, or you're stuck trying to figure out what to do with a pile of IRV ballots from CA, a pile of range votes from NJ, and the FPTP votes from the rest of the country, and that seems just about intractable.

Actually, with races being close, I also wonder whether CA dividing its electors would get us more attention... The problem is that it favors whoever the majority doesn't like, in the short term.

[1] It must, now, be democratic; this wasn't always the case, though - many states originally had their legislatures choose.

[2] Obviously this language would be tightened up, but hopefully it's understandable.

Every known voting system that's deterministic and non-dictatorial is subject to strategic voting. This is fairly obvious, since a group's preferences can be non-transitive. For example, in a (very) hypothetical election with three voters who are also the three candidates, if A votes for B, B votes for C, and C votes for A (all based on their top preference), then clearly any one candidate can swing the vote to his or her second preference by voting for that instead.

Moreover, all voting systems have advantages and disadvantages, because many desirable traits of voting systems are mutually exclusive. Approval voting, for example, isn't even guaranteed to pick a candidate that is preferred by the majority of voters. Approval voting can also elect a candidate who would lose a head-to-head race against every other candidate. See http://en.wikipedia.org/wiki/Comparison_of_instant_runoff_vo....

Just to clarify, is your rebuttal essentially: "No system is perfect, so we shouldn't investigate better systems?"

I personally like approval the best primarily because it is extremely simple. All systems that meat the Condorcet criteria are too complicated IMO.

It also has the advantage over IRV that the votes can be tabulated at each polling place, rather than needing to be done centrally, which means both fewer changes to the current polling system as well as lower impact of fraud.

"Every known voting system that's deterministic and non-dictatorial is subject to strategic voting."

Maybe, but there's the matter of to what degree.

"Moreover, all voting systems have advantages and disadvantages, because many desirable traits of voting systems are mutually exclusive."

True, but that shouldn't stop us from optimizing as best we can.