The more realistic three-way Condorcet race from 2000 would have been between Bush, McCain and Gore, like so:
Bush>McCain: 49,000,000 McCain>Bush>Gore: 1,000,000 McCain>Gore>Bush: 500,000 Gore>McCain: 49,500,001
McCain might very well have won that contest based on second place strength, and I think there would have been a lot of grumbling, but you would have had a lot of Bush supporters saying "well, at least Gore didn't win", and Gore supporters saying "well, at least Bush didn't win". In essence, a compromise would have been entirely appropriate.
Of course, under IRV using these ballots, Gore wins by one vote. We all know he won by more than that.
Rob
James Gilmour wrote:
While I accept that IRV would reject some Condorcet winners, I have real concerns about the political acceptability of the Condorcet outcome in some (very real) circumstances and the effect of the voting system on electors' behaviour.
I have used the popular vote results of a real election, but I've reversed the two largest figures to make a point.
Bush 50,996,064 Gore 50,456,167 Nader 2,864,810
Please do not think this is an attack on the electoral college - that is a quite separate issue. I hope no one will be offended by my omission of all the other candidates from this exercise.
If this were a popular vote FPTP election, Bush would win. But if we assume Nader supporters are closer to Gore than to Bush, some would say we should use IRV so that the majority of voters could be represented.
Now we have to make assumptions (guesses) about how these same voters might vote under a different voting system.
IRV1 Bush 50,996,064 Gore 50,456,167 Nader>Gore 2,864,810
Nader is excluded and Gore wins. Majority of voters represented.
IRV2 Bush 50,996,064 Gore>Nader 50,456,167 Nader>Gore 2,864,810
As before, Nader is excluded and Gore wins. Majority of voters represented.
IRV3 Bush>Nader 50,996,064 Gore>Nader 50,456,167 Nader>Gore 2,864,810
As before, Nader is excluded and Gore wins. Majority of voters represented.
BUT Nader was everyone's second choice and so this is not a properly representative result.
Solution = Condorcet. Result = Nader wins. This makes good theoretical sense, which I fully understand, but I do not believe it would be politically acceptable to the electors of the USA.
Were Condorcet ever to be adopted, I would predict that you would NEVER see the voting pattern given at IRV3. Once the major parties saw the effect of the Condorcet system, their supporters would "bullet vote" only for the parties' candidates. So you would be back to the pattern at IRV1. The "other side" might win, but at least they wouldn't see their own chance denied almost automatically in favour of a minor minority candidate.
I think the real world would be very different from the simulations commonly used here to show the strengths and weaknesses of different voting systems. My conclusion would be that US electors would accept IRV but not Condorcet.
James
----
Election-methods mailing list - see http://electorama.com/em for list info
---- Election-methods mailing list - see http://electorama.com/em for list info
