Paul: You quoted me:
" In other words, the A and B voters are a majority, C is the favorite of the most, and A is a more popular favorite than B." and asked: So who should win based only on your analysis? C "the favorite of the most", or A who is not? [endquote] If you prefer Plurality, then of course you'll just elect C. My 33,32,34 example showed only the sincere rankings. Who should win depends on what the actual ballots look like. Say the method is Condorcet(wv), which what is usually meant by "Condorcet" nowadays. The sincere rankings were: 33: A>B 32: B>A 34: C The assumption of these examples is that the {A,B} voters greatly prefer A and B to C. They despise C. Their preference between A and B is considerably weaker. The A voters and the B voters add up to a majority. They have the power to win, if they can co-operate. So the co-operative, responsible and trusting A voters rank B in 2nd place. But the B voters refuse to rank A. Actual ballots: 33: A>B 32: B 34: C B wins. The B voters have taken advantage of the co-operativeness of the A voters. The A voters wanted to help make the {A,B} majority win. But They were had by the B voters. I'm not trying to make that sound worse than it is. As I said, though Approval has that problem too, it can be dealt with in any of 5 ways that I've listed. But it definitely involves strategy. My point is that Condorcet has that problem too, as shown above. So, don't let anyone tell you that Approval uses strategy, but Condorcet doesn't need strategy. The strategy-free-ness of Condorcet is a myth. But myths that we so want to believe in are persistent. Condorcetists are in denial. They want to believe that Condorcet is the solution, and so they are in denial about its problems. You asked who should win. Anyone but B. It's ok if C wins. If it's known that the above-listed rankings will elect C, then the A voters can safely rank B In 2nd place, knowing that the B voters can't gain by defection. Result: The A voters and the B voters will rank eachother's candidates, and the {A,B} majority will win. A will win. The A faction is larger than the B faction. But the important thing is that {A,B} wins. A majority votes as a mutual majority, and, through co-operation, wins. No one has the co-operation/defection dilemma, the chicken dilemma, that they'd have in Condorcet. ICT elects C with the above-listed ballots. ICT achieves defection-resistance by ignoring the A voters' support for anyone other than their favorite, when choosing a winner from among IC-unbeaten candidates. There are two methods, MMPO and MDDTR that choose A with those ballots. I like that even better. But those methods have problems. For one thing, there are some unfair criticisms that can and would be used against them, worsening their acceptability with the public. Additionally, they're both more vulnerable to successful burial strategy. So I'd prefer ICT to them. But I've told why no rank method is sufficiently winnable, and why the elegantly simple Approval method is the feasible, do-able proposal. You continued: Language is tricky. If C is the favorite of the most, how can A or B have a majority? [endquote] I didn't say that A _or_ B has a majority. I said that the A voters _plus_ the B voters add up to at least a majority. Mike Ossipoff ---- Election-Methods mailing list - see http://electorama.com/em for list info