On Thu, 16 Jan 2003, Tom Ruen wrote: > I judge that Bucklin has pretty much the same strengths and weaknesses as > MCA. Ranking versus Rating is the difference. > > The option of allowing equal rankings is nice, although not clear that it is > in the best interest of anyone to do so. > > Bucklin might be considered more equal between voters since all voters have > the same influence (if they don't truncate their preferences). > > One variation on MCA might be to actually divide one's vote among the > choices offered at a given level, if more than one. > > If I vote: > A - Preferred > B - Preferred > C - Acceptable > D - Unacceptable > > My first round vote could be: A=1/2, B=1/2 > And if a second round is needed: A=2/3, B=2/3, C=2/3 > > Then every voter has the same influence. I'm not concluding firmly whether > or not this is an improvement, but it then would be the mathematically > similar to Bucklin, since Bucklin gives voters one vote the first round, two > votes the second round, etc. (Bucklin could also allow tied rankings and > split votes) > > The advantage of this fixed-vote-value counting is that if a single majority > candidate existed in the first round, you could be sure that candidate was > preferred by a majority, while MCA counting can't promise that in round 1. >
Actually, my first proposal of (what was later dubbed MCA by Joe Weinstein) intended for the favored position to be counted cumulatively, i.e. if three candidates were marked as favored, then they would each get one third vote in the majority favorite stage of the procedure. As you have remarked, this would preclude multiple candidates getting a majority. But extending cumulative rules down to the second level would give strong incentive for insincere order on the ballot, since compromise would have diluted strength if other candidates were marked as acceptable. In fact, the temptation would be to vote compromise as favorite and truncate the rest of the candidates (including favorite) so that compromise would have the best possible chance in the second round. In this regard, see in the EM archive messages 9628 and 9692 for discussions of this issue. The original proposal can be found at the end of message 9571 (April 27, 2002). I'm willing to go with the version that appeals most to the public, since the original version comes close to satisfying the (strong) FBC, much closer than IRV does. In fact, suppose that your favorite candidate is A, and that there are two front runners B and C, of which your preferred is B, so that your sincere preferences (restricted to these three candidates) look like A>B>C. What would be the worst that could happen if you voted your MCA ballot sincerely (under the cumulative rules for preferred status and approval rules for acceptable status)? There are realistically only two bad possibilities: (1) Candidate C wins with more than 50 percent of the favored status votes. (2) No candidate wins in the "majority choice" round, but candidate C wins in the approval round. In case (1) it wouldn't have helped to put B ahead of A on your ballot, because only one candidate can have a 50 plus percent majority in the cumulative stage, and (in the case under consideration) candidate C is that candidate. In case (2) it wouldn't have helped to put B ahead of A on your ballot, because in the second round, favored status and merely approved status are not distinguished. So if this method does not satisfy the strong FBC, it is because of some other much less likely scenario that we haven't considered here. This may well be the form that would appeal the most to the public because of strong feelings about the meaning of "one person one vote" and "majority rule." I believe most folks want a method that (1) allows them to give strictly greater support to favorite A than to compromise B without penalty (such as wasted vote), (2) satisfies one person one vote, (3) satisfies the majority criterion, and (4) is as simple as possible consistent with the other three requirements. This method would beat IRV on all four counts. In particular, with regard to (3), an IRV majority is more likely to be a result of insincere ballots than a majority under a method like this version of MCA that (at least nearly) satisfies the strong FBC. The only simpler method that I know of that more or less satisfies (1), (2), and (3) is Candidate Proxy: you vote for one candidate, who then represents you as your proxy in an election completion convention where an election completion procedure is carried out by the proxies (candidates) on behalf of the voters. How well this method satisfies the strong FBC depends on the choice of the election completion procedure. And obviously, the voters lose some power because of their vicarious status in the election completion convention. Of course, as soon as the winner takes over as your representative, your control takes a drastic dive. At least during the election completion convention you have true proportional representation. Forest ---- For more information about this list (subscribe, unsubscribe, FAQ, etc), please see http://www.eskimo.com/~robla/em