Actually, Paul understood that very well. If you recall, his original statement was "this is why it's so hard to explain" to non-specialists.
I have something of a philosophical problem with methods that "count" the results expressed as the pair-wise matrix, since the "winner" depends upon which of the Condorcet methods is used to resolve cycles. This example bothers me, too: 1992: 45% Bush > Perot 10% Perot 45% Clinton > Perot Bush: - 45 45 Perot 55 - 55 Clinton: 45 45 - Perot is the clear Condorcet winner, but that cannot be the right result. If you replace those names with A, B, C the result looks ok. But that's not a logical argument. More serious is the "transparency" of a method. Especially with the current controversies about how votes are counted, I think it is critical to be able to map "what is counted" directly back to "what the voters put on their ballots", and since the linear translation from from the #voters by #candidates ranked ballots matrix to the #candidates x #candidates pair-wise matrix is intransitive I don't think you can convince the voters to accept that their vote was counted. (Even if it is). >From the VOTERS perspective, if a majority of the voters are going to be unhappy with the outcome, the purity of a Condorcet method that makes them unhappy is indistinguishable from the unhappiness they get from Plurality. As to getting some empirical data, I think someone mentioned earlier that if we could get ANY ranked-ballot approach approved, we'd have the raw data to analyze different methods with. Until then, there's only the proprietary data the pollsters collect. > -----Original Message----- > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] ] On Behalf Of [EMAIL PROTECTED] > Sent: Sunday, October 10, 2004 7:52 PM > To: [EMAIL PROTECTED] > Subject: [EM] Reply to Paul Kislanko > > In a message dated 10/6/04, [EMAIL PROTECTED] writes: > > > You may not take it that Paul has conceded anything since nobody's > > ever answered the original question. > > > > 5 of 9 voters voted C>A. > > > > Paul's question is how can anyone justify A's win. > > > > No one has addressed that. Until they do, ad hominems are just an > > example of how unlikely it will be to convince voters to change > > election methods. >> > > But in the example you cite, 7 of 9 voters voted B>C, so how > could anyone justify C's win? And 6 out of 9 voters voted > A>B, so how could anyone justify B's win? > > The experts all agree that there is no perfect voting method > that will satisfy everyone in every conceivable case, so the > goal must be to find the method that will result in the most > satisfaction overall compared with other methods. > > One question this list doesn't address very much is how often > the kinds of cycles that bother you (and everyone else) would > occur in actual voting situations. It's an empirical question > for which there is now very little data, because Condorcet > voting has rarely if ever been used in any elections of > public officials, and it has been used only slightly less > rarely in other kinds of elections (e.g., in elections held > by nongovernmental organizations). > > Advocates of instant runoff voting, which has been used in > enough public elections to provide some useful data, argue in > response to criticisms of it that there have been few if any > instances where the theoretical problems it poses have > actually been a factor in elections. > > It may well be that if Condorcet voting were used in a > variety of public elections over an extended period, over 99% > of cases if not 100% would have true Condorcet winners and no cycles. > That is, in each case the winning candidate would be prefered > over every other candidate if matched one to one. If that > were the result, then I suspect Condorcet would be widely > preferred over other methods and you would not have any > problems with it. > > But unfortunately, the empirical data needed to fairly > evaluate different voting methods in real world elections > just doesn't exist right now. That's a problem I wish the > participants on this list would devote more attention to. Has > anyone proposed any promising ways to test different methods > empirically? Has any such testing been done, and if so, what > have been the results? > > -Ralph Suter > ---- > Election-methods mailing list - see http://electorama.com/em > for list info > ---- Election-methods mailing list - see http://electorama.com/em for list info
