<DIV><B><I><andru@cs.cornell.edu></I></B> wrote:</DIV>
<DIV> </DIV>
<DIV>> I would think that if votes are sincere, the best voting method would not be<BR>> Condorcet at all. It would be for each voter to assign a number of points to<BR>> each candidate representing the utility they ascribe to that candidate. The<BR>> candidate with the largest total utility would win. This method has obvious<BR>> advantages over Condorcet: it is simpler and takes advantage of more<BR>> information from the voters. Of course, it is trivially vulnerable to insincere<BR>> voters.<BR></DIV>
<DIV>Yes I agree, ratings would be sincere. I think the "good performance" of ranking based methods can be explained by saying that they have made the right choice on what user preferences we can take into account and still get quite sincere votes and results. Rankings are thus relatively informative but still relatively safe against strategies.</DIV>
<DIV> </DIV>
<DIV>The cardinal pairwise method of James Green-Armytage is an interesting study where he expands the area of voter preferences that can be taken into account by applying the ratings when solving a cycle of preferences. The borderline between what can be taken into account and what not is thus not quite stable yet. Of cource someone may also claim that already rankings are too much and we should satisfy with something less like Approval type of voting.</DIV>
<DIV> </DIV>
<DIV>Best Regards,</DIV>
<DIV>Juho</DIV><p>Send instant messages to your online friends http://uk.messenger.yahoo.com