The other thing I wanted to mention about why Bucklin and MMPO might complement each other is that MMPO potentially makes more use of the information in the lower ranks than Bucklin (especially in a many level cardinal weighted pairwise version), while MMPO tends to encourage equal ranking at the top (in order to satisfy the FBC), which is a service that Bucklin can perform for MMPO by collapsing the top levels down to the median of the candidate ratings..----- Original Message -----From: Date: Saturday, April 24, 2010 11:17 amSubject: ParticipationTo: [email protected],> If I am not mistaken, both Bucklin and MMPO satisfy Perez' weak > version of> Participation: if the winner changes when a ballot is added, > then the old winner> was not ranked top on the added ballot.> > I wonder if some kind of hybrid between these two methods might > be better than> either without losing this form of Participation.> > For one thing, the pairwise oppositions in MMPO would need to be > replaced with> some kind of weighted pairwise opposition to ensure clone > independence. For> that we need two or more levels if not full blown cardinal > ratings. > > What if the three levels are (1) anything from the top rank down > to the level of> collapse that would be needed if Bucklin were used, (2) other > ranked, and (3)> truncated? Then if a ballot has x in level 1, and y in level > 3, that ballot> contributes two to the opposition of x against y, whereas if x > is only one level> above y, the ballot contributes only one to the opposition.> > Any other thoughts on this?
---- Election-Methods mailing list - see http://electorama.com/em for list info
