He suggested assigning candidates a score based upon "number of bullet votes required to be the condorcet winner". (except, in his description, he suggested the beatpath winner would always have a score of 0, regardless of whether he is the condorcet winner). I noted that (aside from the exception for the beatpath winner), that this is simply the minmax (margins) score. (well, technically i suppose it would be off by one, since according to his wording, it would require the additional vote to actually beat, rather than just tie, all candidates. I don't think the difference is significant)
Now....what if we change it so that, instead of considering the number of bullet votes (a bullet vote could be defined, in terms of the pairwise matrix, as adding one to each cell in a candidate's row)...instead we consider the number of *pairwise* votes required to be the condorcet winner. In other words, the *sum* of all losing margins, rather than just the largest one.
What is this method called? it is still a condorcet method, and is still every bit as simple to calculate as minmax.
Does it have advantages over minmax (margins)? It seems like it might actually be a little better, since each score will be affected by a greater number of losing margins, rather than just the largest one.
-rob
---- election-methods mailing list - see http://electorama.com/em for list info
