Dear Forest, Some replies follow, on the subject of cardinal weighted pairwise (CWP) and dyadic weighted pairwise (DWP)... > > >If all of the ratings go to the extremes of zero and 100, then the >ordinal >information is lost. Otherwise the strategic voter preserves the most >important ordinal information by squeezing all of the important >candidates >into ratings of 0, 1, 99, and 100.
Yes, of course I wouldn't want the ordinal information to be lost. Even if the "extreme ratings incentive" is amazingly strong (which I strongly doubt), the A>B>C>D>E>F (100, 99, 98, 2, 1, 0) type ratings that you describe should work just fine. If you're still not satisfied with this, then allow non-integer ratings (e.g. 100, 99.99, 99.98, 0.02, 0.01, 0) or maintain separate ordinal and cardinal ballots. One way or another, the ordinal information should be easy to preserve. > >It may not be a serious problem in most elections, but you might like my >suggestion anyway once you see it. I don't understand it yet, but I'm still working on it. I'll read through your proposal a couple more times, and then send another note if I still don't get it. my best, James ---- Election-methods mailing list - see http://electorama.com/em for list info