Forwarded message: This letter to EM was returned because I'd addressed it to "election-methdos". So I'm forwarding it to the list. Sorry about the resulting ">" characters in the left margin. > > Charles said that one goal of the vote on standards would > be to show whether or not the various count rules really give > different results in a real election, rather than just theoretically. > > But it's known that they'll give different results in real elections. > For one thing, however, if they give different results in this > election, that will especially mean something, since it will > suggest that they _often_ give different results. Additionally, > if your favorite alternative is on the wrong end of a method's > failure, then its failure will be especially well dramatized. > But yes, it's useful to conduct an election, to show the different > count rules in action, maybe demonstrating their failures, maybe > not. If it doesn't bring out their failures, that merely means > that the ones subject to failure didn't fail _this_ time. > > Let's hope that it doesn't take failures in public elections > to convince people of IRO's inadequacy. Sometimes it seems as > if CVD is trying to embarass itself in that way. > > Maybe it's premature to do a count of the rankings, but > the results so far do seem to favor majority rule & sincerity. > > Though those 2 standards can be interpreted different ways, > something can be said about those ways: > > Majority rule: > > Don's interpretation of majority defines majority as a > majority between a reduced set of the alternatives, a set thkat > has been reduced according to his method's rules. If his method > dumps candidates so that there are only 2 left, and a majority > prefer one remaining one to the other, then he calls that > compliance with majority rule if that candidate then wins. > Though anyone can define any standard they want to, please be > advised that a definition of majority in terms of the method > that one proposes differs from what the word means to most > people. > > Blake says that he prefers to consider a majority to mean > a majority of the people who have indicated preference between > a particular pair. Fine. Any method meeting the Condorcet Criterion > won't avoidably violate the principle that if A has such a > majority over B, then if we choose A or B, it should be A. > Condorcet Criterion methods honor Blakes majority. > > But when Blake proposes that differences are important, be > advised that he's now talking about something different from > a majority principle. I'm not saying that differences, as a > principle or standard in its own right isn't valid. Only > that no Condorcet Criterion method violates Blake's majority > principle. > > The majority principle that I've suggested for rank elections > with more than 2 alternatives, however, is violated by Margins > and IRO. If the majority of the participants in a multi-alternative > choice indicate that they'd rather have A than B, then if we > chose one of those 2 it should be A. Blake said he disagreed > wit that, but it probably is as obvious to most as it is to me. > > The example Blake gave to show a reason for disagreeing > contained the implied assumption that margins are more > important. And if margins are important to some because > they can be used to try to determine which pairwise propositions > are more likely to be "true" (A really is better than B), then > that must be considered a whole different standard, apart from > majority rule. > > More subsequently. > > Mike > > >
