At 03:55 AM 8/24/2007, Jobst Heitzig wrote: >[I wrote:] > > > There are, I believe, ways to improve the performance of Range, and, > > as it happens, the one I've been proposing also makes Range MC > > compliant in the overall method, including a possible runoff. > > Obviously, Range *cannot* be MC compliant directly, for it can pass > > over the favorite of a majority, when this is only by a relatively > > small preference strength, to elect a stronger preference of a > > minority. > >Range *is* a majoritarian method since a majority can elect whomever >it wants by bullet voting.
That does not contradict what I wrote. Being a "majoritarian" method does not make the method Majority Criterion compliant. Absolutely, I have argued that Approval satisfies the majority criterion as written. If the majority prefers a candidate over all others, and strictly indicates that, it must prevail in Approval. However, I only got a little support from others in this contention. The idea of this unexpressed preference is very strong, even though nobody that I've seen writes a definition of the Majority Criterion that explicitly makes it clear that unexpressed preferences are not moot. But Range is a little different. There, too, I have argued that we must take votes as written, that the concept of "sincere" Range vote is problematic. If I prefer Bush to Adolf Hitler, and somehow Hitler ends up on the ballot (why should that matter? -- after all, I'm free to write in anyone!), must I rate Bush nonzero to vote "sincerely", assuming I do, in fact, prefer Bush to Hitler by a proportional amount (considering the other candidates) to cause me to notch up my vote? Many writers gloss over the problem. The argument that was made by those opposed to what I was saying was that the majority could not freely express its preference without incurring some loss. Yet, generally, it is considered that Plurality satisfies the Majority Criterion, even though it is true with that method, as well, that expressing a strict preference for your favorite can mean your vote is wasted. What is clear is that all of these methods do guarantee a majority that knows it is a majority can prevail, regardless of preference strength, if they choose to do so. Thus the original problem is, in a single-step method, almost certainly insoluble, unless votes are bids and thus presumably connected to real preference strength. If the majority knows that it is the majority, and it doesn't mind giving up the extra public funding from the minority with a strong preference (and thus reduced taxes for themselves), then it can choose to let its preference go. However, this is a trick: what has happened is that the preference of the majority has been shifted by expected consequences of the election. Anything could cause that shift. *It is no longer the preference of a majority.* However, most of us are concerned about the zero-knowledge, or, as well, the only-approximate-knowledge case. I never claimed that Range satisfied the Majority Criterion, because the method so clearly invites intermediate ratings, and if a majority does not know it is a majority, it may hedge its bets, and thus the majority favorite can lose. Note that in a single step method, this violation of MC is *essential*. MC is actually a poor criterion to judge elections by, *but* it is related to a very important principle of democracy: majority rule. Majority rule in aggregative systems is oppressive, which is why few seriously propose pure aggregative, direct democracy. However, in the context of full deliberative context, it is crucial, for, in fact, the alternative to majority rule is not supermajority rule or consensus, it is minority rule, where the status quo favors the minority. Majority rule is the foundation of deliberative democracy. And, of course, a wise majority knows that to use its power thoughtlessly could be very dangerous. Majority rule *approximates* social utility maximization, but fails badly under some circumstances. And if the majority makes decisions by a slim margin, based on small preference, contrary to the strong preference of a minority, it harms society overall, and, as they say, what goes around comes around. The Republicans in the last Congress were somewhat restrained in using their majority power as a blunt instrument. The majority has rule-interpretation power (the House and Senate are, in fact, democracies of their members, though the systems can be somewhat defective, they retain most standard deliberative procedures), and so it can essentially do what it wants, by interpreting the rules as they want to interpret them. But it would set a precedent and could easily come back to bite them, and the wiser Republicans knew this. > > Given that I don't "believe" in measurable utility, am I an "other"? > >Yes, and I ask you to understand the given example in the second way. However, I'm quite happy *using* a kind of "measurable" utility, which is in simulations, where we know the quantified utilities as initial assumptions. Assuming such utilities as was done in the example provided, and then trying to draw overall conclusions from such, is hazardous, because a method may be vulnerable or display some bizarre results when encountering a particular pattern, but if that pattern is impossibly rare, it is moot. But if reasonable preference distributions are generated randomly, and then used to determine voting patterns assuming various strategies, the utility is known *as an assumption* and then we can study how voters would vote and how methods would respond. And all of this would obviously be a fruitful field for research, but in theory with simulations and in the field, studying actual voting patterns as compared with, for example, range poll results and even possibly such results. Note that it would be possible to study, for example, a Clarke tax with some subset of voters who voluntarily participate. I'll leave the details of how this could be done to others, but it would be one way to get, in real elections, sincere votes. ---- Election-Methods mailing list - see http://electorama.com/em for list info