Martin wrote: The problem I have about Mike's criteria (and some others) is that they talk about reversal of preferences, and failure to vote preferences, and recently I've begun to doubt the usefulness of these ideas in general. I reply: If you're saying that "these ideas" aren't useful in general, then it would be good if you'd support your claim in general, instead of just specifically. Your specific objection is simply the result of your having missed definitions that were posted to EM many times. But I can't answer your general objection unless you say what it is. Martin continues: Specifically, if I define IRV(for eg) as "place marks on this piece of paper", then I can claim that the criteria doesn't apply to me, so nah-nah. "I'm not asking for the expression of preferences" I say, "You can read preferences into the ballots if you like, but that's just your *interpretation". I reply: Your "nah-nah" is premature. My definition of voting a preference for one candidate over another has been posted to EM a number of times. That definition has nothing to do with the matter of whether or not the method proponent says that he's asking for the expression of preferences, or how the ballot instruction is written. Here's the definition: A voter votes a preference for X over Y iff he votes X over Y A voter votes X over Y iff he votes in a way so that it's possible to contrive a configuration of the other voters such that if all the alternatives except for X & Y were deleted from the ballots, and the ballots recounted, then X would be the unique winner if & only if we count that voter's ballot. [end of definition] For specificness, let me add: A voter reverses a preference if, preferring X to Y, he votes Y over X. A voter falsifies a preference if he votes X over Y without preferring X to Y. Martin continues: I think this is a reasonable complaint. I reply: It would be a reasonable complaint if it were true. But, as you now see, voting a preference for X over Y has a definition that doesn't depend on the method proponent or the ballot instruction talking about expression of preferences. Martin continues: There are also problems with methods like Approval which don't allow the expression of all preferences - eg, Approval fails SDSC almost by default - before the votes are even counted, Approval fails to pass - that feels unfair, even if (as it turns out) it isn't. I reply: I'm sorry, but I don't know what that means. Approval fails SDSC because a nonrank method can't record enough pairwise preferences. But it's desirable to be able to defeat a "greater evil" without having to insincerely equal-rank more preferred candidates. A desirable guarantee is enough to justify a criterion. Martin continues: One workaround for this is to only define criteria on ranked ballots, and specifically define what a "sincere" vote is on such a ballot, along with what a "preference" is. That's a bit of a kludge, and it doesn't help us compare, eg, Approval or CR or methods which take Dyadic votes. That's right. Blake says that his Condorcet Criterion is useful because it can compare rank methods to eachother. But how useful is a criterion that can only apply within one class of methods, and can't compare methods of different classes? Your system of criteria that replace "voting X over Y" with statements or stipulations about the results of doing so are something that has been discussed some. It would be more elegant to talk directly about the results of defensive strategy than to just speak of voting X over Y. Of course the easy part is suggesting the desirability of such a system of criteria. Completely defining them, showing that they're universally applicable, and act in a way that won't embarrass us, that's more work. I hadn't gotten into that task, because I felt that the existing defensive strategy criteria were useful enough, describing guarantees that I'd like to have, and which measure for standards that are important to me. Your definition of naive voting needs work. But of course your proposed system of criteria would be interesting, and would be useful if they can be met. Do you know what methods meet them and which don't? But I don't think they're really defined until "naive voting" is better defined, or replaced. It really should be something that's the same for all methods, and it won't do to say "simplified 0-info voting", or the like. But I look forward to your further development of that system of criteria. When they're completely defined, and when we know what meets them and what doesn't, they might reveal some important strategy distinctions among methods that weren't known previously. I hope you'll pursue the developement of those criteria. Mike Ossipoff _________________________________________________________________ Get your FREE download of MSN Explorer at http://explorer.msn.com
