Sorry I haven't replied on this thread for a while. I think that I got some agreement on some factual points, but didn't really explain why these points mattered.
I admit that you can come up with examples where, because of irrational voting, winning-votes is advantageous in terms of electing a sincere CW. I also admit that voters will sometimes vote irrationally. But I think we agree that as voters understand the method better, any claimed advantage of winning votes over margins, at least with regard to the truncation-resistance issue, vanishes. This is because, as I've established, there's no strategic benefit to truncation over order-reversal (or random ranking). Note that I'm only talking about knowledge of the method here, not knowledge of how others will vote. My point is that even if you don't know how other people are going to vote, truncation gives you no new strategic opportunities, and so to voters who understand the methods, there's no difference in strategy. But since voters often won't understand the method, one effect will be that voters who give incomplete rankings will tend to be punished for their ignorance, whether these incomplete rankings represent trying to get away with something, or not. You could try to educate them, but that defeats the whole purpose, because the idea of winning votes is that voters will either use truncation or give up strategy entirely in exactly those situations where they should rationally rank randomly or order-reverse. The presence of the partial ranking option on the ballot seems to be there only to trick people into compromising their interests in the hopes that some of those who are fooled deserve it. I think we'd have to be pretty desperate before suggesting a method that only works because people don't understand it. And so the argument is that things are that desperate, that Marginal methods are too affected by strategy to be usable. Now, many people would argue that any Condorcet method will be unusable because of strategy. I think that is false, and that by many measures Condorcet criterion methods tend to be the most strategy resistant of the non-random methods. And groups often use tournament style voting, with essentially the same problems, without much incident. But that's a whole other subject. My point is that those who would argue that Marginal methods are too subject to strategy should really give up Condorcet methods altogether. This is because every example of truncation-resistance is an example where the truncators could have got their way if they understood the method better. Ignorance is admittedly widespread, but I don't see how we can count on it in every body we recommend Condorcet to. So, I think it is inescapable that if we abandon Margins, we must abandon winning-votes as well. The truncation-resistance arguments against Margins are really arguments against winning-votes too. --- Blake Cretney ---- For more information about this list (subscribe, unsubscribe, FAQ, etc), please see http://www.eskimo.com/~robla/em
