David-- When I was replying to some of the criteria you named, I left out the tacatical ones. More about the rest of them in my next posting, but right now I just want to emphasize that the one that says that no one should have incentive to not rank his favorite in 1st place is unquestionably the thing that I consider most important for a voting system. In general, no one should have to rank a less-liked alternative equal to or (especially) over a more-liked one. But when someone has incentive to not rank their favorite 1st, that's surely the worst violation of all. *** It seems to me that the criterion about tyrants is the Majority Loser Criterion. And that the one about split-vote is really equivalent to the Independence From Clones Criterion. Maybe IRO meets that criterion, but Schulze's method meets it too, along with additional especially important criteria-- such as the tactical criterion mentioned above. *** It seems to me that all the VA methods meet that "1st Choice Criterion" (that says no one should have to not rank his favorite 1st). One thing for sure is that no one has to do that to protect a CW. But I don't believe there's any incentive to do so, even offensively. Approval, though it doesn't let you vote your favorite over everyone, when you need to support a compromise, it does let you give a full vote to your favorite--as much of a 1st place vote as anyone can give to any candidate. But Approval probably doesn't meet your 1st Choice Criterion as it was intended, since Approval doesn't let you vote your favorite _over_ everyone. That's why I say that Approval gives half the strategy benefit of the best rank methods. More later Mike Ossipoff
