Approval Voting satisfies the Condorcet criterion (as does Plurality). The idea that it does not is based on the imputation of unexpressed preferences. That is, *if* there were more expressible ranks, and the voters used them, the outcome could change.
(But if the Condorcet criterion *requires* that all preferences be expressible, i.e., that the number of ranks equals the number of candidates, then, of course, any method which does not allow that does not satisfy the criterion. I don't know the exact wording. But I've never seen anyone objecting that a method which only allows N ranks in the presence of more than N candidates does not satisfy the Criterion. Obviously, such a method requires equal ranking. Note also that a voter might prefer to add more than one write-in candidate. Should this too be allowed? If not, why not? -- the answer, ballot complexity, would similarly apply to reducing the number of expressible ranks when there are many candidates.) Sometimes we forget that Approval and Plurality are ranked methods with only two ranks. The problem with Plurality, of course, is that equal ranking is generally prohibited. If overvoting were allowed, with one stroke of the deletion pen in the election code, Plurality would become Approval, which has got to be the simplest of the proposed election reforms. (But Asset voting, using plurality counting, i.e., no overvoting allowed, is just as simple to count and uses the same ballot. If overvoting is allowed, the counting gets more complex because in this case the votes must become fractional votes, i.e., the method becomes Fractional Approval Asset Voting, my current favorite. But what I'll now call Standard Asset Voting -- i.e., vote for one only -- is really almost as good without the counting complexity. Pick the candidate you most trust and vote for him or her, no worry about wasted votes.) ---- election-methods mailing list - see http://electorama.com/em for list info
