Forest wrote: >[...] if we interpret truncations as NO and ranked as YES. > >This time Bush wins, having greater approval than Gore. > >I suggest that this is a reasonable interpretation, and that the resulting >version of Approval Completed Condorcet is not too shabby.
The problem with handling Approval Completed Condorcet in this fashion is that it creates a strong incentive to truncate those candidates you disapprove of, even if you have a preference between some of them. In the example I gave, for instance, it would create a strong incentive for the 12 Gore>Bush>Nader voters to bullet vote for Gore. This denies information to the preceding Condorcet vote; we may end up with the Approval completion corrupting the initial search for a Condorcet winner. As Alex said, it seems that putting a yes/no vote would not be too much extra effort. A 5 grade (4 grade? 6 grade? ABCDEF?) ballot that allows implicit yes/no votes by the grade given is a possible solution here, and allows for very intuitive Approval Completed Condorcet. Has anyone tried to figure out what criteria are satisfied by such a form of ACC? If pure ranked ballots are the only option, then I'm still very comfortable with SSD. It takes a very peculiar circular tie in order to cause any sort of trouble with SSD. -Adam
