Chris-- I'm referring to the procedure that you suggested: When the wiinnner by some method is determined, then each ballot gives an approval to that winner, and also to every candidate whom it ranks higher.
You haven't named that procedure yet, and so, until you do, I'm calling it "Approval-Fill-In". Three things occur to me about Approval-Fill-In: 1. It would improve on any and every method. From the point of view of each voter, it is an improvement, because every voter would want to help, against that winner, the candidates whom s/he likes better. It would elect someone more liked. It would improve social utility (SU). 2. There's no reason for it to only use one rank-count, or, in general, one method. Each voter could designate any method that the (presumably secure) count procedures are set up for, and use it for hir Approval-Fill-In. Different voters may have different ideas of which method best chooses the CW, their least-liked candidate to optimally approve.. 3. In fact, Approval-Fill-In, based on a rank-count could be an _option_ for an Approval election. There's no reason to not allow any voter, in an Approval election, to choose her approvals in any way that s/he wants to. They're hir approvals, after all. All of the Approval election approval-management options that I've described could be offered, and so could Approval-Fill-In, based on the rank-count of the voter's own choice. As I've often pointed out, an option is incomparably more proposable than a new voting system. People could argue the fairness, rightness or results of a new voting system, and it could take forever to get it enacted, if ever. But, since each voter's vote is hir own, to do with as s/he pleases, no one can object to an option. (Though I'd oppose one if it violated FBC). I consider ICT to be the best rank count, and the best way to try to choose the CW. It doesn't meet Condorcet's Criteriion, but it encourages sincerity in ways that Condorcet does not, and is free of Condorcet's sinceriy-discouragements. So, though I've suggested a large set of approval-management options, including MCA, MTA, ABucklin, MTA2, MTA/MCA, ABucklin2, and conditional versions of all of those--I considerICT to be better than those, at least partly because of its somewhat better defection-resistance. So my first suggestion for an approval-management option would be Approval-Fill-In. My suggested method for a voter to designate for hir Approval-Fill-In would be iCT. I suggest that the best route to enactment of a good rank method would be Approval. I've spoken of how Approval is nothing other than an elimination of Plurality's ridiculous forced falsification requirement. And I've mentioned how therefore Approval is the only method that would obviously be an improvement over Plurality, and only an improvement. And, as I said above, no one can object to the fairness of an _option_. And the Approval-Fill-In option could use any method as its basis. This approach would be the best for the person who is reluctant to replace Plurality, since it merely proposes the method that can only improve on Plurality. The minimal correction of Plurlity's ridiculous problem. It would also be best for the advocate of any rank count, because it can lead (as an approval-management option) to something demonstrably better than that rank count (that rank count with Approval-Fill-In). Mike Ossipoff
---- Election-Methods mailing list - see http://electorama.com/em for list info
