It should be noted that for some methods, like the various Condorcets, there is no meaningful ranking of losers. And do we really want a Condorcet winner??Torco wrote:http://www.cs.wustl.edu/~legrand/rbvote/calc.html
this seems to work wonderfully
EDIT: damn, it doesn't give the ranked list of losers. it does tell us that preliminary results indicate the condorcet winner is lyra.
Personally, I'd just go with a plain ol' Borda Count, ideally the Kiribati subtype. Would anyone have any major objections to that? The big problem I can see is that since there was no restriction on entries, cloning will skew the results. So, I'd suggest a two-stage process, combining an initial Approval round to eliminate the least popular candidates and minimise cloning issues, and then a Borda round to finalise the results. [I wouldn't just use Approval all the way through, because IMO Approval Voting is too worried about avoiding upsetting people (which is less of an issue here because there are multiple winners and nobody's getting shot) at the cost of paying less attention to who people actually prefer]. This needn't be too complicated for voters, because you can just ask for a ranked ballot and impose a threshold on it to indicate approval; then, candidates who fail the approval round can be struck off the ballots. We get the benefits of two rounds of voting, but we only have to vote once.
[While I'm at it:
- Range Voting: too vulnerable to subjective calibration issues. I.e. people who express themselves more enthusiastically get more of a say than leveler heads.
- MBC: not necessary, I don't think. No clear reason in this case why ranking more candidates should increase the voting power of an individual]