At 03:49 PM 5/19/2010, Kathy Dopp wrote:
"Satisfaction Approval Voting" is a new proportional representation approval voting method, devised by political scientist Steven J. Brams, Department of Politics, New York University and D. Marc Kilgour, Department of Mathematics, Wilfrid Laurier University.
SAV is essentially Plurality-at-large, single-vote, with division of votes in case of "overvoting."
I proposed the same method, in effect, with FAAV, Fractional Approval Asset Voting. SAV is just this without the Asset part to handle what would otherwise be wasted votes.
Unlike instant runoff voting, Satisfaction Approval Voting (SAV) can use existing ballot layouts, is precinct-summable (additive), and treats all voters equally and fairly.
It's tricky. It relies upon clones to negotiate on a single candidacy or see their votes wasted. FAAV would empower bullet votes, but also allow 'virtual committees" to do that negotiation, now knowing what voting power they have. Overall, far more efficient, I'd suggest, and SAV can still waste a huge number of votes.
If you are going to do a proportional representation system, SAV relies upon an assumption of equal approval of candidates. That makes it impossible for it to analyze and find true representation. Far better would be an STV-like system that seeks to assign a relatively trusted representative to as many voters as possible.
The system could be, as to ballot, SAV, but with a Range ballot instead of a simply plurality ballot. I won't describe details here. SAV is certainly an improvement over Plurality-at-large, just as ordinary approval is an improvement over vote-for-one Plurality. SAV reduces to plurality if voters vote for a single candidate. I think SAV is a nno-starter because of the effect on voters, who may well regret splitting their vote. But with a system where either votes aren't wasted at all (Asset) or vote wastage is minimized (STV-like systems that devalue or set aside ballots based on the election of representatives named on them and the ranking or rating of those representatives)
I found the paper quite obtuse, unnecessarily difficult. Instead of straighforwardly stating the method, the theoretical justification was first outlined. Frustrating way of writing, I'd say.
The basic method, as I understand it. To elect N representatives, votes vote an approval ballot, and may vote for as few or as many as they choose. If they vote for M candidates, their vote for each becomes 1/M. (The ballot is not considered if it does not contain a countable vote for a candidate, I presume.) The N candidates with the most votes wins.
Most of the paper is justification for why this is a Great Idea, and why the obvious problem (dilution of voting power when a voter votes for more than one) is a feature, not a bug, based on arguments that candidates will cooperate to choose to not split their vote. I find that rather ... speciulative. I.e., it may work sometimes, and fail spectacularly at other times.
Asset voting, FAAV, would use the same ballot and counting mechanism, but would simply assign those votes to the candidates, who *then* negotiate who gets the seat. If I cooperate with you to create, say, a seat for you, you know where your bread was buttered. I'll have some influence. In full-blown continuous Asset systems, I would only be choosing you as a representative in deliberation, I'd retain the right to vote my votes on issues, if I choose. Some would, some wouldn't.... your vote would count all my votes if I don't directly vote.
But short of that, if it's a pure election, final till the next election, no continuous voting rights, I would still ordinarily have privileged access to you. Or, next election, I won't give you my votes again! (If I still get some.... it would depend on the voters who supported me understanding the situation: did you renege on any agreements we had, or was I stupid and to blame for trusting you in the first place .... or both.))
---- Election-Methods mailing list - see http://electorama.com/em for list info
