At 08:53 PM 1/25/2010, Jameson Quinn wrote:
I'd love to hear what others think of these proposals.
I think you are on the right track. Proxy systems introduce deliberative elements to amalgamation systems, which just has to make them more intelligent. They aren't terribly necessary when direct deliberation is possible, but with scale, direct processes become impossibly tedious, rapidly.
Personally, my favorite basis for a hybrid system is Bucklin with equal-rankings and gaps allowed.
Which is almost what was used. Equal-rankings, in Duluth Bucklin, were allowed in third rank, and the tradition of preventing multiple "Favorites" was very strong. I wouldn't be horrified to see Bucklin implemented today with quite the same restriction, I just think it's stupid, it would do no harm, and some good.
(Bucklin does poorly on criteria, but relatively well in practice. Obviously, hybrid systems are never going to be the best on pure criteria anyway.)
Right, because the criteria were designed to study single-ballot deterministic systems fed by pure preference data, equal ranking not allowed and not even considered of importance.
Thus the King of Criteria, the Condorcet Criterion, or its stronger relative, the Majority Criterion, are blatantly defective, as can be shown by real-world situations where, with completely sincere voting, they require a winner that every voter, once the votes are revealed, is very likely to consider a poor choice. Even though it is still their "personal favorite," it's a bad choice for the society, and they will know it, and would have to be pretty sick to insist on it. Quite simply, the vast majority of people wouldn't.
The Pizza election, three friends choosing a pizza. Two prefer Pepperoni, but will accept Mushroom, it's a decent choice for them. One is Jewish, can't eat Pepperoni, period, but prefers Mushroom. Mushroom must lose to Pepperoni, according to the Condorcet Criterion and the Majority Criterion. And it's a terrible choice for the group to make (assuming they must choose one pizza).
In a repeated ballot situation, ordinarily, the majority will accept the strong preference of a minority, as long as their own preference is weak. And this will show up -- shhh... big secret -- in differential turnout in a runoff election. It tests preference strength. The minority will have high motivation to turn out and vote, due to high preference strength, whereas the majority, by the terms of the problem, have low preference strength, they are actually going to be happy with either outcome.
Hence my prediction: as long as the votes weren't distorted in some way in the primary, a Range winner in a runoff with a Condorcet winner has a natural advantage, and will likely win. Absolutely, this would be invisible in the simulations that have been done, which think of a runoff as if it were simply a re-analysis of the ballots or underlying preferences. It's quite different, with different voters motivated to vote for different reasons in a different context.
But the runoff is important precisely because the Range results might be distorted, and the preference information better.
As a democratic principle, whenever the preference of a majority is being discarded, the majority should consent. Not necessarily the same majority, but a majority of the voters who choose to vote on the question. Those who care.
---- Election-Methods mailing list - see http://electorama.com/em for list info
