At 09:29 AM 12/9/2007, Diego Santos wrote: >2007/12/9, Jan Kok <<mailto:[EMAIL PROTECTED]>[EMAIL PROTECTED]>: >Well, I consider almost any form of Bucklin more palatable than IRV, >and of course it is better than Plurality. > > >Bucklin is not so bad, but I still think that a better ranked method >should not provide incentive for strategic truncation.
As I've pointed out, in real elections, there is little such incentive. Bucklin is basically quite like Approval, with *less* "truncation" incentive, and Approval is quite good. Absolutely, in the full universe of election possibilities, one would want to be able to express exclusive preference without thereby making the vote moot. A Bucklin ballot could accomplish this. The question is how to count it. For example, suppose that a three-rank Bucklin ballot were counted as pure approval. Put any vote on it, you have approved the candidate. But you can rank your choices. And then the ranked ballots are analyzed to see if there is a candidate who pairwise beats the Approval winner. If so, *there is a runoff*. But that would actually be rare, I'd predict. The runoff preferentially favors the Approval winner, normally. I've made the argument more clearly with a similar scheme with Range (Range winner vs. pairwise winner analyzing the Range ballot as a ranked one). Basically, if the Range votes are sincere, the Range winner supporters have higher preference strength and can be expected to turn out differentially in favor of that candidate, whereas the pairwise winner is by weaker preference. But if something went awry and the Range votes were *not* sincere and reasonably commensurable, the majority can turn the election to the pairwise winner, easily. Thus the method satisfies the Majority Criterion -- all forms. ---- Election-Methods mailing list - see http://electorama.com/em for list info
