Ah, but you contradict yourself. See below.
Sorry if you don't want me mentioning borda, but the only way to get around what you see as "wrong" is by doing exactly what borda does. And that is bad.Borda isn't the only way to preserve rank differences pairwise.
I didn't say it was. I used it as a familiar and well understood example of something that does that. Anything that preserves such differences is going to have similar problems.
Condorcet allows people to never have to consider strategy when placing their vote. ( ... )Wouldn't it be "strategic" for a bloc to vote "Z>A" at the bottom of the ballot when they want neither Z nor A, but can use that to defeat A in any method that counts pairwise-matrix numbers instead of ballots?
My wording was a bit sloppy, because it is true that condorcet SOMETIMES can reward insincere voting. It is not perfect. It is, in my opinion, really really close....close enough that insincere voting will not have a significant effect on elections, and most importantly to me, will allow middle ground candidates to be elected rather than forcing people into two opposing clusters.
In most cases you will not gain much from doing what you describe, assuming they don't actually prefer Z to A. Only if Z, A and the preferred candidate are in a condorcet cycle would that have any effect whatsover. And if they are in a condorcet cycle, that get's pretty hard to predict whether A or Z is the biggest threat to the preferred candidate. So I just don't see it being a significant factor.
In most cases you will not gain much from doing what you describe, assuming they don't actually prefer Z to A. Only if Z, A and the preferred candidate are in a condorcet cycle would that have any effect whatsover. And if they are in a condorcet cycle, that get's pretty hard to predict whether A or Z is the biggest threat to the preferred candidate. So I just don't see it being a significant factor.
---- election-methods mailing list - see http://electorama.com/em for list info
