[EM] thoughts on the pairwise matrix

[Abd ul-Rahman Lomax]

At 01:18 PM 11/29/2005, Paul Kislanko wrote:
>And in some methods I can say A=B>C, which may or may not be closer to my
>preference. I also prefer A>B and for heaven's sake NEVER C. Which gets into
>the value of truncation and the subjective interpretation of what that
>means.

The strength of preference indicated is not a part of the Condorcet 
method at all. Most ranked ballots have no difficulty expressing the 
former, A=B>C, the problem is only election counting rules which 
might prohibit overvoting.

And A>B, NEVER C, is, quite simply, A>B>>C, which requires the 
addition of an Approval cutoff; some Condorcet-compliant methods use 
this to determine the winner if there is no Condorcet winner. 
Otherwise it is irrelevant except as dicta.

>It's the subjective interpretation of ranked ballots accompanied by the
>method's claim that it is objective with no unstated assumptions that I find
>bothersome. Not bothersome enough to dislike such methods, but enough to
>want more precision in the descriptions and "proofs" accompanying them.

A ranked ballot is a ranked ballot, nothing less (unless truncated by 
design) and nothing more (unless the feature is added as something 
other than pure ranking information).

Methods make no claims, people make claims. In the absence of a 
specific claim, the objection is not intelligible.

Nothing is purely objective, but interpreting a difference in rank 
expressed on a ballot as a voter preference is relatively objective!