Definition of "Pairwise Method"

Mike Ossipoff dfb at
Tue Apr 16 02:13:48 PDT 1996

Bruce said that he disagrees with my use of English in the matter of
whether the term "Pairwise" should be limited to methods that get their
result using only arrays p & q.  ...Which is to say: methods that 
go only by how many people voted x over y, how many people voted y over
x, and how many people voted them equal, or ranked neither.

I said that I've proposed refinements of Condorcet, and I've proposed
other Pairwise methods too, which use other information than those totals,
and that those methods shouldn't be excluded by the definition of
Pairwise methods.

Now just what is it that you disagree with about that English, Bruce?
If Pairwise methods must use only arrays p & q, I I've propsoed methods
that require more information than that, but which are based on pairwise
comparisons, and which I claim should be called Pairwise methods, what's
wrong with that English?

What your pencil, paper & slide rule example seems to be saying is that
you didn't really mean to define Pairwise methods as methods that need
only arrays p & q, but you only meant to say that methods needing only
p & q are Pairwise methods. No, I'm sorry, but you were including the
condition of requiring only arrays p & q as a condition for a method
being a Pairwise method.

I'm going to take another look at your Jean-Charles method, and then
comment on it.



More information about the Election-Methods mailing list