[EM] Craig's posts

[Roy One]

I'm finding it a bit hard to get the gist of what points Craig is
trying to make in his recent posts. Maybe it's because they refer to
private conversations that preceded them.

By "Condorcet finds the wrong number of winners sometimes," I presume
he means that some elections do not have Ideal Democratic Winners
(Craig: please see http://www.electionmethods.org for definition),
and he has taken "Condorcet method" to mean precisely a method that
finds the IDW if there is one, but does not find a single winner when
there is not.

Now, addressing Craig directly:

That is a mistaken belief. There are several Condorcet methods,
including Plain Condorcet (the original), which find single winners
even when no IDW exists. Finding the "wrong number of winners"
happens in plurality in the case of an exact tie, as well as in IRV.

A Condorcet method is simply one which finds the IDW if one exists.
The various Condorcet methods differ in how they find a winner when
no IDW exists.

"The pairwise comparing idea is undesirable from the point of view of
voters."

I don't see why. If you ask a voter to rank a list of candidates, you
know how he feels about any two of them relative to each other. Do
you think that perhaps a voter likes A better than B, unless C is
running, too, in which case he like B better than A? That doesn't
make any sense. Perhaps you've surveyed voters about this, or have
some other support for this claim?

The "exact algebra of principles of preferential voting" seems to
refer to some existing work that evaluates current voting methods by
these means. I would be interested to see it, if a reference could be
supplied.

The "axiom of no harming of the candidate voted for" is an
unsatisfactory one. It necessitates plurality rule, which (it is my
experience) is something that is "undesirable from the point of view
of the voters."

Roy

__________________________________________________
Do You Yahoo!?
Get email alerts & NEW webcam video instant messaging with Yahoo! Messenger
http://im.yahoo.com