Intuitive Loser Criteriion

[Mike Ossipoff]

I said that I'm going to reply pre-emptively to arguments that Bruce
might use after I leave this list, as I will soon be doing. One 
thing he might post about is an "Intuitive Loser Criterion".
This criterion, which I'll define & discuss briefly here, is
a good example of how far it's necessary to reach in order to
try to find a criterion for Smith//Codorcet to fail.

I'll talk about that criterion, but first I'd like to
comment about Bruce's claim that Smith//Condorcet can pick
some "strange" winners, or that it has a "flaw". The "flaw"
is that it doesn't pick its winner by Copeland's candidate-
counting standard. Excuse me, but any violation of Copeland's
standard is a flaw? Condorcet isn't Copeland, and it won't
pick according to Copeland's standard. It doesn't pick according
to how many alternatives a particular alternative beats &
is beaten by. If it did that it would be Copeland. I've
talked about why Condorcet's standard, properties, &
criteria it meets are more important than how many alternatives
a particular alternative beats & is beaten by.

Strange winners? As I've said, Bruce's example of that is indeed
strange, as are all of his Condorcet examples in that posting.
They're strange because in all of them the alterntive least
disliked, in terms of how many people say something else is
better, that alternative somehow is beaten by many more
alternatives than are the others. No explanation or scenario
is given for how that could happen. Strange? I agree.

Intuitive Loser Criterion:

A method must not choose an alternative with the worst Borda
score & the worst Copeland score.

Come again??

I've used this analogy before, but that criterion sounds like
when Kenneth Patchen's giraffe said, "Ok, then, lets just say
that the one with the longest neck gets all the jellybeans."

I could write my own intuitive loser criterion:

Mike's Intuitive Loser Criterion:

A method must not choose an alternative with the worst Borda
score & the worst Condorcet score.

Or (why be shy):

A method must not choose an alternative with the worst Condorcet
score.

Or, better yet:

A method must not choose any alternative that isn't the winner
by Condorcet's method.

***

When we write criteria that feature the method that we're
advocating...guess what?: The method we're advocating
will be the method that doesn't violate them.

I assume that Bruce is serious when he says that Copeland
is better because it meets a criterion that can only be
failed by a method that can pick the alternative with the
worst Copeland score.

***

Ordinarily I wouldn't reply to this argument till Bruce uses
it, but, as I said, it's necessary to reply to it pre-emptively,
since Bruce will likely use it later, perhaps after I've left
this list, as I'll soon do.

***

Mike




--