[EM] Condorcet Criterion for plurality.

[Markus Schulze]

Dear Mike,

you wrote (13 Dec 2000):
> Markus wrote (13 Dec 2000):
> > The intention of IIAC is to guarantee that the result of
> > the elections cannot be manipulated by running additional
> > candidates who change the result of the elections without
> > being elected. But you define IIAC is such a manner that
> > even plurality --which is highly vulnerable to this strategy--
> > meets this criterion. Therefore your definition of IIAC has
> > nothing to do with the original intention of this criterion.
> > Therefore --and this is what Blake is saying-- you have to
> > present an additional justification for your definition of
> > IIAC. And in so far as you are unwilling or unable to
> > justify your definition of IIAC, the usefulness of your
> > "universally accepted" concept is questionable.
>
> Justify IIAC? I've repeatedly said that IIAC doesn't mean
> anything to me. Its importance derives entirely from the fact
> that lots of people, including IRVies, keep bringing it up.
> That's great when they do that, since the only complete IIAC
> definition that I've heard of says that Approval is better than
> IRV and that even Plurality is better than IRV. Obviously there
> are many reasons why those methods are better than IRV, but when
> the IRVies' own citation of Arrow counts against IRV, that makes
> IIAC useful. _That's_ my justification of IIAC. Maybe Arrow's
> own definition of IIAC is different from mine. Maybe it's
> justified in some way that mine isn't. Fine. I don't care.

Of course, when you define IIAC in such a manner that IIAC has
nothing to do with the original intention of this criterion,
then --unless you can give an additional explanation why your
IIAC should be desirable-- the observation that plurality
meets your IIAC is quite meaningless. The usefulness of your
"universally accepted" concept is questionable, when you cannot
explain why the resulting criteria describe desirable properties. 

Markus Schulze