VA, Margins, & voter wishes

[Mike Ositoff]

Replying farther down in this message:

On Fri, 2 Oct 1998, Blake Cretney wrote:

> On Thu, 1 Oct 1998 17:51:11    Mike Ositoff wrote:
> >
> >
> >I'm going to back out of my agreement with Blake's stat4ment
> >that VA is out of the rank-count discussion. For 1 thing,
> >the fact that I don't believe that rank-balloting is a
> >practical public proposal doesn't mean don't believe that
> >1 rank method has more merit than another. Some people do
> >advocate rank methods for public use, and I suggest to them
> >that VA is the one that does what they want, if they want
> >what most people want from a rank method. And the merit of
> >methods can be discussed regardless of what some of us
> >believe in proposing publicly. Also, pursuing this discussion
> >helps prove my claim that rank count discussion will always
> >be a hopeless mess.
> >
> So, one of your reasons for advocating VA is to discredit Condorcet
> methods in general.

I replied here, and now reply farther down. Each reply is to
a remark farther down.

> 
> >Also, I'm not the only one here to speak for VA vs Margins.
> 
> ...
> >So you see, Blake, we all agree with the desirabilty of
> >that measure. But the Condorcet Criterion, by itself, isn't
> >enough. That's why I added:
> >
> >If a majority of all the voters indicate that they'd rather
> >have A than B, then if we choose A or B it should be A.
> >
> >Do you disagree with that, Blake? Or other Margins advocates?
> >

> Yes, I disagree.  What your saying is that a pair-wise vote
> where the winning side has a majority of voters must always
> take precedence over a pair-wise vote where this is not the
> case, no matter what the margin of victory is.

Margin of victory? You've neglected to show why margin of victory
is important. Majority is widely agreed to be important. If you
believe that margin of victory is important, you need to show
why. As a personal standard of yours, in its own right, or
with respect to some agreed-upon standard? If you use it, can
you show why it's important?


> So, when one must be over-ruled, a victory of
> 52 to 48  with 100 voters
> must take precedence over a victory of
> 49 to 4
> despite the fact that for those expressing a preference, the second
> vote seems clearly more decisive.

If you consider margins important.


> 
> Furthermore, if you accept VA, it has the side effects that
> 60 to 40
> must take precedence over
> 59 to 3
> And
> 48 to 47
> must take precedence over
> 47 to 2

It's always easy to find examples where compliance with one standard
results in really extreme, & intuitively bad, violation of some
other standard. One must judge by which of the standards seems
more important.


> These results seem contrary to our usual idea of what constitutes a
> decisive vote, and are not required by GMC, but are the result of VA.
> 
> The problem here is the basic belief that a majority of people
> participating in a pair-wise vote is not a real majority, that
> only a majority of participants in the election as a whole is a
> real majority.

That's what you said before. We've been over this. We agree with
your pairwise plurality being important. It's the basis of the
Condorcet Criteion, important to all of us (except Don).

That criterion isn't enough. Often it won't apply. Then we
need something else. Looking at majorities doesn't contradict
the idea that pairwise pluralities are important; it takes
over when the criterion based on them doesn't apply because
no one beats everyone else.
> 
> But consider that a majority of those participating in the election, is
> not necessarily a majority of those elligible to vote.  Typically,
> however, we interpret majority rule as a majority of participants.
> So, in a pair-wise vote, I think it is reasonable to think that it is the participants in that pair-wise vote that count.
> 
> See above.

Mike

> 
> -----== Sent via Deja News, The Discussion Network ==-----
> http://www.dejanews.com/  Easy access to 50,000+ discussion forums
> 
>