[EM] Unranked IRV versus Approval - divergent winners exist!

[Tom Ruen]

----- Original Message -----
From: "MIKE OSSIPOFF" <nkklrp at hotmail.com>
To: <election-methods-list at eskimo.com>
Sent: Monday, April 02, 2001 3:33 AM
Subject: Re: [EM] Unranked IRV versus Approval - divergent winners exist!


>
> >Because there IS no best that will be agreed by all.
>
> But presumably there are some methods that seem to _you_ to be better
> than others. So why should you advocate less.
>
> Majority rule is not a
> >well defined concept with 3 or more choices.
>
> ...certainly not by IRVies. There are 2 definitions of majority--the
> universal definition, and the IRVie definition.
>
> What does majority mean, no matter how many candidates there are?
> More than half of the voters.
>
> What does majority rule mean, no matter how many candidates there are?
> A majority getting its way. With nearly all single-winner methods
> (but maybe not with Borda), a majority can always get its way.
> If it wants to elect someone, that's easy. It also has the power to
> ensure that a certain candidate _won't_ win. If we were to just say
> that majority rule means a majority getting its way, and not qualify
> that, then, no how many or how few candidates there are, majority rule
> would be a useless standard, since all methods meet it.

If a majority supports a single candidate, any method may do. The question
is how do you convince people to compromise to find a majority among 3 or
more strong choices.

>
> So how about a majority getting its way by sincere voting? That's
> something that the majority defensive strategy criteria measure, and
> IRV fails all of those criteria. Condorcet's method meets all of them.
> Approval meets one of them, WDSC, in addition to the additional
> defensive strategy criterion FBC.
>
>
>
> Approval and Condorcet are
> >not single votes - they are N and N*(N-1)/2 elections which must be
merged
> >somehow to pick a winner.
>
> It isn't clear why you think that Approval is about N elections.

Approval asks voters N independent questions. Will you vote for A? Will you
vote for B? ...

Condorcet asks the voters N*(N-1)/2 questions. Who do you support more
strongly A or B? Who do you support more strongly A or C? ....