[EM] Don't be so quick in handing out those Nobel prizes:

[MIKE OSSIPOFF]

I don't usually reply to Don, but I'll comment on his example.

He says I propose...

>scheme
>to allow a minority party candidate to win over a majority party candidate.

Actually Approval honors majority rule and IRV doesn't. In Approval
a group consisting of a majority of all the voters has a way to
make a certain candidate lose, if there's someone whom they all prefer to 
him,  without any of that majority
reversing a sincere preference. That isn't true of IRV. That's
the basic minimum requirement for a single-winner method. Without
that, you have no majority rule.

>This is Mike's ongoing agenda.  You should take that into consideration
>whenever you read any of his tripe.
>
>      Consider an example: 60 A,  30 B,  10 C
>      Now it would appear that the 60 A faction is a shoo-in to win this
>election, but it will depend on one of four methods winning the methods
>election, Plurality, Irving, Condorcet, or Bucklin.
>      If the 60 A faction is careless and votes at random, each of these
>four methods will only receive about fifteen votes.

If people vote at random, don't expect much from any voting system.

For an Approval vote,
maybe our IRVie friend is afraid that the 60 A voters will feel
that they need to vote for B (I assume that B is a middle compromise).
The only reason why the A voters would need to vote for B would
be if C is going to beat A. But C only has 1/6 the support that A
has. That would be a very big mis-estimate on the part of the A voters, to 
believe that C was going to beat A.

In any case, if your point is that strategic judgement is needed
with Approval, and that therefore Approval isn't Condorcet, that's
something that I've never denied. But that's irrelevant to the
comparison between Approval & IRV.

>      The 30 B faction needs Borda in order to have a chance of winning,
>Approval would be better.  So, Borda gets thirty votes and is the winning
>method because this is a `Plurality' election of the methods.

Wrong. Voter's Choice is a Plurality election among the candidates,
in which people place their Plurality vote guided by the result
of the method that they designate.

In such an election, if you, Don, believe that IRV
in some way picks the best compromise that you can get, then you
should designate IRV. Most would agree that Condorcet does that best.

>      Now, if enough of the 60 A faction carelessly made a choice for
>candidate B, then candidate B is the winner, this is possible.

...only if the A voters thought C would beat A, when actually
A has a majority, and C has 6 times fewer votes than A has.

It takes a big misjudgement, typically, to give away a majority
win in Approval. I posted at length on this matter some time ago.
Does Approval need strategic judgement? Sure. You want a method that
doesn't? Then Condorcet is what you want.

>      In order for the 60 A faction to protect itself under Mike's scheme,
>it will need to always be on its toes by making sure that every one of
>their supporters vote for the same method, Plurality, and bullet votes
>their candidate A.

No, each voter should designate the voting system that s/he
believes will do the best job of picking the best compromise that
s/he can get. Certainly, if A is that much more poplular than C,
the A voters would know that, and their Approval & CR ballots
would reflect that information: They wouldn't vote for a 2nd
choice. They'd only vote for A in Approval & CR.

What method should they designate? Condorcet does the best job
of picking the best you can get. Designate Condorcet. In particular,
designate SSD if it's a public election, or Cloneproof SSD if it's
a small committee vote.

Mike Ossipoff



_________________________________________________________________
Get your FREE download of MSN Explorer at http://explorer.msn.com