[EM] Approval Strategy A- Question for Rob LeGrand

[Gervase Lam]

> Date: Fri, 21 Nov 2003 01:21:16 +0100 (CET)
> From: =?iso-8859-1?q?Kevin=20Venzke?= <stepjak at yahoo.fr>
> Subject: Re: [EM] Approval Strategy A- Question for Rob LeGrand

> My thoughts:
> Plurality will be the most proportional because it can occasionally
> elect a fluke candidate to represent weird voters.
> IRV will be next, electing only mainstream flukes (no other flukes).
> Borda will be somewhere around here, although I don't think you'll be
> able to model it accurately.  Specifically, how can you account for the
> incentive for a party to run multiple candidates?
> Approval will be among the least proportional.
> Condorcet methods will be last, picking the centrist whenever the voters
> decide not to truncate him.

Some people argue that (Plurality) PR can cause too much diversity.  This 
can be detrimental when the elected chamber try to vote on things.  This 
is the reason why Italy went from proportional to a single seats election. 

So, there could still be room for Approval and Condorcet.  Nevertheless, I 
did (and still do?) agree with you that if you want PR, why not just stick 
with Plurality.  Plurality is simple, which is a bonus.

However, can Cardinal Ratings be put above plurality?  To me, it seems the 
most proportional of all of the methods around.

Immediately below that would be Borda as it is the "next step" down from 
Cardinal Ratings.  Borda would be followed by Approval then Condorcet.

May be the hierarchy should be more like a "Y" shape.  Cardinal Ratings 
and IRV would both branch off from Borda.

This makes me wonder about the word "proportional" in a method.  In a 
given method, what part of the ballot is proportional?

Take Plurality and Cardinal Ratings.  Cardinal Ratings is proportional for 
every candidate on the ballot.  Meanwhile, Plurality is only proportional 
with respect to the top most voted candidate.

The closest reasonable method that I can think of that uses Cardinal 
Ratings is your Gradual Information Approval method.  Instead of stopping 
when there is 1 candidate left, why not stop when there are N candidates 
left?

Thanks,
Gervase.