[EM] Re: Approval/Condorcet Compromise

[Russ Paielli]

Ted Stern tedstern-at-mailinator.com |EMlist| wrote:
> On 9 Mar 2005 at 10:29 PST, Forest Simmons wrote:
> 
>>Most of the proposed Approval/Condorcet Compromises assume that the CW is 
>>more desirable than the Approval Winner when they are not the same 
>>candidate, i.e. the Approval Winner is only to be considered when there is 
>>no CW available.
>>
>>That seems to me like a kind of one sided approach to "compromise."
> 
> 
> Well, if one is a Condorcet advocate, one might view Approval as means toward
> the end (intuitively appealing cycle resolution), and not necessarily as an
> equally valid end.

That's right. I don't see the need for a compromise between Condorcet 
and Approval. Condorcet is fine except that it needs a good way to break 
out of cycles, and Approval is made to order for the job.

Perhaps I am fooling myself, but the more I think about the method I 
proposed (or resurrected) for combining Condorcet and Approval, the more 
it makes sense to me. Here's the method again for reference:

Each voter ranks the candidates and specifies an Approval cutoff. The CW 
wins if one exists, otherwise the least-approved candidate is dropped 
until a CW emerges.

The need for each voter to specify an Approval cutoff does complicate 
the implementation a bit, but it also gives the voter a critical mode of 
expression.

Think of the Approval counts for each candidate as filling in the unused 
diagonal elements of the pairwise matrix. Whereas the off-diagonal 
elements contain the scores for each pairwise race, the diagonal 
elements contain the scores for each candidate vs. the "expected value" 
of the election itself.

As I said before, this method is much simpler than traditional Condorcet 
methods that drop defeats rather than candidates, and I believe it will 
have a much better chance of being accepted for public elections. It may 
still be too complicated, but at least it will have a better chance.

Methods that involve dropping candidates are too complicated and 
arbitrary to be acceptable for major public elections for the 
foreseeable future -- and then some.

--Russ