[EM] Condorcet Flavored PR Methods

[Forest Simmons]

On Fri, 27 Sep 2002, Alex Small wrote:

> Forest Simmons said:
> > If the method yields a "beats all" combination, then that combination
> > is by definition better than any other combination according to
> > whatever standard of "better PR representation" is being used in item
> > (2) above.
>
>
> I guess that formulating a criterion for "better PR representation" will
> be addressed in a subsequent post.

Yes, in fact various possible criteria will be addressed.

> The other major question that I see is
> resolving cyclic ambiguities, although the stakes aren't as high as in
> single-winner races because all of the M-member lists in the innermost
> unbeaten set will presumably satisfy a large number of voters.
>

Yes, once you have the pairwise matrix you can process it according to the
rules of SSD or RP, etc.  The processor doesn't have to know whether the
matrix came from a single winner election or a multiwinner election or
whether the election is supposed to be PR or not.  All of that is taken
care of before submitting the matrix to be processed.

> Question:  As interesting as it is to contemplate PR via Condorcet or
> Approval, what are the odds of the public accepting these methods?

Not good for large scale public elections, but remember that those are not
our only applications.  There are already software development
organizations that use SSD(wv) as their official voting method, for
example. One of my students is a member of such a society.


Joe Weinstein has opened up great vistas for election methods WITHIN
representative bodies that have themselves been chosen by lot.

In any case, a good Condorcet Flavored PR Method would be one (among many)
standards for comparing results among various methods being proposed for
public consumption or for theoretical purposes.


----
For more information about this list (subscribe, unsubscribe, FAQ, etc), 
please see http://www.eskimo.com/~robla/em