[EM] Strategies for RRV/RSV and BR for multi-member constituencies

[Raph Frank]

On Wed, Jan 28, 2009 at 8:39 PM, Kristofer Munsterhjelm
<km-elmet at broadpark.no> wrote:
> The "construct the assembly and then let it vote" approach might be
> reducible to ordinary Bayesian Regret. The idea would be this: single-winner
> BR assigns utilities to all candidates and voters. Call the candidates'
> utilities their "defined utilities".

You would need one for each dimension that is being tested.

However, a assembly made up of clones would also vote the same way as
the people, if they were purely centrist.

This is kinda the same point I made in my post.  Would a voter prefer
an assembly with higher average utility of the candidates, or one
where he has 1 candidate who is an exact clone of him.