[EM] Call for Ideas on Automatic Approval Cutoff Finding

[Forest Simmons]

On Sat, 13 Sep 2003, [iso-8859-1] Kevin Venzke wrote:

... [skipping past many great insights to this proposal] ...

>
> When there are cycles, I think there's necessarily a game of chicken going
> on among the voters.  I think an approval measurement is the best way to
> resolve the game, forcing voters to pick sides without "perfect" polling info.
> A defect is that there's the possibility that not all voters give approval
> votes with relevance to the cycle...  For instance, if someone disapproved
> all cycle participants, or none of them, then he doesn't participate in
> resolving the cycle.
>
> Maybe CR is a good resolution to this defect?  Once we have a cycle, all
> cycle participants have similar odds (don't they?), so it's a trivial task to
> convert a CR ballot to an approval ballot: Approve everyone preferred to the
> average CR score of cycle participants.
>

Excellent idea!

In connection with this idea keep in mind that there are several methods
that generate cycles, and this idea could be used to resolve a cycle from
any of these methods.

In particular, Cumulative Repeated Approval Balloting (CRAB), exhibits
cyclic behavior in some cases.  Since some of the candidates appear more
frequently (as leaders) in the "cycle"  than others, their empirical
frequencies could be used as probabilities in the expected CR calculation.

Forest