[EM] Poll Ballots, from poll-topics poll

[MIKE OSSIPOFF]

>It's as if a sliding scale is needed: Approve my favorite of the two 
>leading
>candidates, plus all candidates higher than that choice, and extend the
>selection downward by some degree depending on the closeness of the
>two leading candidates. Don't just extend downward for exact ties, in
>other words. Do so for near ties, as well.

Yes, that's a worthwhile refinement to pursue for future polls. As
you suggested, it would be too big a new issue to apply it to the
Voting Systems poll. It would be a thorough overhaul-refinement of
Voter's Choice, which now only looks at _winners_ by the designated methods.

Myerson & Weber, in their article about voting equilibrium,
discussed a relation between past scores and current tie-probabilities,
it seems to me. Lorrie Cranor's method & Hoffman's method also estimate
tie probabilities from past vote totals. Cranor's & Hoffman's methods
can be found at Cranor's website, available from Rob L.'s Condorcet
website, reachable from http://www.eskimo.com/~robla. I'll look for
the Myerson & Weber article, and if it has anything relevent to this
issue, I'll post where to find the article.

Actually, Voter's Choice is rather similar to Cranor's Declared Stragegy
Voting, except that DSV repeats the count, and the stratgey
calculations based on each count, till the result converges, and
she uses Plurality instead of Approval as the base method, and uses
Pluralilty only, rather than letting people designate other methods
to guide the placement of their vote.

But if Cranor's or Hoffman's method were to be used with few voters,
then there'd be the question of how to adapt those methods to few voters.

Mike Ossipoff


_________________________________________________________________
Get your FREE download of MSN Explorer at http://explorer.msn.com