[EM] CR/Approval method, RWE

[Forest Simmons]

This version of Chris's could be called "runoff done right."

It preserves the idea of not "lowering your vote" until lack of viability
forces it upon you, but it has a better measure of viability than ordinary
runoff, and it doesn't make "irrevocable decisions based on partial
information" (a nice phrase I first heard from Mike Ossipoff).

We could also get a method called "Coombs done right" by setting the
approval cutoffs just above the least preferred viable candidate on the
ballot at each stage.

Kevin's method is a compromise between "runoff done right" and "Coombs
done right."

Another compromise suited to ranked ballots would be to set the cutoff at
the median viable candidate (instead of Kevin's mean CR candidate) at each
stage.

Forest


On Wed, 1 Oct 2003, Chris Benham wrote:

> I  like Kevin 's  "another  CR/Approval  method", and  I think it would
> be good if  the name contained a reference to the
> type of  ballot  used. Maybe something like  "Venzke Ratings"?
> It has occurred to me that the method  is similar in concept to  "Runoff
>  Without Elimination". This quickly led me to
> realise that RWE  could be simply improved  by instead of candidates
> being permanently "shunted", we adopt Kevin's
> mechanism of  marking candidates (maybe temporarily) "not viable", and
> also we use Kevin's stopping rule (when all
> but one candidate is marked "not viable").
>
> "Inproved" RWE  (Venzke CR/Approval  adapted  for plain ranked ballots):
>
> Ranked ballots, equal preferences and truncation ok.
> First iteration: Count the first preference votes  and  mark the
> candidate with the lowest tally as "not viable".
> Reset the tallies to zero.
> Second iteration: Each ballot contributes a vote each (no split votes)
> to their favourite candidate/s of those not marked
> "not viable", and also to any marked "not viable" for whom  they show
> equal or higher preference .Now mark the two
> candidates with the lowest tallies as "not viable", and reset all
> tallies to zero.
> And so on, each time marking as many candidates "not viable" as there
> have been iterations(inclusive). The tag expires if
> it is not renewed at the next iteration. When all but one candidate is
> marked as "not viable" , then that candidate wins.
>
> Chris Benham
>
>
> ----
> Election-methods mailing list - see http://electorama.com/em for list info
>