[EM] Re: public acceptability

[Forest Simmons]

From: "James Green-Armytage" <jarmyta at antioch-college.edu>
Subject: [EM] I forgot something important...


James wrote...

 	I forgot to mention something important before I sent my last post, "CWO
may be worth fighting for". I wrote:

>	Here is one possible progression for single winner elections (to decide
>on representatives):
>1. plurality and runoffs
>2. IRV
>3. CWO-IRV
>4. ranked pairs(wv), with CWO
>5. cardinal pairwise (with CWO?)

James went on to say that he hoped that Direct Democracy could furnish a 
shortcut to Condorcet methods.

Forest replies ...


I have a shortcut to Direct Democracy that meets all of Russ's simplicity 
criteria:



[Start Method Description]

Voters distinguish "favorite" and "also approved" candidates, as in 
Majority Choice Approval.

While two or more candidates remain in competition, all candidates (acting 
as proxy electors for their direct supporters) eliminate (by majority 
vicarious vote) one of the two remaining candidates with lowest approval.


[End Method Description]


Explanation:

This is TAWS (total approval winner stays) with the voters directly 
supplying the approval information, and the ordinal information coming 
from the candidates that they directly support.

In the USA we already use electors.  But because of the severe 
discretization error inherent in the Electoral College, these electors 
represent the voters in only crude proportion at best.

["Best" means those states that do not require their electors to vote as a 
block.]

Furthermore, since the voters usually don't even know their electors, they 
don't really have any reason for trusting them as representatives.

Wouldn't you rather have your favorite candidate as an elector?

The proxy weight of each candidate is the number of ballots on which s/he 
was designated as favorite.

To minimize the possibility of spoiled ballots, we allow voters to 
designate more than one favorite. If a ballot has k candidates designated 
as favorite, then that ballot contributes only 1/k to each of their proxy 
weights, i.e. favorites are counted "cumulatively" for proxy purposes.


There are lots of possible variations.

For example, instead of basing the method on TAWS, we could sort the 
approval seeded list using sink sort, bubble sort, or some other way.

My favorite sort in this context:

While some candidate (pairwise) beats an adjacent candidate with more 
approval, swap the members of the adjacent pair with the least approval 
difference.

This sort has a nice reverse symmetry to it: reverse the approval order 
and the pairwise beats, and the final order is reversed, as well.

Forest