[EM] Five Slots and Cranor

[Forest Simmons]

Yes, I keep coming back to the five slot method too. It really gives ten
choices if you count AB as half way between A and B, and you count no
grade as below F.  That should be plenty of choices for any single winner
election.

I'm beginning to like Cranor's method which starts with CR ballots, like
the five slot ballot, and automatically calculates everybody's best
Approval strategy based on the CR information.

>From that point there are two options: (1) inform the voters what their
optimum Approval strategies are and let them decide whther or not to vote
them. (2) just assume that the voters want their optimum strategies, and
calculate the winner accordingly.

It seems to me that the second method would be more convenient for the
voters. They just vote their honest grades and let the computer figure out
the winner.

It also seems to me that the method could be refined as follows:

Start with Five Slot Ballots. Let Cranor figure out the optimum Dyadic
Approval strategies. Calculate the winner accordingly.

Voters that want to know the details can study (Dyadic) Approval and
Cranor's method. Otherwise they can be content that the computer is
picking the optimum candidate based on the grades that they assigned, as
they now trust that our current congressional apportionment method (which
not even one citizen in ten thousand understands) is fair. 

Forest

On Tue, 24 Apr 2001, Joe Weinstein wrote:

> Forest,
> 

[...]

> 
> I have appreciated (though do not agree with all contentions in) your 
> subsequent postings to the EM list.  Your analysis of Green psychology was 
> especially notable.
> 
> For now I still find that your 'five-slot' scheme (i.e., comprising 
> approving grades A,B,C and disapproving grades D,F; or something like this 
> scheme, i.e. several approving grades and several disapproving grades) is 
> the best compromise among virtues of simplicity, expressivity, monotonicity, 
> strategic integrity, and feasible sensitivity to Green psychology.  (I 
> continue to find Condorcet methods too convoluted to sell to myself, let 
> alone to the public.)
> 
> Your post even motivated me to concoct and investigate an apparently new 
> virtual runoff method to answer better to the Green psychology. In brief, a 
> voter marks each candidate as fully approved, conditionally approved, or 
> disapproved.  If round #1 gives an approval victory to a candidate you 
> disapprove (e.g. Bush), then each of your 'conditional approval' marks 
> (e.g.Gore) is rescored from disapprove (=0) to approve (=1).  However, this 
> method's instrumental results typically either match straight Approval (of 
> both fully and conditionally approved candidates), or else seem to be 
> irrational departures from it.
> 
> Your five-slot (or similar) mod of Approval is simpler and better.  A Green 
> could have in good conscience voted Nader=A, Gore=C, Bush=F; and thereby 
> helped both to stop Bush and to credibly tell the Dems that next time, if 
> they don't come much closer to Nader, they can expect Dem=D (or Dem=F).
> 
[...]
> 
> Joe