[EM] Equal Rankings in Real World Voting Systems

[Gervase Lam]

> Date: Mon, 7 Mar 2005 10:17:43 -0800 (PST)
> From: Alex Small
> Subject: [EM] Equal Rankings in Real World Voting Systems

> As to whether equal rankings complicate the Approval cutoff, I prefer to
> just have separate yes/no boxes by each candidate's name (or even a
> separate Approval section of the ballot if it makes ballot layout
> easier).  Otherwise poll workers will have to answer the same question
> over and over:  "When I mark this candidate as the cutoff, does that
> mean that I'm approving him and everyone above him, or that I'm
> approving everybody above him but not him?"  (Actually, that would be
> the most cogent phrasing of the question.  Most people would probably
> phrase it in more confusing ways so that the poll workers aren't sure
> what answer to give.) 

I don't like the idea of yes/no boxes because it would be possible for a 
voter to approve a candidate who is ranked lower than a candidate who is 
disapproved.  May be the ballot paper could be divided into a left side 
and right side.  Any candidates ranked on the left side are approved while 
any candidates ranked on the right side are disapproved.

I can see the motivation of using Approval to break a Condorcet Cycle.  
However, there have been posts to this list describing methods that used 
Ranked ballots in order to derive Approval results.  Why can't such 
methods be used thereby removing the need for voters to mark off an 
Approval cut-off?  Is it the fact that deriving Approval cut-offs from 
rankings would make ranking strategically more likely?

Maybe using (Spruced Up) Gradual Ordinal Information Approval would be 
better.  That is, Forest Simmons "adaptation" of Kevin Venzke's Gradual 
(Cardinal) Information Approval.  Whereas the cardinal version uses the 
mean average Utility of the candidates that are marked Viable in each 
iteration, the ordinal version uses the median average ranking of the 
candidates marked Viable.

Thanks,
Gervase.