IRV inconsistency

Forest Simmons fsimmons at pcc.edu
Fri May 18 12:55:00 PDT 2001



On Thu, 17 May 2001, Buddha Buck wrote:

<snip>

> 
> 3) Ask the voters to rank a mythical "None-Of-The-Below" (NOTB) candidate 
> as well as the real candidates.  Ballot: A>B>C>NOTB>D>E
> 

I like the NOTB option, which is equivalent to my MAC (Minimum Acceptable
Candidate) proposal, but perhaps less confusing to the voters.

Still, the redundant Yes/No check boxes next to each name on the ballot
might make the most sense to some voters, since the notion of a
fictitious cutoff reference candidate might seem too far out to some
voters.

Demorep has resolutely stuck to his proposal as long as I have been
watching the EM list, and the longer I think about it, the higher it gets
in my ranking of practical methods in terms of likelihood of good results.

Here's how I would modify it:

Use simple Grade ballots (A through F) with A,B,and C considered above
NOTB/MAC/YES/NO/Pass/Fail cutoff and D, F, and blank considered below.

If no candidate has more passing than failing grades, then .... (whatever
consequences appropriate for the type of election in question).

If there is a clear "beats all" pairwise winner, great!

Otherwise, take the candidate with the greatest number of passing grades.

If the voters could understand the following modification and accept it, I
would insert the following in the pairwise comparison part:

In pairwise comparison of candidates X and Y, those ballots that have both
candidates on the same side of the pass/fail cutoff are to be given only
half as much weight as those ballots that assign them to opposite sides of
the cutoff. 

I believe that using the grade ballots only for pass/fail and pairwise
order information might partially satisfy Demorep's (and other's) concern
about grades and rates having no measureable units that have standard
meaning to all of the voters. 

The extra simplicity and familiarity of the grade ballot compared to the
ranked ballot might tip the balance in favor of its use, even though it
wouldn't have the order resolution of Demorep's ranked/Yes/No ballot.

Forest



More information about the Election-Methods mailing list