[EM] Alphabetical ballot

[matt at tidalwave.net]

On 20 Sep 2002 at 19:49, Elisabeth Varin/Stephane Rouillon wrote:
>  ....
> An example of completed ballot: 
> Shade at Most One Grade per Candidate 
> And Indicate your Limit (|) Between Acceptable and Undesirable Candidates: 
> | 
> Amy Jones AB C D# | F GH I 
> Barbara Smith A BC D E |F G #I 
> Charles Jones #B C DE | F GH I 
> Daniel Smith AB C DE | F GH I 
> Eve JonesA # CD E | FG H I 
> Fibbie Smith AB # DE | F GH I 
> Georges Jones AB C DE | F GH I 
> Harry Smith A# C DE | F GH I 
> Isabel JonesA B CD E | #G H I 
> | 
> that we would translate as: 
> Charles(yes) > Eve(yes) = Harry(yes) > Fibbie(yes) > Amy(yes) > Isabel(no) > Barbara(no) > 
> Daniel(no) ? Georges(no) 

Your translation loses the extra information made available by grading that is useful 
for accurately summing ballots.  There are at least two alternatives: Use a dummy 
place holder candidate (an underline character) or place the grade in the 
paranthesis.  And since we have the approval cutoff (which is indicated by the bar 
after the last approved candidate in the ballot) we may as the well place the 
unranked candidates just after the cutoff in the ballot:
C>E=H>F>_>A|>D?G>_>I>_>B>_
C(A)>E(B)=H(B)>F(C)>A(E)|>D?G>I(F)>B(H)

The place holder method is more compact and easier to read.  Ballots with unlike 
rankings (different grade sets) can now be fairly combined which is my 
understanding of the primary benefit of using grades.  Some tallying methods may 
want to count '=' differently than '?', for example count as '=' 1/2 vote each but '?' as 
zero vote each, so it may be useful even if negative votes is not a good idea.

> I know some would prefer to put unranked Candidates as neutral but 
> how could we attribute them a (yes/no) index? 
> It does not bother me because I treat the no indicator 
> not as a negative vote. Using negative votes is like giving someone 
> the opportunity to deny someone else any representation. 
> In other words a 45 - 23 result is very different and shall not 
> have the same consequences as a 25 - 3 result. It produces 
> the same outcome for single-winner contests, but it should not 
> for fair (proportional) multiple winners methods. 

> Steph.

----
For more information about this list (subscribe, unsubscribe, FAQ, etc), 
please see http://www.eskimo.com/~robla/em