[EM] Student government - what voting system to recommend?

[Gervase Lam]

> Date: Tue, 24 Apr 2007 20:32:01 -0400
> From: "Tim Hull"
> Subject: Re: [EM] Student government - what voting system to
> 	recommend?

> I know the Condorcet winner is preferred to every other candidate - however,
> I'm in particular assuming ballots like this:
> 
> 48% - 10 D 2 PW 0 R
> 47% - 10 R 2 PW 0 D
> 5% - 10 PW 5 D 0 R
> 
> (the numbers being the sincere range rating for the candidate)
> Under Condorcet, PW would win despite the fact that he or she is barely
> liked by anyone.
> Under range and IRV, D would win.

I suppose this shows that extra expressiveness (i.e. precision) in
voting would be better here to prevent the celebrity Pro-Wrestler (PW)
winning against the Democratic (D) and Republican (R) candidates.
However, I do have concerns with this, whether it be Range voting, Hay
voting, etc...

I don't know whether I got this from this list or somewhere else, but I
heard of cases where voters marked a cardinal ballot in an ordinal
style.  That is, 10 is given to candidate A, 9 to candidate B, 8 to
candidate C, etc...

They did this in a naive attempt to give equal "later-no-harm"
protection to each of the candidates on the ballot.  In other cases, it
was a genuine misunderstanding by the voter (e.g. I saw 10
numbers/spaces on the ballot, so I thought that meant I was supposed to
rank the candidates).

Whatever the cause, each ballot does not reflect the true (cardinal)
opinions of each voter.  Yes, cardinal ballots have better precision,
but it is at a cost of "less accuracy".  I'd rather be up front and just
stick with ordinal voting.  Sure, it isn't as precise, but it is much
more accurate.

One alternative that I thought of that has also been mentioned on the
list is for the strength of preferences in a ranking to be indicated by
voters.  For example, with A>>B>>>C>D, the BC preference is the 1st
strongest, the AB is the 2nd strongest, with CD being 3rd.

I think Mike Ossipoff has mentioned using it in order to
collapse/compress preferences one at a time if there is no Condorcet
winner.  The collapse starts from the weakest preference moving to the
next strongest preference until a Condorcet winner is found.

However, after thinking about this, I also felt this to be susceptible
to voters ranking like A>>>B>>C>D in order to give each candidate
sufficient later-no-harm protection.

Thanks,
Gervase.