[EM] Citation for immunity to strategic voting?

[Andrew Myers]

Date: Mon, 5 Sep 2005 18:51:40 -0400
From: Andrew Myers <andru at cs.cornell.edu>
To: Stephane Rouillon <stephane.rouillon at sympatico.ca>
Subject: Re: [EM] Citation for immunity to strategic voting?

On Mon, Sep 05, 2005 at 05:55:01PM -0400, Stephane Rouillon wrote:
> Actually as many people will tell you,
> this claim is wrong.
> 
> I see that Rob already gave you a counter example.
> 
> Maybe you would like to know that using winning vote as
> criteria to make pairwise comparison instead of margins
> can make your claim true for strong Condorcet winners
> (ones which have a more than 50% majority against every
> other candidate).  Using margin as a criteria your claim is only valid
> for stronger Condorcet winners (having a 2/3 majority against
> every other candidate).
> 
> Finally, no method is know to garantee the election of a weak
> Condorcet winner against unsincere preferences. This
> is understandable because absentees can always alter the balance
> against the Condorcet winner and hope to unsincerely create
> a cycle containing one of their better choice.
> 
> Hope it helps,
> Steph.

That's very helpful and makes perfect sense. I guess we could guarantee no weak
CW by requiring that voters order all candidates, but this might be seen to be
onerous. On the other hand, there will probably be a set of "plausible"
candidates and if voters know they should make sure to rank all of them it
would help create a strong CW. One could also imagine employing a runoff
election mechanism in the case of a top cycle (as Juho suggested) where
additionally voters were required to give a total order on all candidates to
defend against strategic voting.

Thanks,

-- Andrew