[EM] Election-Methods Digest, Vol 235, Issue 30

[Kristofer Munsterhjelm]

On 2024-02-20 09:10, robert bristow-johnson wrote:
> 
> Robert's claim:
> 
> "Cardinal methods demand too much tactical thinking from voters"
> 
> All's I wanted to say is, if there are 3 or more candidates, then 
> Cardinal methods *inherently* require some tactical consideration by the 
> voter the minute they go into the voting booth.
> 
> I really think that voters are partisans and are legitimately 
> differently motivated than Olympic figure skating judges.  We're 
> partisans and the reason we vote is to try to get our preferred 
> candidate elected.  But our political interests may include preventing 
> the election of someone we loathe and realistically fear may get elected.
> 
> The unavoidable question is how much should a voter is how much should I 
> score or approve my sec ond-choice candidate?  If I score them too high 
> (or Approve them), I reduce or lose my vote for my favorite.

Yes. And that fits with the aside I made in the other post, that a 
rank-consistent (Brams' sincere) method can be more prone to strategy or 
have more serious consequences for misfire than a method that isn't.

I think I found a concrete example, even. Durand's paper about 
Condorcification[1] states results that imply that if an election is 
manipulable by some voters under Condorcet//Approval (explicit Approval 
cutoffs), then it's manipulable under plain Approval as well. Yet 
Approval is rank-consistent and Condorcet//Approval, to my knowledge, is 
not.

This seems quite intuitive to me, and I would imagine that there are 
elections where Approval is manipulable and C//A is not, e.g. two good 
candidates and and awful candidate: the better good candidate is the CW, 
but Approval voters have to decide whether to put their cutoff between 
best and good, or good and awful, as you mentioned.

> I don't accept assurances from CES and Clay Shentrup that I should 
> simply vote my sincere preferences and trust the system.

 From my impressions from reading Reddit, Clay's defining characteristic 
is his hostility towards people who don't agree with him. Reddit isn't 
known for its relaxed atmosphere, but I think his original account even 
got banned.

But returning to Approval and Range, one could easily say "*which* 
sincere preference"? If it's vNM sincerity, that's not necessarily 
optimal. If it's Brams sincerity (rank-consistency), choosing the right 
one is what makes it hard.

-km

[1] DURAND, François; MATHIEU, Fabien; NOIRIE, Ludovic. Can a Condorcet 
rule have a low coalitional manipulability?. In: ECAI 2016. IOS Press, 
2016. p. 707-715. 
https://ebooks.iospress.nl/pdf/doi/10.3233/978-1-61499-672-9-707