[EM] Hare (aka IRV) versus STAR

[robert bristow-johnson]

On 04/13/2024 3:55 PM EDT Closed Limelike Curves <closed.limelike.curves at gmail.com> wrote:
> 
> 
> Wait, 45% of the ballots in this election violate the strictly-single-peaked assumption. Center seems to be just as polarizing/hated as Left and Right.

The Left voters are opposed to both Center and Right.

The Right voters are opposed to both Center and Left.

How is that surprising?  This is the classic scenario where IRV demonstrates the Center Squeeze effect.  Get the numbers from Table 2 in this paper and scale them down to 100 voters: https://drive.google.com/file/d/1jIhFQfEoxSdyRz5SqEjZotbVDx4xshwM/view 

> Under these ballots, "center" isn't a centrist candidate, he's some kind of third position. (Maybe you could get this result if you had a fascist/populist running against extremely polarized conservatives and liberals? You need some kind of circular ideological spectrum for this election to happen with sincere ballots.)
> 

The Left candidate is Vermont Progressive Party.
The Right candidate is Republican.
The Center candidate is Democrat.  As Figure 1 shows, he's really the Center.  He gets far more 2nd-choice votes from either Left or Right voters than the candidates on the Left and Right get from voters on the opposite wing.

More info here: https://en.wikipedia.org/wiki/2012_Burlington_mayoral_election 

> So I have no intuition for who should win, because this isn't a single-dimensional election.

Nope.  Classic Center Squeeze.  It is not a cycle like the Minneapolis Ward 2 election in 2021 or the Oakland School Board District 4 election in 2022.  It's a straight Left - Center - Right distribution with a very few Left and Right crossover 2nd-choice votes.

> If the ballots are perfectly honest, your "Right" candidate should win because he's the optimal-utility candidate.

That's assuming utility is the correct ethic for elections.  I would take issue with that.  Utilitarianism is the correct ethic for distribution of resources.  But valuing our votes *equally* is the only correct ethic for elections (which is what Condorcet is committed to).  People have died in the United States for the value of valuing our votes equally.

> The single-peaked preference assumption is violated too badly for Condorcet to be a good guide here.
> 

The "single-peaked preference assumption" is what sane voters, that understand how STAR works, would do with their preferences.  This is in my other response from a couple minutes ago.  It's up to you to demonstrate otherwise.

--

r b-j . _ . _ . _ . _ rbj at audioimagination.com

"Imagination is more important than knowledge."

.
.
.