[EM] So I got an email... / IIA

[Kristofer Munsterhjelm]

On 10.04.2022 20:52, robert bristow-johnson wrote:
> But it's *because* of the Condorcet failure that the IIA failed. The
> Condorcet winner (if one exists) must be kept out of the final round. 
> The candidate that loses in the final round and displaces the CW from
> the final round is the spoiler. Then voters for the spoiler who ranked
> the IRV winner lowest are punished for not betraying their favorite.

I agree. In the fully general sense of IIA (that you can remove any
subset of the losing candidates and the winner will still win),
Condorcet methods must fail IIA when there's no CW, as the typical
Arrow's example shows. So must non-Condorcet methods. But non-Condorcet
methods also fail some of the time when there *is* a CW[1]. In that
respect, Condorcet methods are superior.

> But I think Precinct Summability is what's getting the most scrutiny
> here.  We look at Maine who had to haul 600,000 ballots from every
> corner of the state and didn't have an outcome for 4 days.

Is there a chance, then, that Burlington will go directly to a summable
Condorcet method instead of a patch to IRV?

-km

[1] ... because eliminating everybody but the method winner and the CW
leads the CW to win by majority rule.