[EM] Majority

[Kevin Venzke]

Hi Colin/Forest,

Colin Champion wrote:
> I'm confused... I thought Forest was providing a philosophical argument
> in favour of minimax, but I can't read Kevin's reply as expressing
> philosophical doubts over that method.

To me minimax is more of a framework than a method, so I would not have so many
doubts about that, but rather about the specific treatment I (somewhat incorrectly)
understood Forest to be suggesting.

It looks like Forest's intended method was what Woodall called "MinGS."

Indeed this does not violate Plurality. It also satisfies LNHelp, mono-raise, and
mono-add-top.

But not only is it not a Condorcet method, it doesn't ensure mutual majority even
with three candidates. So I think the spoiler concern is still relevant.

Kevin


> On 29/05/2022 18:07, Kevin Venzke wrote:
> > For me, it would be necessary to "lock" the actually voted majorities in some way,
> > so that the thought experiment majorities don't override them and refute what they
> > stated.
> >
> > Otherwise I think it's too spoiler-prone: If you're barely on the losing end of a
> > majority, but the majority was burdened with an additional candidate, then you
> > are more likely to win. You just need to squeak past the majority, while they need
> > far more additional votes to beat each other.
> >
> > I would maybe form the CDTT (i.e. the Schwartz set resulting from replacing
> > non-majorities with ties) and only allow the thought experiment's winner to be a
> > candidate from the CDTT.
> >
> > But this would still violate Plurality (and certainly lacks elegance).