[EM] Problems with finding the probable best governor

[Bart Ingles]

MIKE OSSIPOFF wrote:

> Bart Ingles wrote:
> >Ratings and probability estimates are just a way of putting what
> >you know in one place -- you ultimately have to decide how to weigh the
> >information.  And the 'lottery' part mostly just consists of identifying
> >frontrunners.
> 
> You mean that the probably part mostly just consists of
> identifying the frontrunners. Maybe now, but, for one thing,
> I don't know if, even right now, it would still seem that way
> if Approval were being used instead of Plurality. Plurality
> greatly exaggerates, maybe outright manufactures, the appearance
> of some certain 2 media-favored candidates being frontrunners.
> And it would be especially less likely to be that way after
> a better voting system ended the 2-party illusion, with the
> result that more kinds of candidates could run, and more
> running candidates would be considered viable. So, that
> statement that the probability judgement is mostly a matter of
> identifying 2 frontrunners is likely to only apply while Plurality
> is in use. Later it might not be so obvious who the frontrunners
> will be, and frontrunner probabilities will be what we have.

There could be more or fewer than 2 frontrunners.  I would expect very
large numbers to be increasingly unlikely though, even with good
election methods.

So in a 10-candidate field, if you can identify 3 or 4 who are
front-runners (meaning they all have a reasonable shot at winning), you
still narrow the possible strategies considerably.


> [...]
> The reason why I argue against these arguments is that IRVists
> are using them against Condorcet. If the middle candidate is
> so mediocre, so little better than the worst, that just means
> (to me) that those people need to run and vote for someone better.

 
If the IRVists are using these arguments, then they are on shaky
ground.  IRV's SUE is very mediocre (even Plurality is likely better
when polling info is available), and its worst-case SUE is very small
for elections with more than a few candidates (proportional to 2^(1-k)
where k = the number of candidates).  That's why I no longer support
IRV.

It would be no problem at all to nuke an IRV supporter attempting to use
social utility to argue in favor of IRV.

Bart