[EM] Majority-Judgement using adjectives versus alphabetical scales versus numerical ranges.

[Jameson Quinn]

2012/12/7 Ted Stern <araucaria.araucana at gmail.com>

> On 07 Dec 2012 08:13:09 -0800, Jameson Quinn wrote:
> >
> > I tend to favor letter grades for MJ. Since the MJ (or CMJ)
> > tiebreaker itself assigns plusses and minuses, you can simply use
> > the letters A,B,C,D,F. That's only 5 categories; if you wanted 6,
> > you could add an explicit "A+" option, because without that the
> > tiebreaker could never assign a + to the highest grade.
>
> Hi Jameson,
>
> Balinski and Laraki make a very clear argument about why Majority
> Judgment should use named grades instead of letters or numbers:  they
> are trying to avoid implicit ranking.
>
> I agree with that argument, as it regards numbers. I think that it may be
true in France for letters, but in the US letters are grades, not ranks;
everyone is used to the idea that two people in the same class can get the
same grade.

Also, the idea is that the effectiveness (ie, social utility, if that were
knowable) of the BR process is best if every voter has the same a prioriti
probability distribution over grades. That is, it's not that you shouldn't
calibrate your grades to the quality of candidates that you typically see,
it's just that you shouldn't recalibrate every single election. I think
that using letter grades will encourage this correct behavior.

Jameson
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20121207/d566c443/attachment-0004.htm>