[EM] Condorcet criterias

[Stephane Rouillon]

I compared margin, relative margin, winning votes and relative winning votes.
If we assume votes are either in Favor, Against or Without opinion those criteria
are defined like that:
Total votes = F + A + W (a more detailed case would consider Equal votes)
margin = F - A
relative margin = F - A / (F + A)
winning votes = F
relative winning votes = F / (F + A)

I found that only winning votes is able to garantee victory to a strong Condorcet
winner
(which received more votes in favor when compared to every other opponent)
despite
truncation strategy.
However relative margin seems the most fair criteria if we know all ballots are
sincere.

Steph.

Jobst Heitzig a écrit :

> Dear Adam!
>
> > You may be right that in this case (due to unequal "votes" in different
> > contests and the lack of opportunity for strategy) margins or even relative
> > margins could be a better measure here than just winning "votes".
> >
> > -Adam
>
> I also think that *relative* margins would be the best measure of defeat
> strength here!
>
> > By Ranked Pairs, Boston and Anaheim tie.
>
> Also River reports a tie between Anaheim and Boston when using winning
> "votes".
>
> Jobst
>
> ----
> Election-methods mailing list - see http://electorama.com/em for list info