[EM] Matthijs van Duin criterion

Warren Smith wds at math.temple.edu
Tue Mar 6 17:47:31 PST 2007

>Criterion:  Adding a candidate who is pairwise defeated by the current 
winner may not cause a different winner to be chosen.
>wds: are there any pure rank-order methods that obey this criterion?

Equivalent criterion: removing a candidate pairwise defeated by current winner,
cannot alter the winner.

So in an A>B>C>A cycle, in which (say) A wins, removing B cannot make C win.
However, C must win for any method reducing in the 2-candidate case to simple
majority vote (since C>A pairwise).

I conclude that Matthijs van Duin's criterion is failed by every reasonable
voting method based on pure rank order votes.

He says DMC satisfies the criterion.  I may be confused, but I don't think so.

(please correct me if I'm confused.)

More information about the Election-Methods mailing list