[EM] Condorcet methods - should the cycle order always determine the result order?

Kristofer Munsterhjelm km_elmet at t-online.de
Tue Nov 4 08:49:41 PST 2014


On 11/04/2014 04:38 PM, Toby Pereira wrote:

> With Schulze, I thought its beatpath method was transitive. I wasn't
> thinking of removing the winner and recalculating. Maybe I'm wrong about
> that. But yes, River obviously only affirms the pairwise results that it
> needs to in order to get a winner, so certainly can't be said to produce
> an overall ranking in all cases.

Beatpath is indeed transitive.

As for Ranked Pairs, it passes local IIA, which means that if you remove 
the winner from every ballot, whoever came in second (by the ordering 
given by Ranked Pairs) will win, and that if you eliminate the loser, 
whoever came next to last will be the new loser. So it doesn't matter 
whether you go by its ordering or by sequential elimination.

River doesn't give a full ordering, but to my knowledge, there are 
extensions of it that do (see
http://wiki.electorama.com/wiki/River ). I don't remember if River 
passes local IIA.


More information about the Election-Methods mailing list