[EM] Ranked Pairs tiebreaker scheme

[Kevin Venzke]

Hello,

Is this Ranked Pairs tie-breaking method usable?:

"Suppose A>B and C>D are of equal strength as strength is being defined.

If B is not D and one of B and D defeats the other pairwise, then rank
A>B ahead of C>D if D defeats B, or vice versa if B defeats D.

Else (when B is D, or B and D tie pairwise) rank A>B ahead of C>D if A
defeats C pairwise, and C>D ahead of A>B if C defeats A pairwise."

Maybe this or something like it is already someone's version. MAM uses
Random Ballot by this stage, and Eric Gorr's "Deterministic #1" says that 
"[equivalent] defeats are considered together with previously kept defeats,"
which sounds to me like an equivalent defeat could be discarded even if
it doesn't contradict anything.

Any other approaches?

Kevin Venzke



	

	
		
__________________________________________________________________
Découvrez le nouveau Yahoo! Mail : 250 Mo d'espace de stockage pour vos mails ! 
Créez votre Yahoo! Mail sur http://fr.mail.yahoo.com/