[EM] Spoiler Effect on Wikipedia

[Paul Crowley]

On Fri, 12 Nov 2004 23:25:55 -0500, Eric Gorr <eric at ericgorr.net> wrote:
> Well, use the one that Dr. Tideman proposes.
> I would assume that it would fall into the resolute model.

Isn't the "resolute model" just, well, wrong?  No entirely
deterministic method can satisfy universality, anonymity, and
neutrality.  Since we nearly always want our election methods to
satisfy all these three criteria, we have to sacrifice determinism,
and thus in order to be precise we have to phrase our criteria in
terms of probabilities.  Schulze's restatement of IIA and ICC seem
like the best and most natural way to do that. And so he's right, IIA
does not imply ICC.

I guess you could sacrifice universality instead, but then you have a
hard time stating your criteria, because you have to handle the case
where an election method fails to produce an answer altogether.
-- 
  __
\/ o\ Paul Crowley
/\__/ www.ciphergoth.org