When is Simmons going to settle on one good method? Wake Me Up When It's Over :-)
I am going to try to use vivid language to make this new method visually memorable:
Let's call the alternatives that are pairwise beaten or tied by X, its tamed set T(X), and the complement of this set, it's wild set W(X). So the complete set of alternatives is the disjoint union of T(X) and W(X), no matter which alternative is represented by X.
An epsilon neighborhood of a set K is the set of all points that are no farther than epsilon away from K.
Our new method is called "Wild Tamer:"
Elect the alternative X who's wild set is contained in the smallest epsilon neighborhood of its tame set.
This method has all the nice properties of any of my previous metric based methods along with a tightening of the ship so to speak.
[epsilon neighborhood? What do you expect from a topologist?]