[EM] Re: RAMON

[Forest Simmons]

I wonder if it would be easier to get monotonicity if we were to draw only 
one ballot and deal with the order of the candidates on that ballot.

How about this:

(After figuring out the pairwise win/lose matrix) draw a ballot at random. 
Let N1 > N2 > ... > Nk, be the order of the candidates on the random 
ballot (with ties broken at random).

For each i start a transitive chain with Ni and go through the other 
candidates in order, passing over the candidates that beat Ni, and passing 
over the candidates that would not fit into the chain without violating 
transitivity.

If no candidate beats all of the candidates in the resulting chain, then 
Ni is the winner, else go on to N(i+1).

In other words, the winner is the highest ranking candidate on the 
randomly drawn ballot that ends up being the head of a maximal chain.

Forest