[EM] Simmons' recursively defined second-place method

[Warren Smith]

>Elect the alternative X with the least pairwise opposition from the alternative X' that
>would win if the winning alternative X were withdrawn.

It is not clear to me that this method is useable to compute
the winner in polynomial time. The full tree of recursions could have
N! branches if N candidates, and might have to be explored.  Or
2^N candidate subsets might have to be considered.

The definition is short but the work might be very large.

Also,  even if the number of voters V is very large (so that ties are
unlikely) it might be that N! is even larger (consider N=20) in which case it is not obvious 
that you won't get killed by ties buried somewhere inside 
the massive tree of recursive
exploration...         
wds