Wow. I'd forgotten that the new voters would change what
constitutes a majority. Bucklin indeed violates No-Show.
And now, before someone else tells me, I must admit that the
Smith//Condorcet(EM) with subcycle rule 2 violates Pareto,
and GMC. And, which is the same thing, it violates my
basic democratic principle that says "If a majority rank
A over B, then if we choose A or B, it should be A".
I've never understood quite what's so important about
Pareto. With a million voters, does it really matter if
someone is beaten, pairwise, 999999 to 1 or 1000000 to zero?
To me, violating Pareto just shows how badly something can
violate GMC.
I don't know how important those violations are when we're
going from one cycle-order to the next. I suppose it
depends on how badly we want what that method offers.
One could argue that if one member of a subcycle wins with
respect to the larger cycle, then even a subcycle member
who is badly beaten from the larger cycle could still be
considered in some sense at least an equal of everyone in
the subcycle, especially if he wins within the subcycle.
Mike
