[EM] Simulation results (Approval, utility, Schulze efficiency)

[wclark at xoom.org]

Kevin Venzke wrote:

> It's surely a fluke that "Two Evils" outperforms "Zero-Info" here.  I
> have to doubt that random information could be better than none at all.

I wonder if it might make sense to think of the random information as a
signal that can be productively exploited to organize cooperative blocs of
voters.

I'm reminded of some of Robert Axelrod's work in game theory, whereby
letting agents in a "Prisoner's Dilemma" simulation use an otherwise
meaningless marker or tag, the agents were able to self-organize
cooperative groups that could sustain consistently higher scores in the
game.  Basically, agents could use the marker to identify in-groups of
other agents that they knew they could trust, and cooperate to their
mutual benefit.

Or, think of the story of the indecisive donkey that starved to death,
because he was exactly halfway between two equally large piles of apples
-- and thus had no way of choosing one pile over the other to eat. 
Sometimes a little randomness is precisely what a system needs, to break
some form of symmetry that's holding it back.

-Bill

-- 
Ralph Nader for US President in 2004
http://votenader.org/