Ken Johnson wrote: > > I was, until recently, a fanatical advocate of Approval. I tried to > demonstrate by empirical simulation the superiority of Approval over > rank methods, based on the criterion that the election method should > maximize "social utility" as defined by sincere CR. (See my earlier post > in Election-methods digest, Vol 1 #597, Message 4.) Unfortunately, it > didn't turn out the way I expected. In a single-issue election with many > candidates, Approval exhibited abysmal performance, worse than all other > methods (including Plurality). So I gave up my crusade. Oh well.
I have no idea how to locate "Vol 1 #597", so I don't know what Ken means by "exhibited abysmal performance". Was this a series of simulations? A single example? In Merrill's 'Monte Carlo' simulations, approval voting did pretty well in terms of social utility-- better than all other methods in the three-candidate case, and only marginally worse than Condorcet and Borda with four or more candidates. Bart ---- Election-methods mailing list - see http://electorama.com/em for list info
