>Well, for my money, in an election with multiple candidates, Condorcet
>voting is the way to go. Each voter ranks each candidate in order from
>best to worst, and there is a simple algorithm to determine the winner.
>
>
http://www.electionmethods.org/
>
>There's even a site that will run a Condorcet election for you, without
>anyone here having to understand or implement the underlying algorithm. I
>forget the URL, but IIRC it's linked from the above site.
Ugh. I think that if a real election were ever held on that basis, the
results would be widespread dissent. It's far too likely to elect a
candidate who was nobody's first choice. The theory is that if say 45% of
the electorate vote for A, 40% vote for B and 15% vote for C, but A's
supporters prefer C to B, and B's prefer C to A, then you should elect C as
a compromise candidate. But only 15% of the electorate wanted C, so 85% are
dissatisfied.
Pete