[EM] possible improved IRV method

[Jonathan Lundell]

At 1:22 PM -0400 6/27/06, Dave Ketchum wrote:
>Strategic voting seems to me to get called worse names than it deserves:
>       Some makes sense, to select among and vote for the best combination
>of acceptability and electability.
>       Some is less deserving - to, somehow KNOW how others will vote, and
>have your clan defeat their will with exotic voting.  How do you KNOW
>this, and pass the word to your clan, without other clans learning and
>changing their votes to respond to yours?

There needn't be any such word-passing; in some very common cases, 
all that is needed is standard publicly-reported polling. If an 
election is closely contested between two major candidates, it 
requires no real coordination for supporters to bury the leading 
competitor.

I prefer the term "insincere voting" to denote any deviation from the 
sincere ranked preferences of the voter. Of course we want to elect 
the "sincere Condorcet winner", if there is one. But if we employ a 
voting method that encourages insincere voting, then there's no 
assurance that we're doing that.

I'm a true believer in neither IRV nor Condorcet voting. Each has 
very different, really incommensurable, flaws, and to confidently 
proclaim one or the other as technically superior is to miss the 
point.

IRV's problem with ignoring later preferences and not always finding 
a Condorcet winner is a direct consequence of the way it avoids 
Condorcet's problem with encouraging insincere voting. Which flaw is 
more serious is a matter for subjective judgement, and not objective 
determination. Informed IRV advocates are not fools--at least no more 
so than informed Condorcet advocates.

Strong advocates of either system tend to minimize the flaws of the 
system they support. IRV advocates generally claim--and may well be 
right--that examples in which IRV falls on its face tend to be highly 
artificial constructions, unlikely to occur in a real election. 
Condorcet advocates generally claim (often implicitly) that voters 
will cast sincere ballots regardless of the counting system. Neither 
claim is provable a priori.


-- 
/Jonathan Lundell.