[EM] my letter to CVD

James Green-Armytage jarmyta at antioch-college.edu
Sun Apr 4 14:35:02 PDT 2004


Adam Tarr <atarr at purdue.edu> writes:
>I think you're being a bit too kind to IRV here.  Say it was a general 
>election in an area with a large progressive base, and the 22% Democrat
>was 
>actually a Green.  If the Green loses first, the Democrat wins 55%-45%,
>but 
>if a few Republicans push the Green over the Democrat by putting him in 
>first place, then the Republicans will win as long as roughly a third of 
>the Democratic voters prefer the Republican to the Green.
>-Adam

	Yes, you're probably right that my wording was just a bit too generous on
that point. I am aware of the sort of example which you're referring to
here. I don't know how often that would actually happen. I tend to assume
that it would not happen often, but I may be wrong about this. 
	It does seem rather counter-intuitive for a Republican>>Democrat>Green
voter to strategically vote Green>Republican>Democrat. I would imagine
that it would take a good deal of convincing to make a Republican believe
that such a vote was in their best interest, unless they were very
sophisticated with voting methods. On the other hand the burying strategy
(which applies to Condorcet but not IRV), strategically voting
Republican>Green>Democrat, to bury the Democrat, seems a little more
obvious and easy to grasp. It is somewhat intuitive to put the closest
rival in last place, to hurt him as much as possible, even if there is a
more obscure candidate who you actually like less. (I sort of imagine that
whatever ranked ballot method is implemented, there will be some
uninformed voters who generally assume that the tally method is something
like Borda.)
	Lack of monotonicity and vulnerability to the push-over strategy isn't
the primary problem in IRV. The problem is lack of Condorcet efficiency
and more frequent incentive for the compromising strategy. But I'm
agreeing that the lack of monotonicity ain't good, of course, and I hope
that my e-mail didn't totally obscure that point. I guess I thought it
would be good to concede some points in order to strengthen others.

my best,
James




More information about the Election-Methods mailing list