[EM] Re: Election-methods digest, Vol 1 #607 - 3 msgs

[Bart Ingles]

Jeffrey O'Neill wrote:
> 
> >Merrill's simulations show plenty.
> 
> To be specific, Merrill's simulations show that IRV is much better than
> PR.  Merrill is fairly supportive of IRV in his book.

I don't recall Merrill saying anything about PR, but "better" is a value
judgement.  In both the Condorcet and Social Utility results, he shows a
strong sensitivity to the policy positions of candidates under IRV. 
When the candidates bunch together relative to the voters' positions,
IRV's Condorcet and S.U. performances both drop even faster than that of
Plurality.

Merrill also concludes that voter strategy is easier under Plurality
than under IRV.  In an obvious two-party election, any voter who cares
about the two-way race will vote for one of the major candidates.  Even
in Florida in 2000, around 80% of Nader's supporters ended up voting for
someone other than Nader.  I'm not willing to concede that either Bush
or Gore was the rightful claimant to the remaining 20 percent.


> Another point is the we have an entrenched two-party system in the US and
> that isn't going to change anytime soon (though I wish it would).

If the two-party system weren't propped up by an anti-competitive voting
system, I believe that this would change.  There is no evidence that it
EVER would under IRV, in which case I would just  as soon stay with
Plurality for partisan elections.


>  Under a
> strong two-party system, most elections will have two strong candidates and
> a few weak candidates.  In such a situation, IRV will generally elect the
> Condorcet winner.

But even when IRV chooses the Condorcet winner, it's apt to be an
artifical one, achieved by eliminating any possible independent
Condorcet winners before the ballots are even printed.  You might as
well outlaw third-party candidates altogether, so that you can guarantee
a majority-favorite winner regardless of the method used.

I no longer view third-party candidates as automatic spoilers.  With a
healthy mix of 3rd-party and independent candidates competing for
dissatisfied voters, neither of the major parties is a net long-term
beneficiary of a "spoiler effect".  Instead, 3rd-party predation will
tend to draw voters away from the more lackluster of the top-two, thus
tipping the balance in favor of the Social Utility winner.

I believe this also serves to put the focus on the relative merits of
the top two candidates, and not just on their party affiliations.  Under
IRV, if everyone dutifully ranks down to one of the top two, all you
will end up with is a referendum on which of the two parties people hate
the least.  In other words, I believe IRV outcomes would more closely
follow voter registration while Plurality would be more influenced by
comparisons of the actual candidates.

Bart