[EM] RE: Let's found an organization to oppose IRV

[Bart Ingles]

"Donald E. Davison" wrote:
> 
> Greetings List,
> 
> Craig Layton wrote:
> >I continue to be perturbed by arguments that this voting system or that
> >voting system elects centrist candidates, compromise candidates, etc.
> >Voting systems should be judged on how well they reflect the will of the
> >people.  It is quite often the case that centrist candidates represent no
> >one.
> 
> Donald: They, centrist candidates, make fair `Caretaker' officials.
> 
> >What voters would do with an approval vote is very difficult, if not
> >impossible, to predict.  The support is close enough so that both the Hitler
> >and Stalin voters know that if they approve Washington, Washington will
> >almost certainly win, but if they only vote for their own candidates, there
> >is a very good chance that they will win.  I also think that it's fairly
> >safe to assume that Hitler and Stalin voters will be fairly parochial.
> >Voting for only your own candidate in such a close race might be the optimal
> >strategy.
> >
> >Craig Layton
> 
> Donald: Craig is correct. Smart voters will only make one choice when faced
> with a method like Approval or Borda  or that other guy, what's him name.
>        That will result in turning these weird methods into Plurality(FPTP).
>        `The more things change the more they remain the the same.'
>        T.J. Max Depo-rep-man should take another look at Instant Run-Off - Ha Ha
>        Now there is a good single winner method, right under your nose.


So you are saying that with Approval, most Nader supporters will
continue to vote only for Gore, as they did in the most recent election?


When reasonably good polling info is available, FPTP isn't completely
without merit, as far as its ability to select candidates.  It does have
the property that for each vote cast, you know that the candidate voted
for is either the voter's favorite, or at least considered good enough
by the voter to be worth abandoning his favorite.


With runoff elections or IRV, since there is no real disincentive for
ranking all candidates, you can easily get a winner whom very few people
actually like.  With five candidates, it's possible to have a winner
supported by only 6.25% of the population, and despised by the other
93.75% (vs. 25/75 for FPTP or Approval).

A big problem is IRV/Runoff's insistence on electing a majority winner,
whether true majority support exists or not.  You can't infer from
preference order alone (other than first or last choice) whether a voter
actually supports or despises a particular candidate.  IRV/Runoff can
only guarantee a majority winner by making unwarranted assumptions about
the meaning of voters' second and subsequent choices.  When no candidate
has true majority support, the runoff election ends up counting
"throw-away" votes toward a fictitious majority.

This doesn't happen with FPTP because the giving your vote to a
compromise candidate is so costly (you can't vote for your favorite). 
Approval voting is in between the two extremes, in that the cost of
compromise is less than for FPTP, but still enough to keep a voter from
giving out votes willy-nilly.


As for Approval electing centrist "caretaker candidates", consider the
following:

        Candidate(rating)
45%    A(10)    B(1)   C(0)
30%    B(10)    C(1)   A(0)
25%    C(10)    B(1)   A(0)

Here B is the centrist, and not a very good one -- 70% of the voters
despise him.  Yet he wins under IRV.  Under Approval, few voters are
likely to vote for more than one candidate, so A would win with the
strongest plurality.