[EM] Why I think IRV isn't a serious alternative

[Aaron Armitage]

--- On Tue, 11/25/08, Jonathan Lundell <jlundell at pobox.com> wrote:

> From: Jonathan Lundell <jlundell at pobox.com>
> Subject: Re: [EM] Why I think IRV isn't a serious alternative
> To: stepjak at yahoo.fr
> Cc: election-methods at electorama.com
> Date: Tuesday, November 25, 2008, 3:20 PM
> On Nov 25, 2008, at 1:07 PM, Kevin Venzke wrote:
> 
> > Hi Jonathan,
> > 
> > --- En date de : Mar 25.11.08, Jonathan Lundell
> <jlundell at pobox.com> a écrit :
> >> It's not an unreasonable
> >> conjecture that Bayrou would have gotten a larger
> percentage
> >> of first choices (some from Sarkozy and Royal)
> under IRV.
> > 
> > Could you explain this viewpoint?
> 
> The Wikipedia article doesn't give us the actual poll
> questions, but I take them to be something like "which
> candidate do you intend to vote for". So the answers
> need to be interpreted in the context of the first round of
> a TTR election in which polling suggests that the
> front-runners are Royal and Sarkozy. The actual results of
> the first round are consistent with that interpretation.
> 
> In that context, how many voters whose sincere first choice
> was some other candidate (including Bayrou) were
> strategically voting for Sarkozy or Royal? We simply
> don't know, and I merely assert that we can't assume
> that the votes in the first round of a TTR election are
> identical to the first choices in an IRV election.
> ----
> Election-Methods mailing list - see
> http://electorama.com/em for list info

My understanding is that Bayrou was positioned "between" Sarkozy and
Royal, which makes it highly unlikely that his supporters were
compromising by supporting one of the other candidates or claiming to rank
them higher when asked by polls. This is because of how candidates
position themselves. If the spectrum runs from left to right, with the
center-most possible candidate marked C, like this:

---------C---------

Then the main center-left and center-right candidates want to be as close
to C as possible, or even occupy that position, because they are
guaranteed the support of everyone further out than them who cares about
strategy, and are guaranteed all their transfer votes in the final round
of counting. The most rational positioning for them is:

--------ACB--------

Or, even better:

--------AB---------

Or:

---------AB--------

Even though voters are distributed along a bell curve rather than evenly,
there are very few people who will sincerely rank C first. An IRV
supporter has to either say what you said, that somehow behavioral changes
encouraged by IRV will make it work out, without any explanation of what
would induce people to vote for a candidate further away from them rather
than one closer except raw strategy of the kind IRV supposedly frustrates,
or else denounce C as a turkey and declare that the hypothetical
possibility than a candidate C could exist who had campaigned by laying
low or avoiding issues invalidates the majority C>A and C>B preferences,
without even knowing how this particular C campaigned.

------D-ACBE-------

Now what? Under Condorcet, we know. C still wins. Unless the bell curve is
very steep, it appears E would win under IRV. Why is this better?