[EM] The Allure of IRV...

[Alex Small]

I was talking to a co-worker about France.  I said this indicates a flaw of
runoff methods.

She doesn't know much about voting methods, so she started saying that
the "successive elimination method" is much better because in France a
whole bunch of liberals divided the vote and kept any liberals out of the
finale, whereas "successive elimination" (her name for IRV) would have put
a liberal in the finale.  Basically, she'd heard somebody describe IRV once
and she thinks it's great.  (In all fairness, I also thought it was great
when I first learned there were alternatives to plurality.)

I tried to explain Condorcet to her.  She kept insisting that "successive
elimination" wouldn't have the problem seen in France.  She said some
friend from Ireland told her that successive elimination elected a third-
party candidate who was more moderate.

Well, we all know that the problem with IRV isn't that in one particular
race it gives a bad result.  IRV can elect a Condorcet candidate, it just
doesn't guarantee the election of a Condorcet candidate (if he exists).
For that matter, Condorcet can give a "bad" result if all of the candidates
suck.  The issue is bigger.

She kept saying that it worked in Ireland in one anecdotal election, so it
must be good.

This is the problem with dissuading people against IRV:  The idea of a
guaranteed majority looks so great, never mind that the majority is very
path-dependent (which leads to non-monotonicity, IIAC, etc.).  It also
looks so great that if our favorite is gone, well, at least we can weigh in
on the remaining candidates.  Face it, folks, IRV _LOOKS_ really good.

I don't know how to break through that psychology for others.  I myself
broke through it because I enjoy math, so I was drawn to investigate
election methods further.  Also, I went to a talk by Saari, and although I
don't agree with him on BC he at least illustrated issues I'd never thought
of.  Even though I don't share his conclusions, I'm glad Saari showed that
the issue is much bigger than simply making sure you get a majority
(however manufactured) in the end.

But, not everybody will attend a talk by an expert who doesn't like IRV.
Not everybody will bother to read up on it.  They just hear that it worked
in this one election, and if your first choice is gone, well, don't worry,
you still have some say.  How to dissuade them?

Don't say "Hitler-Stalin-Washington."  The example is so extreme that it
will be ignored.

Don't say "monotonicity" or "IIAC".  Most people will fall asleep.

I guess the only way to beat the IRV psychology is to reach them before the
IRV people do.  I got started on my Eastern European research by contacting
Steven Brams.  He admits that his knowledge of E. Europe is dated, so the
info posted to the list is likely right.  Well, time to get going faster on
it...

Alex

P.S.  This person is an engineer.

----
For more information about this list (subscribe, unsubscribe, FAQ, etc), 
please see http://www.eskimo.com/~robla/em