[EM] The Repoman strikes again

[LAYTON Craig]

Bart Ingles wrote:

>I think Merrill's study was really showing IRV's vote-splitting problem,
>in which having several candidates clustered together will tend to
>benefit candidates who are outside of the cluster.  The studies assumed
>normal distributions of candidates, which of course places the cluster
>near the center, thereby favoring the more extreme candidates to either
>side.
>
>When he reduced the standard deviation (tightened up the cluster) for
>candidates but not for voters, the effect was intensified.  The
>conclusion was that the candidates who moved toward the center were in
>effect punished for doing so.

I suppose there are two ways of expressing why this shows that a condorcet
system is superior to an IRV system.  The first is that IRV unfairly rewards
some candidates, and punishes others, leading to a situation where two
candidates with relatively strong primary support can squeeze out a group of
candidates, despite the fact that that group (or any candidate in it) is
really more popular.  Basically, the results can be erratic and
counter-intuitive.  The second is to say that IRV elects extreme candidates
and that is bad because extreme candidates are bad.  I really prefer the
first style of reasoning to the second.

Although I expressed some concerns with social utility arguments in the
past, I do now think that they are valuable in demonstrating which methods
return the candidates that the people are most happy with, at least as far
as their expressed preferences are genuine.  This is the true measure of a
democratic system.  What I have more difficulty with is assumptions about
voting patterns without any actual evidence of people voting in this
fashion.  It is difficult to get accurate information about full preferences
in our elections (the main information available being how votes are
distributed when a candidate is eliminated), but it appears that voting
patterns are more erratic than people might think.  What I have seen doesn't
suggest that votes flow smoothly from 'exreme' candidates towards centrist
candidates.  I can imagine cases where preferences from extreme candidates
might go to the other extreme rather than to the middle.  The most recent
Russian presidential elections showed some indication that in the period
between that election and the previous two, a block of voters moved from
(excuse the spelling) Zhirinovsky, the extreme right candidate, to Zuganov,
the Communist candidate.