[EM] Real IRV Election, Disputable Result

[Jonathan Lundell]

At 4:46 PM -0500 3/15/06, RLSuter at aol.com wrote:
>So my questions are, given the many different possible numbers
>and varieties of candidates in different elections, combined with
>the varying concerns different voters will have about different
>candidates, how often are voters likely to use a burying strategy,
>and how easy will it usually be -- even for voters who really excel
>at strategic thinking (undoubtedly a very small percentage) -- to
>decide that burying is a good strategy and how deep a particular
>candidate should be buried? In short, how big a factor is strategic
>burying ever really likely to be?

My conjecture (based on extrapolation from smallish ranked-preference 
elections that I've helped run) is that 1) you're entirely right that 
most voters aren't going to be very good at strategic thinking, but 
that 2) they're going to engage in (bad) strategic thinking, and 
voting, anyway.

I recommend spending some time at the blackjack tables for a 
confirmation of this kind of thing. (Learn basic blackjack odds, 
which give you an expected return of about 98%, so the process 
doesn't get too expensive.) It seems like every player has a system, 
and most of them (even most of them that get published in how-to 
books) are deeply silly.

What we want, it seems to me, is to be able to persuade the voter to 
vote sincerely, and to do so honestly. The problem with burying is 
that it's a both simple and relatively obvious, and in particular 
elections might become widespread.

The problem is that claims like "almost all real-world IRV election 
will choose the Condorcet winner", or "voters in real-world Condorcet 
elections won't vote strategically" are close to impossible to 
demonstrate to someone who's skeptical of one claim or the other--or 
both.

I worry more about insincere/strategic voting because I'm convinced 
that most people are prone to act that way, especially in emotional 
high-stakes elections, but if you don't share my worry, I doubt that 
I can convince you. I tend to worry less about IRV because I think 
the IRV winner is generally justifiable even in the (apparently 
unlikely) event that s/he isn't the Condorcet winner.

I'm no doubt biased, in that my interests lie more in multiple-winner 
STV elections, where the Condorcet winner doesn't figure.

Also, I see either IRV or Condorcet methods as so much better than 
the alternatives that we shouldn't overstate the defects of either 
one, compared to the much greater defects of (especially) simple 
plurality elections.
-- 
/Jonathan Lundell.