Spread spectrum analysis: IR vs Condorcet

[Steve Eppley]

   cc: John Taplin

Bruce A wrote:
>Steve asked if that book discussed ranked voting methods.
-snip-
>I also suspect that such a definitive reference does not currently
>exist because, as both Steve and my Oct. 26 message imply, much
>remains to be done in this area. 

Are you familiar with the references cited by Gary Cox in the
message I forwarded to EM a few days ago?

Ordeshook's book had a bit about candidate strategies under
plurality, enough to get me thinking about Instant Runoff's
problems.  In another message I'm posting today (Re: hippopotamus...)
I discuss a one-dimensional analysis of IR given 3 candidates and a
bell-shaped distribution of voters, and show how IR fails to elect
the center candidate in this most basic of scenarios if the voters
vote sincerely.  Which runs contrary to Rob Ritchie's contention 
that IR's failure scenarios would be rare.

John Taplin wrote that his analysis of IR (a.k.a. single-winner STV)
data from Australia shows that 80% of the winners were 1st choices of
majorities, and that only about 6% of the winners appeared to be
undesirable deviants from majority rule.  But his analysis didn't
take into consideration candidate and voter strategies.  I don't know
how one could begin to take candidate strategies into account, but
perhaps voters' strategies could be deduced by also examining the
pre-election polling data to which the voters had access.  Then it
might be shown that some of John's 80% were probably not the sincere 
first choices of majorities.

---Steve     (Steve Eppley    seppley at alumni.caltech.edu)