[EM] thread from IRV-Freewheeling

[Bart Ingles]

Eric Gorr wrote:
> http://groups.yahoo.com/group/instantrunoff-freewheeling/message/829
> 
> Some of the more "interesting" comments by James Salsman
> 
> "Because Condorcet voting can more easily be manipulated by strategic
> voting (i.e., marking whichever of the top-two candidates you do not
> want to win dead last after people you like even less), Instant Runoff
> Voting has been shown to produce the Condorcet winner more often in
> practice than the Condorcet method."
> 
> He provides references.
> 
> Has anyone read these preferences? Care to chime in?

The first reference claims to prove that manipulating STV is an 
NP-complete problem.  First, even if true, NP-completeness is not an 
issue when there are only 3 competitive candidates.  Second, 
"manipulation" under STV is frequently desirable, if it helps to bypass 
an irrelevant alternative or prevents a voter from falling prey to a 
monotonicity violation.  Third, to the extent that STV helps to 
perpetuate a two-party duopoly, manipulation is not a concern.

In the fourth reference, Salsman is probably referring to a statement by 
Merrill to the effect that, in the face of strategic voting, Approval 
Voting is the method most likely to elect a Condorcet candidate, 
followed by IRV, and then Condorcet & the other methods.  The Condorcet 
method used by Merrill used Borda as a completion method, so it may well 
have been more manipulable.  I don't know whether it is possible to 
choose a completion method that would reliably punish (or at least not 
reward) burying strategy, in the event that burying resulted in a cycle. 
  Maybe using an anti-democratic completion method-- such as electing 
the candidate with the _worst_ Borda score-- would be necessary to 
effectively punish burying that could result in a cycle.