[EM] Corruption & IRV, slight revision.

[MIKE OSSIPOFF]

EM list--

It might seem as if a weak point in my discussion of IRV &
corruption is where I brushed-off the possibility that the
A position voters might vote B in 2nd place instead of A,
when they put Bc in 1st place. It could even seem as if I
tried to dismiss that possibility because I'm biased against
IRV, and that I was just trying to find a way to say that IRV
fails.

Actually, that possibility didn't occur to me till I was already
writing that letter. I figured it was too late by that time,
forgetting that I could save the letter as a draft, till I
look more closely at the problem.

But, for one thing, I think my answer in that letter had validity:
An A voter who votes Bc 1st and B 2nd is voting his favorite in
last place. Unless we have reason to believe otherwise, it
certainly isn't certain that they'd do that. The example is about
people making Bc win by upranking him to 1st place, and it's
not unreasonable to say that they don't completely reverse all
their preferences too. Contrived? All it takes is to contrive one
plausible failure example, to be able to say that the method
fails. Frontrunner probabilities that encourage skipping in Approval
are not plausible. People not reversing all their preferences
and voting their favorite last isn't implausible. So it's a fair
failure example.

Sure, a person might figure that, because B is in middle policy
position, they might cover both bases by voting B 2nd, since
they want to compromise anyway. If you want to compromise, protect
both middle candidates, including the better one. Reasonable, but
not certain; voters aren't always reasonable.

But in fact, there could be some broadcast or printed advice
for A voters to vote B 2nd, and in general, for people to
rank an uncorrupt counterpart of a corrupt candidate 2nd when
they feel compelled to rank a corrupt one 1st. It could be pointed
out that that can get rid of voters' fear of individually
withdrawing their support for the corrupt candidate.

Looked at that way, IRV, in this special case where we want
to defeat a corrupt candidate over whom another candidate is
unanimously preferred, is significantly better than Plurality--
but still not as good as Approval, which doesn't let the electorate
get stuck on Bc, and doesn't need any strategy sophistication to
avoid that problem.  Approval is better than IRV in that way,
just as in every way that means anything.

But one more thing: IRVies can't have it both ways. They boast
that IRV is simple, and free of strategy. But only strategic
information-sharing and organizing saves IRV from failure in
that corruption-encouragement test.

Mike Ossipoff
______________________________________________________
Get Your Private, Free Email at http://www.hotmail.com