[EM] candidate withdrawal & IRV - what should the rules be?

[David Cary]

--- "Simmons, Forest" <simmonfo at up.edu> wrote:

>  I like the idea of giving candidates the option of withdrawal
> before each potential elimination.
>  
> It seems true that most candidates would be more reluctant to
> withdraw than their supporters might want, but if no candidate
> withdraws, then IRV with withdrawal reduces to plain IRV, no harm
> done.  It's a cost free, harmless option that could do some good.
>  

It is not quite so clear that post-voting withdrawal under IRV in
general, or specifically withdrawal at each elimination step is cost
free or without disadvantages.

For withdrawal at each step:
 -- It lengthens the time to run a vote count.
 -- It requires a candidate to make a withdrawal decision while
withholding important data / info:  what are the actual votes and how
will a candidate or an opponent's withdrawal effect future steps. 
That sets the stage for critical misjudgements.

As has been illustrated here recently, in a large field, the order of
early elimination can effect who the IRV winner is, so this is not
necessarily a game just for the major candidates.

Some general issues about withdrawal under IRV include:
 -- It gives the candidates an opportunity to negotiate the winner of
the election.
 -- To the extent that withdrawal is effectively a game in which even
when the payoffs of an outcome and all the votes are known:
    -- Determining an optimum strategy can be a non-trivial task.
    -- Optimum strategies may result in cyclical stalemates.
 -- It encourages candidates to seek recounts, just so they can have
a second chance at playing the withdrawal game.

Some may still see withdrawal as an improvement to IRV.  If so,
setting rules to minimize its disadvantages might be worthwhile.

An alternative would be to offer a one-time withdrawal opportunity: 
1) determine the winner without withdrawal and publish the results
and votes, 2) Give all candidates an opportunity to withdraw, 3)
Determine the winner with all withdrawing candidates eliminated at
the beginning.  It seems to me that would reduce some of the
complexity, while still preseving many of the benefits.  

A way of dealing with cyclical stalemates could be to give candidates
about a week to declare their withdrawal, but stagger the declaration
deadlines so that candidates that were eliminated earlier in the no
withdrawal IRV have to declare withdrawal earlier.

__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com