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

[raphfrk at netscape.net]

From: Jan Kok <jan.kok.5y at gmail.com>

> A couple weeks ago Forest Simmons suggested the candidate withdrawal
> option as a way of improving most election methods. I just realized
> that the candidate withdrawal option would greatly mitigate the center
> squeeze problem with IRV. (Sometimes it takes a while for other
> people's ideas to sink into my thick skull :-)

Yeah, it might also help to shift the result towards condorcet, if the
candidates withdraw in alignment with their supporters.

You could have people saying "I will withdraw if XXX is going to be
eliminated".

> Maybe I could propose this to the Denver group, if they are determined
> to go ahead with IRV.

> So, what should the rules be for candidate withdrawal with IRV? Should
> candidates be allowed to withdraw at any time (including after the IRV
> count is done, causing the IRV count to be restarted from the
> beginning, or restarted at a particular step), or is there a need for
> restrictions?

I would probably do it after each step.  If nobody is elected, the
election official would ask if any candidate wishes to withdraw.  If
no candidate does, then the bottom candidate is eliminated as per 
normal.

One issue with withdrawl is that it is one area when candidates and 
voters
have misaligned interests.  The voters don't really care about the
candidates, they care about the issues.  They only care about the
candidates because they trust some more than others.  However, getting
elected or not, does matter to the candidate.  This means that a 
candidate
would be willing to risk a poor candidate being elected with 40%
probability over a certainity of not being elected if the candidate
withdraws.

Also, when you start dealing with fewer people in a democracy, fairness
and power start exhibiting "non-linearities".  If you have a council 
with
members having vote weights of 2, 2, 1, all 3 members are equally 
powerful
as any 2 can force a majority.

I read an interesting discussion of the electoral college on

http://www.madore.org/~david/misc/us-voting.html

This shows that as you increase the weight, the power increases with
the square of the weight.