The Allure of IRV

[Forest Simmons]

I want to take a shot at The Allure of IRV.

If I understand correctly from various postings on this list over a period
of time, Australian voters routinely vote "above the line."

To me that means that they consider STV to be too much effort.  The ranked
ballots are a chore for the average voter.  IRV has its allure for voting
activists and reformers, but not for typical voters, at least not after
the shine wears off.

Typical voters just want anything that ameliorates the spoiler problem and
doesn't cost a lot in taxes or voting effort.

Suppose that we stick to the traditional one bit (two level) ballot, so
that existing equipment can handle the count, in an effort to keep the
cost down.  Then Approval is the obvious choice, even though it requires
the voter to come up with an "approval cutoff" or otherwise strain the
brain on how to vote.

Suppose that all voters decided to approve their plurality choice (the
one they would have voted for under lone mark plurality) and anybody they
considered better.  This would be as easy as voting under plurality, and
would yield vastly superior results on average.

Another option for one bit ballots is for voters to use the ballots to
designate candidates as proxies, and then let the candidates vote on
behalf of their respective supporters in a more complicated style "runoff"
or election completion, whether by Condorcet (of whatever stripe) or
Approval.

The Australian situation, where voting "above the line" is the rule rather
than the exception, has (in effect) degenerated into a defective version
of proxy runoff, where you can designate a party as proxy, but not a
candidate, and where the runoff is by STV instead of Condorcet or
Approval.

It's like paying for an Edsel and ending up with a Nova.

Another possibility is an approval/proxy hybrid: all ballots are deemed
Approval ballots except that bullet votes are considered proxy
designations, etc.

Now, if we go to ballots more complicated than one bit per candidate, then
we need really great voter appeal to justify it.

A three level ballot might have significantly greater appeal than a two
level ballot if used as follows:

If no candidate gets more than fifty percent of the top votes, then the
candidate with the fewest bottom level votes wins.

Essentially it is Approval with an expressive feature that wouldn't make
any difference in practice.

Also, it could be considered Bucklin done right.

If it isn't instrumentally superior to Approval and it is more costly in
ballot type, counting equipment, etc. then why consider it?

It would be worth the extra cost if it appealed to the voters more,
because of their desire to show clear favoritism towards their favorites.

>From my experience, that is the main allure of IRV over Approval among
those who have been exposed to both but still prefer IRV; they want a
method that allows them to exalt their favorite strictly above their
compromise, whether or not they understand that IRV can and will penalize
them for this luxury, from time to time.

Who would object to "Bucklin Done Right" if they were told ... "Put two
check marks next to your favorite's name, and one mark next to the name of
each acceptable alternative.  If nobody's favorite gets a majority, then
the candidate acceptable to the greatest number of voters is declared
winner."?

What do you think?

Forest


----
For more information about this list (subscribe, unsubscribe, FAQ, etc), 
please see http://www.eskimo.com/~robla/em