[EM] Approval Voting vs Instant Runoff Voting

[Forest Simmons]

On Wed, 14 Feb 2001, LAYTON Craig wrote:

> Mike wrote:
> 
> >In Approval, people will vote only for their favorite if they think
> >they don't need their 2nd choice, because their favorite has a win.
> >Actually many who now voted only for Nader, because that was the only
> >way to express support for him, would vote for Nader & Gore in Approval.
> >So fewer, not more, people would vote only for their favorite.
>
(Craig wrote) 
> Hold on.  You only recently finished telling us that Nader could win, if
> only you had a different electoral system.  So, which is it?

Forest: Suppose the system were CW, and the preferences were as follows

40%  Gore>Nader>Bush>Buchanan
20%  Nader>Gore>Bush>Buchanan
10%  Bush>Nader>Buchanan>Gore
20%  Bush>Buchanan>Nader>Gore
5%   Bush>Gore>Buchanan>Nader
5%   Buchanan>Bush>Nader>Gore

Nader beats Gore 55% to 45% .
Nader beats Bush 60% to 40%
Nader beats Buchanan 70% to 30%

I have two reasons for giving Gore and Bush only a total of 75% of the
first preferences:

(1) Under Condorcet there is no incentive to vote insincerely (in contrast
to the recent November election). 

(2) Under Condorcet, there would be more voter turnout.  In November half
the eligible voters stayed home because it was a foregone conclusion that
nobody with their interest in mind had a chance of winning (under
plurality).  The Nader/LaDuke Super Rallies, where upwards of ten thousand
paid ten to twenty dollars admission at each rally, showed that Nader had
a lot more grass roots support than either of the big corporate supported
candidates.

I've given Gore the benefit of a doubt here.  Almost nobody in my circle
of acquaintances (job, church, neighborhood, relatives, inlaws, etc.) 
actually preferred Gore over Nader, but because the media did not portray
Nader as a viable candidate, most of these acquaintances voted for either
Gore or Bush. Among them Gore was the spoiler, not Nader. 

If the preferences were indeed as I have hypothesized (or approximately
so) then Nader would have (and should have) won under CW.

Note well that this last statement is of the form "If hypothesis is true,
then conclusion is true." You cannot logically challenge this statement by
contradicting the hypothesis. If IRVies (or others) start defending a
different hypothesis (and they are perfectly free to do), they are not
challenging this statement, they are just starting a new scenario. I hope
they won't pretend it is the same scenario with a different outcome. 

The original question was (implicitly) "Could Nader have won under a
better system?"  I think my example shows that he could have.  To
challenge the sufficiency of this example (not the statement we were
talking about in the previous paragraph) you would have to show either (1)
that CW is not a better system, or (2) that nothing close to the
hypothesized preference scenario is plausible. 

Of course, even if my example is not sufficient to prove the affirmative
nature of the answer to the question, that doesn't mean that there isn't a
more accurate scenario with (perhaps) an even better method under which
Nader would have won, however far that may stretch the brittle imagination
of some of our readers. (Not you Craig)

How about under IRV? If the voters expressed these same preferences, then
Gore would have won. More likely the voters would have voted insincerely,
shrinking the Nader and Buchanan first place preferences, with Gore and
Bush neck and neck as in the actual November election.

Under Approval, with these preferences the election would in all
likelihood go to either Nader or Gore depending on the strength of the
various preferences and the accuracy and availability of information from
the polls.

>  ...... 
> largely random.  See my example of opinion poll ratings & expected utility
> outcomes and try to decide how to cast your Approval vote.
> 

I would like to see that example.  You must have submitted it before my
time.

Forest