[EM] Dave on approval, ranked ballots

[Abd ulRahman Lomax]

At 10:01 AM 7/27/2005, Kevin Venzke wrote:
>Dave,
>
> > Please give an example, but:
> >       No IRV - let that be a separate project.
> >       No cycles - likewise, unless you state that there is no problem
> > without cycles being involved.
>
>When no cycles are involved, Condorcet is 100% perfect. So I have to use
>examples with cycles:

Kevin indicated later approval of Approval, so this is a technical comment:

Condorcet is perfect, absent cycles, if one believes that the winner of a 
majority of pairwise contests should be the winner of the election. 
However, if the winner of the election should be that candidate who is most 
widely approved, and if voters vote sincerely and are willing to stretch 
themselves a little (i.e., to approve a candidate other than their 
favorite), Approval does a better job, and this is not merely in some 
artificial, unlikely situation.

It's a question of election philosophy.

Condorcet:

99 A>B>C

98 C>B>A

Assuming no truncation, A Wins, winning the pairwise contests of A vs. B 
and A vs. C, by a vote of 99 to 98.

In Approval, B wins in what would be called a landslide under present 
conditions, assuming that the voters all approve their top two candidates.

A: 99
B: 197
C: 98

I'd say that in this situation, the Approval outcome is far more likely to 
produce effective government, more general satisfaction with the outcome, 
than Condorcet.

"Perfect" implies "without flaw." I'd call the failure to elect a 
compromise candidate a flaw. But certainly many people do believe that the 
pairwise victory is sufficient; excepting cycles, Condorcet is perfect if 
pairwise justice is the criterion.

It might be interesting to put the pairwise victor in an actual runoff with 
the Approval victor. This would oppose A and B.

It is my view that, while there is a problem with the dictatorship of the 
majority, the minority should never be allowed to impose its views on the 
majority.

If a majority of people actually prefer that the (say) President be 
approved by only 99/197 of the voters, compared to one acceptable by nearly 
*all* voters, if even a few of the A supporters would prefer to see their 
favorite lose than to have him win, ratifying such a polarized choice, the 
victor would become B. (And I assume, indeed, that B supporters -- i.e., 
the Cs in the example -- would not change their votes. Preferring the 
choice of a mere majority over a compromise candidate would be odd for a 
voter in the minority.)

This question is different from preference as to candidates, it is 
preference as to how the society functions. The runoff, with the results 
from the first race, would make that clear.