The None of the Above Chorus:

[Bart Ingles]

Forest Simmons wrote:
> 
> A more recent example: even a dud like Gore would have received more than
> 50% approval in the last election. Do we want to lower our standards below
> that level?
> 
> Bart, I know that the fallacy in my reasoning is that I am taking the word
> "approval" literally instead of strategically.  I know that there might be
> strategic advantages for approving fewer than half of the candidates in a
> large pool, so that the average candidate would receive less than 50%
> approval.

I wouldn't say it was a fallacy, it's just that I didn't know you were
talking about the concept of approval, and not the method "approval
voting".


> But I think that argument is balanced by the one I gave before, i.e. the
> more students, the greater the potential for at least one passing.
> 
> There is a tension between these two arguments, so let's think about the
> extreme cases:
> 
> We already considered the one extreme of a two way race, and found that a
> 50% cutoff was reasonable there.
> 
> At the other extreme, suppose we have 100 candidates for a single
> position.  Shouldn't at least one of them be outstanding enough to get
> more than 50% approval?  If not, I would say, "What a dismal bunch of
> clown clones!"
> 
> It's like measuring goodness of fit with least squares methods. The
> more degrees of freedom, the better the fit has to be in order to measure
> up. If the error doesn't have a decrease commensurate with the increase in
> degrees of freedom, then the additional explanatory variables are
> considered insignificant and discarded.
> 
> Increasing the number of candidates is like increasing the number of
> degrees of freedom. If an increase in candidates doesn't bring a
> commensurate increase in voter approval, then the additional candidates
> should be discarded, in my opinion.
> 
> But maybe I'm way off base here. I've been assuming that you thought 50%
> was too high. Maybe you meant it was too low.
> 
> Forest


I wouldn't say 50% was necessarily too high or too low, just that it was
completely arbitrary.  It really depends on the makeup of the
electorate.

If the voting public consists of four mutually feuding groups of roughly
equal size, then you might be lucky to find someone acceptable to 35%. 
You could keep disqualifying candidates and re-running the election, but
there is no guarantee that the next time will be better.  You may just
deplete the pool of willing candidates, and end up with someone even
worse than in the first round.

Then again, you may have a homogenous society with plenty of competent,
well educated people willing to serve.  So maybe you have 90 excellent
candidates, and 10 clunkers who drag down the average.  In that case,
the optimal strategy would be to vote for most of the candidates, and a
winner with 70-80% might not be unreasonable.

Maybe one reason I naturally recoil at the 50% figure, is the fact that
the pro-IRV groups (or group) use that as part of their "majority"
argument, which is really based on compound fallacies.  Except when
there are exactly two candidates (or when doing a pairwise comparison),
the 50% figure has no particular significance, other than the fact that
it happens to be a round number.

Bart