Instructions for how to make a method look bad:
Contrive an example in which the main contending candidates barely differ
by the method's own standard, but in which those candidates differ humungously
and outrageously
by some other standard. In order to achieve the latter condition as strongly as
possible, it's typically
necessary that the candidates are within one vote of eachother, in terms of the
method's standard.
Then say, "Look how wrongly that method can choose!" Try to sound especially
outraged when you say that,
as some here are already practiced at doing.
And yes, it's true: In Kevin's MMPO bad-example, A, B, and C do almost
identically, in terms of MMPO's
own standard. No choice would be significantly worse than another in terms of
that standard.
But, in terms of the favoriteness standard, they differ drastically and
dramatically.
Has someone followed my above-supplied instructions? Sure. To the letter.
So, because the candidates don't significantly differ by MMPO's standard, but
differ outrageously
by the favoriteness standard, guess which standard we notice? Yes, intuitively,
you look at that
and say that A or B should win. The winner should come from [A,B ]. In other
words, if A doesn't
win, then B should win. If B doesn't win, then A should win.
You know it. I know it.
Problem: The voters don't think so. (Remember them?)
Why don't the voters think so? Because that's necessary in order to create the
favoriteness-outrageous outcome
of Kevin's MMPO bad-example.
Early on in the discussion of that example, I asked who was wronged in that
bad-example.
Someone answered that the [A,B ] voters as a whole, were collectively wronged.
But, as I discussed in my previous post about this, the A voters couldn't care
less whether B or C wins. Therefore,
it's a bit creative to say that they're wronged because B didn't win instead of
C. C won precisely because and only
because the A voters didn't care about B vs C, and the B voters didn't care
about A vs C.
In fact, strictly speaking, if you took a poll among the A voters, between B
and C, C would win that poll.
It's obviously fallacious to speak of [A,B] as a "person" who has been
wronged--a resort of desperation needed
because no one can point to a particular individual who was wronged.
Mike Ossipoff
----
Election-Methods mailing list - see http://electorama.com/em for list info
