Six criteria

[Mike Ositoff]

I said last week that, when using a flexible point assignment
method (sometimes called a "cardinal measure method"), voters
would probably use Approval strategy, but that better maximization
of a voter's utility expectation could be gotten by calculating
different point values to give the different candidates. I was
mistaken. Authors demonstrate that, in a flexible point system,
a voter maximizes his utility expectation by voting exactly
as in Approval, giving the highest point score to the alternatives
that he'd vote for in Approval, and giving the lowest point
score (even if it's negative) to those he wouldn't vote for
in Approval.

Earlier, Steve & Mike S. were debating whether voters would
vote as in Approval, using Saari's point system. The answer is
they would, if they want to maximize their utility expectation.

It's obvious from today's strategic voting behavior that voters
do want to optimize the outcome, and only a few prefer to vote
sincerely.

If Mike S. insists that there'd be no incentive to strategize,
then I ask him what a results-oriented voter would do in the
following situation:

There's a Green, a Democrat, & a Republican. The voter is a Green.
His sincere ratings (utility ratings) of the candidates:

Green: 100
Republican: 0
Democrat: 30

But he knows that there isn't even one chance in 1000 that the
Green will win. So what's this pragmatic voter going to do?
Is he going to give the Democrat 30 points? :-)

He'll do just what an Approval voter would: Give the maximum
point score to the Green & the Democrat, and the minimum to
the Republican. Even if the minimum is negative. (So if it's
-100 to 100, then the Republican gets -100 from that voter).

Mike