[EM] extending Myerson's test--more policy positions

[MIKE OSSIPOFF]

>
>I wonder what strategy is optimal under a pairwise system?  Should you
>refuse to rank candidates whom you rate below a certain number?  If your
>favorite is one of two known front-runners with a 50-50 chance of
>defeating the other front-runner, it would seem advantageous to refuse
>to rank anyone whom you would rate below the middle of the scale.
>
>Does this sound valid?  If so, is there a corresponding strategy for
>candidates you would rate above a certain level, e.g. should you rank
>some of them as equal to your favorite?  This seems less likely, but I
>wonder if anyone has calculated the optimal strategies for
>utility-maximizing voters.  I'm ignoring the possibility of using
>order-reversal here for simplicity.

That's probably a reasonable assumption. Due to the Condorcet
versions' compliance with SFC, and, better yet, the Cycle
Condorcet versions' compliance with GSFC, if no one votes
a false preference, then you need no strategy at all. You can
sincerely rank all of the candidates.

Voting a short ranking in Condorcet is a defense against
offensive order-reversal. Probably won't be needed. When
introducing Condorcet, I don't talk about offensive order-reversal
or the defense against it.

***

But I'd probably limit my ranking anyway, just in case.
One strategy would be to not rank anyone you like less than
the most likely median candidate or CW. Actually, if that likely
median candidate isn't your favorite, and if we're allowed to
rank as many as we want to (so I can rank above someone all
the ballot-candidates & registered write-ins that I like better
than him), then I'd feel free to rank the low candidates too,
or most of them, & might do so.

Your strategy of not ranking anyone worse than the middle of
the 2 most likely winners is similar to my suggestion to not
rank people below the median or the CW.

But you never have reason, with Condorcet, to rank anyone else
equal to your favorite. With Copeland, or Margins? Then you
might have to.

I haven't dealt with mathematical strategy for Condorcet, just
these rule-of-thumb ideas, because strategy is so much less
needed in Condorcet.

***

As for strategy for general pairwise-count methods, other
than Condorcet--let's just hope Condorcet gets enacted instead.

But the general pairwise defense against truncation is to
rank the likely CW or median candidate equal to your favorite
along with everyone you like better than him). The defense against
order-reversal is to rank him alone in 1st place.

If that candidate is your favorite, you could insincerely
rank the side less likely to truncate or order-reverse over
the side more likely to. Or if it could come from either side,
then you could announce that your cohort will vote probabilistically
so as to make it impossible to predict which side will beat
the other side. Or of course you could also publicize your
sincere voting intentions with regard to the 2 sides of
your favorite, if you're sure that the side that you like better
doesn't have voters who will take advantage.

You say that's a mess? Of course it is. That's why Condorcet
is the only pairwise method that's any good. As for rank methods,
Bucklin is ok, but not as good as Condorcet.

Mike Ossipoff


Mike Ossipoff


______________________________________________________
Get Your Private, Free Email at http://www.hotmail.com