[EM] Problems with finding the probable best governor

[MIKE OSSIPOFF]

So if that compromise wins, nothing wrong with
> > that, right? If he were really between Clinton & Bush.

>
>That kind of compromise wouldn't satisfy me.  But then I still suspect
>that a utility-maximizing voter would have incentive to truncate, so it
>may not be a major concern.

I often won't like it either, but the problem is that so many
voters are addicted to it. Condorcet & Approval let them
indulge their addiction to compromise without abandoning candidates
who are better.

Sure they'll truncate, not vote for Middle, if Middle seems likely
to be a dishonest opportunist. That's why I'm not concerned about
the problem. Also, there will be plenty of selection, with genuine
good candidates near the voter median.

>From an individual voter's perspective, it would make no difference
>whether he was compromising by choosing someone farther away on a
>bipolar or monopolar scale.

Sure, so if, by positional & nonpositional considerations, Middle
is worse than the opposite extreme, then you won't vote for Middle.

>That's something that I would like to try, if I ever get time to learn
>Python.  It would be interesting to see, both with and without a
>nonpositional utility component.

Yes, and also how about a simulation that compares Tideman(m)
and Tideman(wv) when strategic truncation is used. For each
candidates & voters configuration tested:

1. Run the simulation with everyone voting sincerely. If there's
no CW, discard that election & go to the next candidates & voters
configuration.

2. If there's a CW, then run the election again, with the same
candidates, but with everyone voting only for candidates whom they
like better from the CW (except for the voters who voted the CW in
1st place--their ballots don't change).

3. For every voter who likes the result of that 2nd simulation
better than the CW, leave his ballot truncated. Otherwise un-truncate
it. Run the simulation one more time.

4. The result of that 3rd simulation is the final election result.

***

I already know that Tideman(m) will do considerably worse than
Tideman(wv), but it would be interesting to find out how much
worse.

***

I'd suggest using 1-D, and 3-D with strong correlation (.9?).

***

Mike Ossipoff



________________________________________________________________________
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com