[EM] Markus: Simulation question

MIKE OSSIPOFF nkklrp at hotmail.com
Thu Dec 26 23:39:34 PST 2002


Markus wrote:

Dear Mike,

you wrote (17 Dec 2002):
>IRV's poor social utility scores in simulations resulted
>from the frequent elimination of median candidates,
>probably happening largely as described above.

you wrote (24 Dec 2002):
>As I said, this kind of thing is the reason why IRV
>did so poorly in social utility simulations.

At least Merrill's simulations say that IRV does it
better than Top-2 Runoff.

I reply:

...better at what? Social utility or Condorcet efficiency, or both?

I'm not denying it, just asking.

But I didn't say that simulations showed IRV doing worse than
Runoff.

You continued:

Which simulations do you
use?

I reply:

Merrill.

The fact that simulations show IRV's compromise-dumping doesn't mean
that I'm claiming that Runoff beats IRV in the simulations. That's
another issue.

I don't have Merrill's book nearby, and so I don't know if IRV
beat Runoff in SU. You said in your earlier posting that IRV is
more Condorcet-efficient than Runoff, and I assume that that is
based on simulations.

If IRV's SU is better than that of Runoff, that means something for
the comparison between IRV & Runoff. But I just want to mention that
IRV did significantly worse than Approval and Condorcet in Merrill's social 
utility simulations.

But, about the Condorcet-efficiency, one obvious fact is that
, as I said, Runoff always elects the CW if s/he is comes in 1st or
2nd as people's indicated favorite, while IRV can fail to elect a CW
even if s/he is the indicated favorite of the most people. That's a
guarantee, and it isn't obvious whether Condorcet efficiency in
simulations is as important as that guarantee.

If IRV's Condorcet efficiency is better than that of Runoof, it
must be that IRV is electing more CWs who come in 3rd or lower in
favoriteness. But those are just the kind of CWs that IRVists don't
want to win! Is IRV biting its own master? Worse for popular CWs,
while helping less popular CWs? By IRVists' own standards, Runoff is doing 
better by CWs than IRV is.

Mike Ossipoff










Markus Schulze







_________________________________________________________________
Add photos to your e-mail with MSN 8. Get 3 months FREE*. 
http://join.msn.com/?page=features/featuredemail&xAPID=42&PS=47575&PI=7324&DI=7474&SU= 
http://www.hotmail.msn.com/cgi-bin/getmsg&HL=1216hotmailtaglines_addphotos_3mf

----
For more information about this list (subscribe, unsubscribe, FAQ, etc), 
please see http://www.eskimo.com/~robla/em



More information about the Election-Methods mailing list