[EM] Fw: IBCM, Tideman, Schulze

[MIKE OSSIPOFF]

Markus wrote:

>You pretend that I claim that I have discovered that Mike's
>1998 Tideman GMC example was impossible. Actually, you wrote in
>your 13 May 2000 mail that Mike's example was impossible. But
>Mike replied (13 May 2000) that such an example _is_ possible.
>Therefore I have sent a new proof of the impossibility of such
>an example in my 19 May 2000 mail.

That statement on 13 May must have been before I found out
that Bruce Anderson hadn't said what I'd thought that he'd said.

But didn't Blake demonstrate that any pairwise votes table
can be made consistent with some set of rankings by merely adding
the same undetermined constant to each entry in the table?  If
so, then that means that any set of pairwise defeat magnitudes
is useful, if all we're interested in is the differences in
those magnitudes, not in their actual values. That particular
example was about majorities, and it depended on the actual
values of the defeat magnitudes, and so it wasn't a valid
example of what it was attempting to show. But it's still valid
for the differences in the defeat magnitudes.


>You wrote (3 Jun 2000):
> > Markus also nastily replied that I should have let Mike speak
> > for himself about GMC and Beatpath GMC.
>
>The problem is that you always refer to Mike although Mike
>disagrees with you in so many points.

I initially liked GMC as something that underlies other criteria
that I like. But BC serves that role better. And Beatpath GMC
is too specific. That stronger criterion would be justified, for
me, if its compliance guaranteed the compliances that BC guarantees,
and if either Beatpath GMC guarantees compliance with additional
important criteria, or if Beatpath GMC is, of itself, directly
compelling. I don't believe that either of those last 2 requirements
is met by Beatpath GMC. Beatpath GMC just narrows its compliance
down to one method, with no convincing reason for doing so.

What I've heard on this list about Tideman(wv)/MTM & Schulze
seems to all point the same way: Criteria, winner comparisons,
decisiveness. Tideman seems to have Schulze's method beat.

But Schulze's method does have one thing that counts in its favor:

In public elections it chooses the same winner as SD & SSD, which
, when they undefeat a candidate, are undefeating the candidate
who can be elected while overruling as few opposing votes as
possible, consistent with BC compliance. But Schulze's method,
as I was saying before, doesn't have the obvious & natural
motivation & justification that SD, SSD, & Tideman have--something
that's important for a public proposal.

Besides, if the standard of overruling as few voters as possible
were more important than all other criteria & standards, then we'd
be advocating PC. And the people's expressed choice between
the Tideman winner & the Schulze winner may be more compelling
than a general reckoning of which winner has more voters
voting against him--for the purposes of choosing between Tideman
& Schulze.

***

Mike Ossipoff


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