[EM] Tideman and GMC

[MIKE OSSIPOFF]

Replying to Markus:


>Please read my mails more carefully. I wrote (29 Jan 2000):
> > Ossipoff's GMC presumes that there usually is a SACW. There is -to my
> > opinion- no justification for this presumption. That's why I

I no longer use GMC. BC underlies & generalizes the defensive
strategy criteria.

I realize that there won't always be a CW, but when there
is, then his majority defeats of other candidates are
especially important to uphold. I completely agree that
it's important to have criteria that are useful even when there's
no CW.

By "CW" I mean the candidate who'd beat everyone else if
all the voters sincerely ranked all the candidates. That seems
the usual meaning. I use "BeatsAll candidate" to refer to someone
who actually beats everyone.

***

>You wrote (26 Feb 2000):
> > And based on private email correspondence with Mike Ossipoff over the
> > last few months, it's clear to me that Mike also does not believe
> > Schulze is better than Tideman in any important way, and Mike no
> > longer considers GMC or Beatpath GMC to be important.
>
>I don't think that Mike Ossipoff needs you to speak for him.

It's true that I don't believe that there's any important
merit difference between the Cycle Condorcet methods, which
include Tideman, Schulze, SSD, SD, DCD, & IBCM. They're the
top class of voting systems.

Sure there's a case for electing as President the Tideman
winner, if we know that the Tideman winner will usually
beat the Schulze winner when the two are different. When
2 methods meet the important criteria, there's a case for
saying why not elect the winner who's more likely to be preferred
by the public to the other winner. But the important thing is
that both methods, and all the Cycle Condorcet methods are
really just as good, especially in public elections. It seems
to me that debate about Tideman vs. Schulze should be done
lightly, since we're talking about a _detail_, and since we
agree on what class of methods is best.

Any criticism energy could be more effectively employed by
directing it against IRV, which is currently being proposed
in San Francisco, California; Santa Clara County, California;
and which has been proposed to the California League of Women
Voters. They're pushing IRV through, and we're fiercely battling
about which Cycle Condorcet method is better?

It could also be pointed out that Schulze, SSD & SD do a
better job of minimizing the strength of the strongest ignored
defeat than Tideman or DCD does. For pure principle, the
pairwise comparison of the 2 winners seems more convincing.
For public acceptance of the result--I don't believe either
method will give the public anything to really object to, but,
considering whatever small acceptance difference there could
be, it could probably be argued either way.

There's someone in my town how considers Schulze's method the
best one to propose, even though I described SSD to him too.
That doesn't bother me, since all those methods are almost
the same in regards to merit. I'd join a campaign just as
determinedly whether the method is Tideman, Schulze, SSD, SD,
DCD, or IBCM.



I no longer use GMC. I quit using it because it was awkward for
me to apply it, and SFC covers what I meant by GMC. Also, as
I said, BC generalizes & underlies the defensive strategy criteria.
If a method meets BC, then it meets all 5 defensive strategy
criteria.

We all often quote others, as when talking about how the opinions
line up on these issues. For instance I can say that you (Markus)
& Steve & I agree that the meaningful & useful measure of the
strength of a defeat is the number of people who voted for
that defeat (That of course is the measure that we call
"votes-against" or "winning-votes"). That's a much more crucial
issue. Sometimes we focus too much on small details on which we
disagree, like which of the Cycle Condorcet methods is better.
All of the Cycle Condorcet methods--Tideman, Schulze, SSD, SD,
DCD, & IBCM--meet all the criteria that are important to me.
The margins methods meet none of the criteria that are important
to me.

So that's the lineup that I point to: It's 3 to 1 in favor
of winning-votes, as opposed to margins, on this list. If Tideman
defined his proposal in terms of margins, that's surely because
he hadn't been in discussions about other measures of defeats.

(When the name "Condorcet method" is used, it goes without saying
that it refers to a votes-against method. And I feel that
methods that meet Condorcet's Criterion should be called
"Condorcet Criterion methods", rather than "Condorcet methods",
since Condorcet defined a specific type of method, in addition
to his more general Condorcet Criterion).

>It is always very amusing when somebody criticizes a given election
>method so vehemently and simultaneously proposes another election
>method that is almost identical and that is based on the same idea
>(here: beat paths).

IBCM is different from Schulze, and there's nothing wrong with
debating minor differences, like the possible merit differences
between the Cycle Condorcet methods. But that difference shouldn't
be blown out of proportion, in comparison the the really drastic
merit differences between votes-against vs margins, and
Condorcet vs IRV.

>
>In so far as until recently your favourite election methods were
>JITW//plurality, JITW//IRO and JITW//MinMax, I have convinced you
>of almost everything I wanted to convince you. Remember that it is
>me who introduced the Tideman method, beat paths, reversal
>symmetry and clones to this mailing list. If I didn't do that, then you
>would still discuss with Mike Ossipoff ad infinitum whether the
>MinMax method or the Copeland method is the best method.

Now wait a minute. It's true that you (Markus) first intruduced
a Condorcet version other than MinMax & Smith//MinMax to this
list. But, soon after, I suggested SD, which is equivalent to
Schulze's method when there are no pairwise ties or equal defeats.
Then, some time later, SSD was proposed. I wrote to you about
that method at the time that it was first proposed. SSD is a
refinement of SD that is also equivalent to Schulze when there
are no pairwise ties or equal defeats.

Neither SD nor SSD was derived from Schulze's method, and
neither owes its invention to Schulze's method. Those are
completely different approaches, separately invented. Likewise
IBCM & DCD.

And no current participant in this discussion was arguing that
Copeland was better than MinMax. No current participant here
was advocating Copeland at all here at any time.

And, as I said, the methods SD, SSD, IBCM & DCD weren't derived
from Schulze's method, and don't owe their invention to it.

As I said, this discussion about the relative merits of different
Cycle Condorcet methods should be taken more lightly.

Mike Ossipoff





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