[EM] 04/03/02 - Mr Stéphane Rouillon wrote:

[Elisabeth Varin/Stephane Rouillon]

Mr Davison,

1) I do agree about your politic concerning forwarded e-mail.
I consider I do not need to talk behind anybody's back and
I think what I say can benefit discussions including anyone.
All I ask is not being quoted out of context, in the same manner
I do not take everyting relayed for granted....
However, If I am asked to keep something private I will without problem
-- as some of my interlocutors asked -- and I would expect someone to do
the
same if I asked.

2) You missed the plus (+) sign:

1000: A,B
1008: A,C
1000: B,C
1005: B,A
1000: C,A
1004: C,B

This example seems very realistic to me,
because a lot of alike problems can be produced and it is very
realistic (probable) that one would occur.
So you could build a lot of more complex and even more realistic
problems...
And monotonicity is still a problem using IRV....
for at least this whole class of problems:

Base + Alpha            : A,B
Base + Beta + Diff1  : A,C
Base + Gamma         : B,C
Base + Delta + Diff2 : B,A
Base + Epsilon          : C,A
Base + Fi + Diff3      : C,B

0 < {Alpha, Beta, Gamma, Delta, Epsilon, Fi} < Diff1 < Diff2 < Diff3 <
Diff1 + Diff2 < Base

If you do not think it is a lot you can count the possibilities for
10,000 votes...

3) Stéphane: "I use IRV in my model but I still search for a better
method."
Donald: If you keep searching you should come full circle back to
IRVing.

No. If you know how to search you really can improve such a problem.
Most of you are obsest with your own ranking of the available criteria
and just try to convince the others your priorities are the good ones.
I do not. I am not changing the ponderation used to qualify the best
solution.
I am trying to change the problem so it will bring a new solution with
different
properties. I know it seems vague for you at the moment. But it is not
to me,
I already have done the job. And you will be able to see it in some
weeks
when I have finished translating my model.
I have seen some persons on this list are mathematicians with some
knowledge
of the simplex phase I method. Mr. Dantzig's procedure add some
dimension to the problem in order to find a first feasible solution.
In the same way, I am adding more freedom to the winner selection
process,
in order to obtain a better set of representatives.
An election should be a representation exercise, not a ball game.
It is not about winning, it is about fairness.

Regards,
Steph.