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

[Donald Davison]

04/03/02 - Mr Stéphane Rouillon wrote:

Date: Sat, 30 Mar 2002 01:25:55 -0500

1000: A,B
1000: A,C
1000: B,C
1000: B,A
1000: C,A
1000: C,B

Maybe I am mistaken, but this example seems very realistic to me.


Hi Stéphane, Donald here, you are mistaken.
  Your example has 6000 votes divided exactly between three candidates,
2000 each, and then divided exactly again, 1000 each for two combinations
of choices for each candidate.
  I would say that is very unrealistic.

Stéphane: "And monotonicity is still a problem using IRV...."

Donald: It is not a problem if it does not exist.  Has it ever been a
problem in a real election in the real world.

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.


Regards, Donald,