[EM] Issues, Condorcet, and IRV (was: IRV vs. plurality)

[Eric Gorr]

At 4:07 PM -0400 8/12/03, Rob Speer wrote:
>On Tue, Aug 12, 2003 at 09:23:18AM -0400, Eric Gorr wrote:
>>  The Participation Criterion seems useless.
>
>It seems to me like a generalization of monotonicity.

Yes...and I am rethinking my stance on monotonicity as well.

>It seems difficult to satisfy, but not useless.

Again, I believe it is.

Any decision can only be based on the information at hand (this 
includes intuition, etc.), knowing full well that should some 
previously unknown information come to light, the decision could very 
well change.

Arguing that the decision should never have the potential to change 
based on different information seems wrong.

What I am beginning to move towards now is something along the lines of...

   based on the information at hand, will an election method will
   select an obviously wrong winner, like IRV will in this case:

   40 A
   35 C > B
   30 B

   and other cases like it.

It would also make a huge difference on how easy it would be to 
manipulate a method via insincere votes, adding clones and perhaps 
others things not coming to mind at the moment.

>Wouldn't it be good if voters knew that voting couldn't make the
>result worse from their point of view?

Well, what matters is what the group as a whole thinks...and that 
decision is based upon all of the combined information to which an 
individual voter can contribute.

Adding those three votes resulted in a different group decision.

Furthermore, it is possible that for every election method an 
individual voter can make the result worse from their point of view. 
For Approval, it would be not selecting the cutoff at the right spot. 
We've already seen cases for IRV. I would imagine modifying the 
preferences of the last three voters, A could be made to win...but I 
haven't played with that.



So, convince me I wrong.