Rich Parties & criteria

[Mike Ossipoff]

First, I want to say (before Bruce does--as he has a right to do)
that the most recent criterion that I posted, my ITA criterion
to measure the Rich Parties fault of Copeland had imprecisions in it,
as posted. I was writing a precise version, but Steve has already
written one.

In fact, as of this time, 5 criteria have been written to measure
that problem, various measures for it by y/n tests, known ;as
criteria. Actually 2 of the criteria for that are equivalent,
though they have different emphasis.

But though there's no shortage of criteria, now, to demonstrate
Copeland's Rich Parties failure, it's also true that we have to
be a little skeptical when Bruce says that only criteria, accompanied
by proof of compliance & non-compliance, can validly establish 
anything. Has Bruce proved that?

The Rich Parties example, and the voting-on-a-movie story
make it quite obvious that Copeland is doing something wrong.
All it takes is 1 example to show that it can happen, and no
other proof is required.

Nevertheless, we have those 5 criteria, now precisely defined,
and demonstrations that they're failed by Copeland but not
by Condorcet (similar proofs of compliance could almost surely
be written for every method but Copeland). And we have 6 other
criteria relating to the lesser-of-2-evils problem & defensive
strategy & majority rule, also with demonstrations of compliance
for Condorcet, and demonstrations of failure for Copeland.
My 40%,25%,35% is all it takes to demonstrate Copeland's failure
for most of them. My 4-candidate example for IDE was an example
where Copeland demonstrates its failure of that criterion--
something that of course need only be demonstrated once for
a particular criterion.


Mike


--