To comment on the last part first:
Kevin says:
(Although, incidentally, Mike's criticism that my scheme can't be used to
show that Approval and Range fail SFC is quite strange. SFC is defined on
sincere preferences. If Mike means "what I use in place of SFC"
I reply:
Yes, I thought Id made it clear that I was asking how, in _your_ system of
criteria, youd show that Approval and other RV versions pass or fail
SFC--your criteria systems way of saying SFC. But if you prefer, just show
that Approval and some other RV version pass or fail Condorcets Criterion,
as your criteria system expresses CC.
And, as I said, you never demonstrated that. I said that years ago, and the
statement remains true now. Your criteria system remains a sketch only, as
long as you cant precisely define it, and then use it to demonstrate
compliance or noncompliance.
Your criteria system has a privileged balloting system. My criteria make no
mention of any balloting system.
By the way, for the kind of approach that youre doing, it might be better
for the privileged balloting system to be ratings instead of rankings.
Ratings contain rank information, but ranks dont contain rating
information. Ratings are more general than rankings are. Therefore, making
them the privileged balloting system isnt quite as unjustifiable. Not
quite. But your criteria system would still be about a balloting system,
something that my criteria make no mention of.
And then theres the fact that your system is based on a false assumption,
even though its falsity is acknowledged, and even though the results might
come out right in the end. Lets pretend that Plurality has rank (or
rating) ballots? What a thing to start a criterion with. What great lengths
some people will go to, to make Plurality fail Condorcets Criterion without
mentioning preference.
These things make your criteria system thoroughly inelegant--an arbitrarily
grafted mess.
Kevin continues:
then I believe it's at least true that there is no way to interpret Approval
or Range such that they satisfy the criterion under my scheme.)
I reply:
Forgive me, but I didnt ask you to show that Approval or RV satisfy the
criterion. I asked you to show that they meet or fail it. Either will do.
More recently I asked that about SFC. But if you dont want to deal with SFC
(though you do claim that you have a votes-only wording for it), then do so
for Condorcets Criterion. I believe that that was probably my initial
request, a few years ago.
Mike Ossipoff
----
election-methods mailing list - see http://electorama.com/em for list info
