[EM] Making UUCC more unambiguous still. New criteria in general.

[MIKE OSSIPOFF]

EM list--

First, let me state the more unambiguous wording of UUCC:

It should never be possible to contrive a configuration of
candidates, voters, voters' sincere ratings, & voters actual votes,
with which there's at least one voter for whom: 1) His ballot
votes Bc over B; & 2) There's no way that he can change his ballot
so that it no longer votes Bc over B without causing the election
of someone whom he likes less than Bc.

[end of definition]

Now that seems completely unambiguous.

I'd felt that there could be a misunderstanding that I meant that
if a voter can find a way of reversing his Bc>B vote that will
elect someone worse, the method fails, or that it only fails if
_anyone_ who makes that change could elect someone whom he likes
less than Bc. The preceding rewording is intended to avoid
those possible ambiguities.

I realize that Blake pointed out that SrDSC wasn't what I thought
it was, with the result that I dropped it, and that I felt that
my original UUCC definition could be interpreted other than as
I intended, with the result that I rewrote it twice, with the
above wording as the result. And I realized that my SARC
definition, as originally written, didn't apply to some methods,
with the result that I rewrote it and posted the new definition
here.

When devising new criteria one is trying to find exact
wording to describe something that one doesn't want to happen,'
or something that one does want to happen. We all know when Plurality
does something that we don't like, but expressing that with exact
unambiguous language isn't as easy. So, when devising new criteria,
it isn't always immediately obvious how to word them, and sometimes
they aren't right the first time.

So if I've been having to correct those definitions, that's because
I've been writing new criteria.

And, by writing criteria that need rewriting, I'm in good company,
because, for example, the meetable wordings of the Condorcet
Criterion are met by Plurality, if they're construed to apply to
Plurality, as they probably rightly should be construed.

And sometimes the Condorcet Criterion is stated in a form that
isn't meetable by any method.

So then one of the most popular & well-known and often-used
criteria isn't well written, and acts contrary to the intent of
those who use it. My point is that it wouldn't be fair to
unfavorably judge my criteria on the basis of the fact that
some of them have required rewrites.

Mike Ossipoff




_________________________________________________________________________
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.

Share information about yourself, create your own public profile at 
http://profiles.msn.com.