[EM] Blake--Your 2 examples. IRV yes, Tideman no.

[MIKE OSSIPOFF]

Blake--

Your IRV example is a UUCC badexample for IRV, since if one
person changes his vote so as to no longer vote Bc over B,
someone whom he likes less than Bc wins as a result, where
Bc previously won, and where everyone likes B better than Bc.

But your example for Condorcet, Tideman(m), & Schulze isn't
an example of those methods failing UUCC, because it took 35
voters reversing their Bc>B vote in order to give the election
to someone whom they like less than Bc. It has to happen if
_one_ voter reverses his Bc>B vote.

If the example is modified so that it's initially 51 to 49
for Bc vs B, and one person reverses that pairwise defeat by
changing his vote, the abovenamed methods continue to elect Bc.
(At first glance).

With the use of truncation, it might be possible to make
Condorcet & Schulze fail UUCC, but it looks more difficult to
make Tideman(m) fail it. That isn't to be construed as
Tideman(m) advocacy over Condorcet & Schulze, because the
criteria met by Condorcet & Schulze, but not by Tideman(m) are
criteria whose failures happen more often in our actual
elections. Plurality & IRV meet Majority Favorite, and Borda
meets Participation, but those are the worst 3 methods.

Also, if a method can only be shown to fail UUCC when there are
equal defeats, or pair-ties other than the tie or near-tie
between B & Bc, that failure wouldn't be relevant to public
elections.

UUCC is intended to measure a particularly extreme & embarrassing
violation, where people feel forced to keep electing Bc even
though _everyone_ likes B better than Bc. Because it's so extreme,
it of course doesn't happen as often as failures of FBC, WDSC,
SFC, GSFC, & SDSC, & BC.

By the way, it true that I'm the only person interested in those
criteria or who has worked on them. Steve Eppley defined GSFC &
BC (the generalization that underlies SFC, GSFC, WDSC, & SDSC),
and proposed other wordings for some of those.

And I believe that there's a general consensus here that the
properties measured by those criteria are desirable, directly
relating as they do to majority rule and getting rid of the
lesser-of-2-evils problem. Russ Paielli has asked for more opinions
about those criteria. If you have an opinion, send it to me, and
I'll forward it to him. Or you could write to him directly if his
website, www.electionmethods.org, gives an address for direct
reply. And, Blake, next time it would be great if you would send
me a copy of your criticism of the criteria when you write to
a website owner,
because surely you don't feel that your arguments' convincingness
depends on their not being replied to.  But thanks for pointing out
SrDSC's unexpected results. Don't hesitate to let me know if
you believe that a criterion works other than as I thought it did.
Why mention it to the website owner but not to me.

By the way, did you write to Barnsdale too?


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.