Re: Constitutional amendment: Condorcet/Clone Proof SSD vote tallying

[Nathanael Nerode <neroden@twcny.rr.com>]

Raul Miller wrote:
>
>On Fri, May 23, 2003 at 08:46:13PM +0200, Jochen Voss wrote:
>> In my example local quorum causes the following problem:
>> dropping an irrelevant option changes which
>> relevant option wins the election.
>> Global quorum does not have this problem.
>
>The way you've apparently defined your terms:  Yes, Global quorum does
>have this "problem".  [However, it's true that your example doesn't 
>show this flaw with what you call "global quorum".]
>
>The reason is that condorcet has this "problem",
>even with no quorums whatsoever.  Martin Schulze's post
>http://lists.debian.org/debian-vote/2003/debian-vote-200305/msg00119.html
>illustrates this principal.

To nitpick, "pure" Condorcet does not have this problem 
(whether you mean the one Jochen mentioned, or violation of 
the participation criterion, which are different), but pure Condorcet 
doesn't always pick a winner.  It's the Condorcet tie resolution method
(Clone-Proof SSD, for instance) which has the problems.

If you *really* want a conservative voting system, use pure Condorcet 
with equal ranking and a default option, and if there's no winner, the 
default option wins.  :-)

In actual fact, there is very often a true Condorcet winner (Ideal 
Democratic Winner).

I mention elsewhere the way in which the quorum requirement can not 
only cause a Condorcet winner to lose, but can do so while causing another 
non-default option to win, and this without a supermajority requirement.  

This, I think, is an actual defect in the Amendment; it causes the 
*Condorcet Criterion* to be violated (which is one of the few good 
reasons to use Condorect in the first place), and to be violated in ways 
*other* than picking the default option (which may be acceptable as a 
measure of conservatism).

--Nathanael