Re: Condorcet cuckoos: promoting the method by having it get the winners wrong
At 03\04\22 10:35 +0200 Tuesday, Matthias Urlichs wrote:
>[] Re Condorcet cuckoos promotin3.ems
>*** PGP Signature Status: unknown
>*** Signer: Unknown, Key ID xD70AAFF9
>*** Signed: 03\04\22 8:35:55 PM
>*** Verified: 03\04\22 10:21:05 PM
>*** BEGIN PGP VERIFIED MESSAGE ***
>
>Hi,
>
>Jochen Voss wrote:
>> [ Cc to debian-vote, bacause it may be of general interest. ]
>>
It was not of interest.
I see that you have not removed the untruthful claim at your
website saying that Condorcet is monotonic.
http://www.mathematik.uni-kl.de/~wwwstoch/voss/comp/vote.html
A good ideal rule will pass only a good ideal preferential
voting method. Instead it could do that and also pass every
undefined constant:
x = undefined
So monotonicity might as well fail the constant "x".
Also it can fail Condorcet too, which has a region that is
undefined.
I am sure Mr Voss has no interest at all in what I write.
But I want to add a bit of detail to my private claim to Mr Voss
that perfectly false information at the website of Mr Voss is
also information that I request to be removed. It is not a
statement of opinion but instead a false statement on preferential
voting.
A background detail is that the person who created the
information never corrects one of the dumb rules.
Mr Voss has them online twice (but zero times would be nicer):
http://www.mathematik.uni-kl.de/~wwwstoch/voss/comp/local-quorum.html
Mr suggests that compliance with the rules of Mike Ossipoff
will shift or change. The rules can not test and any attempt
to assess the health of the Debian Voting system might show
that.
Here is the text on the page:
--------------------------------------------------------------------
>Debian modifications of Condorcet voting
>
>The Debian voting system differs from plain Condorcet voting in
>several important points.
>
>It introduces the concept of a default option. This option is
>either "None Of The Above" or "Further discussion". This option
>should win if the changes below forbid any other decision.
>
>It introduces the concept of a quorum. A quorum should assert that
>there are no decisions without a sufficient number of electors.
>There are two variants of quorum:
>
> global quorum
>
> This is the more traditional form of quorum: if there are not
> enough voters the whole vote is dropped.
>
> per-option quorum
>
> Only options which are not preferred over the default option
> are dropped. I have written an analysis of per-option quorum.
<http://www.mathematik.uni-kl.de/~wwwstoch/voss/comp/local-quorum.html>
This conclusion on "comp/local-quorum.html" is sure to be completely
false. Mr Voss writes:
+ Result: with the introduction of an per-option quorum we loose
+ CC, GCC, SFC, GSFC, SDSC and WDSC. We still have MC.
All repaired 1 winner Condorcet methods that select the Condorcet
winner if it exists, are never monotonic. The website's acronym
for the rule of monotonicity is "MC". It is false.
I emphasize that what is going one here is absolutely not one of
copying in hope or good faith and publishing that and later finding
that it is false. Instead the rules are so stupid and unthinkable
for a mathematician that even considering them to be real should
be about impossible for reasonable persons. Mr Voss is
implying he checked 6 of those rules (CC, GCC, SFC, GSFC, SDSC and
WDSC) in 2 cases presumable against a method that puts a QE solver
at the limit of its abilities and I guess that Mr Voss does not
use a QE solver. However MC is a better defined on the page (less
reliant on people that might not exist).
" We still have MC." says the text. But MC is not satisfied both
before and after (at least, if the Condorcet method selects the
Condorcet winner for the (AB),(B),(C) 1 winner 3 candidate election).
I think Mr Voss made such a packed collection of untrue statements
that the consideration could have been more about Debian than about
mathematics.
Perhaps you would say why the mistakes were made ?.
--
>
>It introduces the concept of a super-majority. Super-majority
>requirements should assure that even a minority of electors can
>inhibit some fundamental decisions. This is used as a special
>protection for the Debian constitution.
...
>It is not finally decided how we will modify Condorcet voting to
>implement these changes. So it is not clear which of the above
>properties of plain Condorcet voting we will have to sacrifice. At
>the time of writing the current draft is Manoj Srivastava's April
>17th Draft of the Voting GR. Comments about this draft should be
>sent to the debian-vote mailing list.
What CRITERIA does Manoj Srivastava *USE* when deciding on whether
the complex changes would make Debian better or not?.
I leave members to imagine the QE equations themselves.
The "Manoj Srivastava" changes:
http://lists.debian.org/debian-vote/2003/debian-vote-200304/msg00044.html
I was writing privately to Mr Voss, since 7 April 2003, and seeking
deletion of false information connected with the Debian project,
at the comp/vote.html page. Nothing I requested go, has been
deleted.
>It would be if he had actually answered the question.
Reply to: