[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

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

Manoj Srivastava wrote:
On Wed, 21 May 2003 14:27:53 -0400, Buddha Buck <bmbuck@14850.com> said: 
2) Would an amendment to (a) to the following effect be acceptable
  and
clear up nomenclature issues:



Replace A.6.2-4 in the proposed amendment with:



2.  Procedural Definitions
a. V(A,B): For any options A and B, V(A,B) is defined as the number
        of ballots cast that rank option A higher than option B
b. margin of A over B: The margin of A over B M(A,B) = V(A,B)-V(B,A)
        Note that M(A,B) = -M(B,A)
c. defeats: For any options A and B, A defeats B if and
         only if M(A,B) > 0
d.  Acceptable: An option A other than the default option is
         considered "acceptable" if and only if M(A,default) >= R,
         where R is the "quorum requirement" for the vote.


	This changes the meaning of Quorum as used by Debian. In
Debian, quorum is used to ensure there a modicum of interest in an
option -- so, if R people vote for an option, there is interest. It
does not matter how many people vote _against_ the option -- Quorum
is not used to ensure a margin of victory.
Ah... then I was confused. Replace "M(A,default) >= R" with "V(A,default) >= R and M(A,default)>0" The V(A,default >=R clause comes from your proposed A.6.2, and the M(A,default)>0 clause comes from your proposed A.6.3. 


	If you meant to change the meaning of quorum, I must confess I
disagree.
My intent was to change none of the semantics of your proposal. My initial intent was to change the terminology from "per-option quorum" (which is subject to misunderstanding and confusion with the standard "total votes cast quorum" concept) to "acceptability". Instead of saying an option that only a small handful of people prefer to the default does not meet quorum, we would say that such an option is not acceptable to enough people. 




e. Superacceptable: An option A with a supermajority
        requirement of N:M is considered superacceptable if and
        only if M*V(A,default) > N:V(default,A).
f.  Pairwise defeat: A pairwise defeat is an ordered pair of
         options (A, B) where A defeats B.
g.  "weaker" A pairwise defeat (A,B) is considered weaker than
         pairwise defeat (C,D) if V(A,B) < V(C,D)
3. Dropped options
a. Any non-default option A which is not acceptable is dropped
b. Any option A with a supermajority requirement which is not
        superacceptable is dropped.
4. Create a list of all pairwise defeats (A,B), where neither A nor
  B
are dropped, sorted by V(A,B).



with related changes elsewhere to use these definition.



	So, to answer your question, no, I would not find this
acceptable for my version of the GR, for the reasons presented
above.
I understand that, since I messed up a crucial definition. Is the new definition better? 



	manoj
Reply to: