RE: Condorcet Voting and Supermajorities (Re: [CONSTITUTIONAL AMENDMENT] Disambiguation of 4.1.5)
Steve Greenland wrote:
> I'm not sure it really makes any sense to have alternatives with
> different majority requirements[1]. The recent case of an ammendment
> that had an (arguably) different requirement than the original GR came
> from two issues:
I agree with you that supermajority requirements don't make much sense when
using the 'Concorde' (Condorcet's) method. Usually, a supermajority
requirement is used to prevent drastic flip-flops in the basic policies of
an organisation. If an organisation is divided into two large factions with
opposing views, standard 'winner-take-all' (plurality) voting will result in
one group or the other becoming dominant, and changing everything their way.
A shift of a few votes in a subsequent election can cause a dramatic
reversal, and this can destabilise/destroy an organisation. However,
Condorcet's method always allows a range of policy options to be considered
(rather than just two), and it will *always* choose the best compromise.
Therefore, provided a good compromise is proposed by someone, there should
never be any radical changes in policy, merely gradual evolution. In this
case, supermajority requirements simply undermine the democratic character
of an organisation, by giving the supporters of the status quo much more
political power than those who prefer change.
That said, if you do want to maintain a supermajority requirement, the
cleanest and most effective way of implementing it is to have voters cast
their ballots normally, ranking the no change/status quo alternative (what
you call FURTHER DISCUSSION) along with all the other options. However, for
any proposal to actually win its pairwise contest against (dominate) the
status quo, it must do so against 3-1 odds (or whatever). Therefore, to
determine the winner, just multiply the votes for the status quo by 3
against every alternative before comparing. For example, suppose we have
the following pair of vote totals:
Proposal X: 55
Further Discussion: 20
Then Further Discussion will win this particular pairing if there is a
supermajority requirement of 3:1, since 20*3 > 55. For Proposal X to win
overall, it must therefore achieve a 3:1 supermajority against the status
quo, and a simple majority against all other alternatives. The same is true
for any other proposal, of course.
> 1. Confusion about whether the constitution even allowed modification of
the
> document in question.
>
> 2. An "ammendment" that was basically a refutation of the original GR.
>
> Manoj and Branden are trying to resolve #1, and a little more care in
> the GRs and ammendments we propose and second will solve #2.
>
> Or perhaps the best answer is to require seperate votes on such issues
> (modifications of the Constitution and other issues that require
> super-majorities:
Obviously, votes on general resolutions (which require a simple majority)
cannot be combined with constitutional amendments that require a
supermajority, since the different majority requirements make it impossible
to compare them. This is true even if they both address the same issues in
different ways. If a constitutional amendment is proposed, the Secretary
should require any other counter-proposals or amendments to *also* be worded
as constitutional amendments, or else rule them out of order. Of course, if
these counter-proposals do not require constitutional changes, they can be
proposed later as ordinary GRs (in which case other proposals which *also*
don't require a constitutional amendment could be considered). Opponents of
the constitutional amendment(s) who think the matter should be addressed
through a GR should simply vote for Further Discussion, of course.
> 1. First vote to determine what combination of GR+ammendment(s) will be
> considered (using a simple Condorcet winner).
>
> 2. Vote on whether to accept the final proposal. Since this is a simple
> yes/no, determining a super-majority is trivial.
I've noticed that there is a tendency among participants on debian-vote to
propose solutions which 'serialise' decision-making, rather than using using
the existing mechanisms. I think this is a result of people's experience
with organisations which use Robert's Rules of Order. For example, when
John Goertzen proposed his GR to amend the DSC, an 'amendment' to his GR was
proposed that eliminated the entire text of the original resolution, and
replaced it with what was described by its proponents as a compromise.
Under Robert's rules, this *would* normally be implemented as an amendment,
since it is only possible to choose between two alternatives at a time
(yes/no) using a simple majority vote. Those who are hostile to the main
motion, but fear that it might be preferred by a majority to the status quo
will therefore propose a compromise amendment, leading to a series of votes:
1. Amendment - YES/NO
2. Main Motion - YES/NO
This is the only way under Robert's Rules that a group *can* choose from
among three alternatives (Amended Main Motion/Main Motion/Status Quo). None
of this is either necessary or desireable when using Condorcet (Concorde)
vote counting, since the method is designed specifically to allow voters to
simultaneously choose the best single alternative from among any number of
proposals. Therefore, when using Concorde voting (or any other pairwise
method), the correct procedure is to treat all 'amendments' as distinct
proposals, that get voted on using a single ballot along with the original,
unamended resolution. If the amendments are genuine amendments, rather than
counter-proposals (that is, they modify only a portion of the text of the
original motion), the amendment text should be merged with the original
resolution, and presented to the voters as one distinct alternative. Of
course, if the amendment is a *friendly* one that is simply designed to
improve the clarity of the original motion, the original may be withdrawn by
its backers prior to voting. Otherwise, the original resolution remains on
the ballot as well. The above election, using Concorde voting should
therefore result in a single ranked ballot with the following options:
A) Resolution
B) Resolution+Amendment (counter-proposal, in this case)
C) Status Quo/Further discussion
The only things that should be required (for sanity's sake ;-) are:
1) all proposals must be *germane*, that is, they address the same issue as
the original proposal, but in different ways
2) all proposals have the same majority requirement (see above).
3) voters have the authority to make the proposed change(s).
The Secretary should have the authority to ensure that these requirements
are met on any ballot that is distributed to the membership. Since (3)
became a contentious issue during the discussion of John Goertzen's
proposal, it would have been appropriate to table that GR and any
counterproposals until the question of authority (and *only* that question)
had been decisively resolved, either by Secretarial fiat or by means of a
constitutional amendment.
--
Norm Petry <npetry@accesscomm.ca>
Reply to: