On Wed, Nov 20, 2002 at 01:14:16AM +1100, Clinton Mead wrote:
> What I've noticed in response to my proposal is that no-one has actually
> mention my proposal, which is unusual, considering the subject title of
> the post, but instead have been discussing criteria, such as "whether
> supermajority votes should be strategy free" as below.
There's no point proposing something until you've decided what you want to
*achieve* with that proposal. I think you're missing the point of what
supermajority is meant to achieve at the moment.
> Then Anthony Towns wrote:
> > But it's exactly what we're trying to achieve with the supermajority
> > requirement, isn't it? Allowing voters to vote strategically so as to
> > knock out candidates they don't like?
> Personally, I think election methods that encourage and reward strategy
> (i.e. insencere voting and candidates) are broken and something to
> avoid, but thats for people to decide.
Sorry, but insincere voting is a subset of strategy. In an ideal Condorcet
voting system (of the sort that provably doesn't exist, but hey), there'd
be only one effective strategy to get what you want: sincerely voting
for the various options in your order of preference.
> My method, which I set out in the original post in this thread, and I
> will replicate at the bottom of this post, trys to meet these three
> critiria.
> 1. Strategy Free, as much so as plain CSSD without supermajority and
> quorum requirements.
> 2. Repeated elections with identical votes produce identical results.
> 3. Provides adequte protection for non-supermajority and non-quorum
> options against options requiring supermajorities and quorums.
(3) isn't well defined, and isn't what's wanted. First, "adequate" isn't
particularly useful for evaluating things; second, we don't particular
want to protect make sure that supermajority options are preferred to
*everything*.
I'd suggest, instead, that you want:
(3) If an option requires an N:1 supermajority, then a minority of
1/(N+1)-th of voters can vote in such a way as to ensure that option
does not win, without inducing any prejudices against their other
preferences.
> If my criteria are wrong, please tell me, but if not, to improve it the
> method, I need someone to show an example which breaks it. I don't mean
> this as any sort of challenge, its just that the only way I can fix
> something is by finding out whats broken first.
> ---
> - Definition: Defeat is the normal defeat, not considering supermajority
> and quorum requirements.
> - Definition: A super-defeat is a defeat after considering supermajority
> and quorum requirements.
> (a) The default option is considered.
Is this further discussion which I see before me? Thread dangling
before my hand? Come, let me post! I see my email not, and yet I write
still. Art thou not, fatal subject, sensible to locution as to audition?
Or art thou a debate of the mind, a false creation, proceeding from
the heat-oppressed brain?
> (b)
> (i) If option A has quorum requirements and supermajority requirements
> less than or equal to option B, and option A defeats option B, then
> option A is considered.
> (ii) If option A super-defeats all options with quorum requirements or
> supermajority requirements less than option A, then option A is considered.
Let's go with a different example for a change. We want to change the voting
system, and have a vote. We end up with three options:
A -- adopt a new Condorcet style voting system, changing the
constitution to document this (3:1 supermajority required)
B -- instruct the secretary how to interpret some ambiguities in
the existing text
F -- further discussion
We discuss the issue, most people are convinced that either will solve
the issue, although a few people don't like the idea of instructing
the secretary outside of the constitution at all. Of the remainder,
most prefer the new system, but some would rather just keep things as
they are. As such, the vote ends up fairly straightforwardly:
70 people vote ABF
30 people vote BAF
10 people vote AFB
with the results:
A defeats B, 80:30 but A doesn't superdefeat B (80:90)
B defeats F, 100:10
A defeats F, 110:0 and superdefeats F, obviously
Quorum is about 45, so isn't an issue.
Under (a), F is considered. Under (b-ii) B is considered, trivially. A isn't
considered under any rules. Under (c) B wins.
> (c)
> Perform CSSD on all considered options, ignoring supermajority and
> quorum requirements.
And that's all very well from a voting system standpoint, but I don't think
it's what we *want*. What we want is a system that will:
(a) let us determine which direction the majority want to go
in, and elect that the winner
(b) in special cases, allow a "superminority" to block something
they particularly object to
Methods that scale comparisons between real options don't meet the first
of those requirements -- they miss the fact that in situations like the
above example, we actually _want_ to do the controversial thing.
A minor point:
> (ii) If option A super-defeats all options with quorum requirements or
> supermajority requirements less than option A, then option A is considered.
The default option doesn't have a quorum requirement, so this is a subset
of the original rule "any option that doesn't beat the default option
by the supermajority requirement is eliminated".
Cheers,
aj
--
Anthony Towns <aj@humbug.org.au> <http://azure.humbug.org.au/~aj/>
I don't speak for anyone save myself. GPG signed mail preferred.
``If you don't do it now, you'll be one year older when you do.''
Attachment:
pgpilAcoCnKb_.pgp
Description: PGP signature