On Mon, Nov 13, 2000 at 04:08:21AM -0500, Branden Robinson wrote:
> There is also A.3.6.
> In cases of doubt the Project Secretary shall decide on matters of
> procedure (for example, whether particular amendments should be considered
> independent or not).
That's probably reasonable.
> > Well, no. People (well, at least one of them) think the only
> > constitutionally sound way of offering an alternative to be voted on
> > is to amend a resolution, whether that alternative completely reverses
> > its meaning and/or effect or not.
> I guess voting NO is not a sufficiently vigorous way of registering one's
> disagreement...
Exactly.
> What should be done about this?
> * * [two other options and..]
> * Should we amend A.3.3 to clearly extend the scope of ballot questions,
> not limiting them to yes/no issues?
> I don't think the first option is really going to get us anywhere. The
> second option might set a precedent for loose interpretation of other parts
> of the consitution. That leaves yet another constitutional amendment...
Which is probably due, anyway.
There are at least two issues which could use correcting:
* the method for resolving circular ties
* how to make alternative resolutions available
Digressing a bit (and rereading some of Mike Ossipoff's mails from June
and February), these are actually related: I imagine what irritates
Branden most about having the "gutting" amendment is that it's voted
on first, so removing non-free had to win a vote just to be considered,
as well as a second vote to be accepted, and that might encourage people
to insincerely vote for the amended version just to try to kill John's
proposal (or conversely). This is called strategic or sophisticated
voting, apparently, and is really a Bad Thing.
So what we should presumably have is a single vote on how to resolve an
issue, with all alternatives presented, and a single resolution chosen.
Some URLs for voting methods:
* http://www.eskimo.com/~robla/em
* http://www.fortunecity.com/meltingpot/harrow/124/
I found the latter most helpful, probably mainly because it states its
results without proof rather than with...
Since we're already using a Condorcet-base scheme, it's probably best to
keep doing that (ie, keeping the "foo DOMINATES bar"). From the latter
URL, it seems that "Tideman" and "Schulze" are probably the most suitable
(they're not vulnerable to most of the nasty strategies). Mike Ossipoff
listed a whole bunch of related systems in his letters too.
From Mike's letter:
Schulze's method:
1. A has a beatpath to B if either A beats B, or if A beats
something that has a beatpath to B.
2. The strength of a beatpath is measured by the strength of its
weakest defeat.
3. If A has a stronger beatpath to B than B has to A, then A
has a "Schulze win" over B.
4. The winner is the candidate who has a Schulze win over each
one of the others.
The Tideman method, according to the second url above is:
Pair-wise victories are considered starting from the strongest, and
working down. If a victory does not conflict (by forming a cycle) with
previous "locked" victories, it is "locked". Otherwise it is
"skipped". The alternative that is left with all defeats "skipped"
wins. Strength of victory is based on either Margins or Winning-Votes,
as described for Minmax.
They're both quite complicated and unusual, but they choose a result
based on an honest ranking of candidates; largely without people being
able to get different results by adding different options ("a vote for
Nader is a vote for Bush"), or forcing people to vote insincerely ("go
ahead! vote for a third party! throw your vote away!") to get something
they'll find acceptable, which apparently is very rare in voting systems.
I presume the best way to handle different possiblities on ballots is
just to vote on them at once (eg, "Remove non-free // We love non-free! //
Status-quo // Further discussion") and have whichever one wins (according
to the voting rules, and any supermajority requirements), win.
The only problem with that is when there are semi-independent options on
the ballot. For example, since the issues in this mail are largely independent,
we might have essentially two ballots with these choices:
[ ] Use the SCHWARTZ method for counting votes
[ ] Use the TIDEMAN method for counting votes
[ ] NO change
[ ] FURTHER discussion
[ ] ONE vote to choose a form of the resolution and accept it
[ ] NO change
[ ] FURTHER discussion
But this means there's no way for someone to say `If we do SCHWARTZ
or TIDEMAN on the first vote, then I'll vote for ONE on the second;
otherwise I'll vote for FURTHER/NO on the second'; or conversely to say
`If we change to ONE on the second; then I want SCHWARTZ or TIDEMAN;
otherwise I'd rather NO on the first ballot'. This probably can't be coped
with particularly well at all, actually (a cross product vote could end
up splitting the TIDEMAN vote, say, and breaking the vote counting).
So, I guess for doing different options, the best way is to essentially
just have a separate ballot for each independent issue, and to let related
proposals and unaccepted but seconded amendments be voted on concurrently.
Yes? No?
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.
``We reject: kings, presidents, and voting.
We believe in: rough consensus and working code.''
-- Dave Clark
Attachment:
pgps4_xjqWCAT.pgp
Description: PGP signature