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

Re: Discussion: Possible GR: Enhance requirements for General Resolutions

Ron <ron@debian.org> wrote:
> > On Fri, 02 Jan 2009, MJ Ray wrote:
> > In the past, I've seen considerable resistance to vote topics being
> > discussed outside -vote, unless they're by one of a few popular DDs.
> > Do supporters of nQ expect this situation to change, only those
> > popular DDs be able to propose GRs, or can someone suggest acceptable
> > ways of recruiting seconds outside -vote?
> Do you advocate the current situation to NOT change? [...]

No.  I accept a change may be worthwhile, but 2Q seems very high and
suggested without reason.  (See my other messages on the topic.)

> Do you really think it would have been difficult to obtain 2Q seconds
> for a resolution to recall the previous vote, and postpone it until
> some of the more obvious glitches had been better ironed out?  [...]

Yes, based on the summary of other votes by Wouter Verhelst and others.

So, are supporters hoping this situation will change, only a few
well-connected DDs will be able to propose GRs, or what?

> We seem to have totally lost the goal of making decisions that affect
> many or all developers by consensus.  The process of building consensus
> revolves around satisfying the concerns of people who see problems with
> your planned course of action to arrive at a Better Solution.  If you
> can't get the consensus of around 30 people to begin with, it doesn't
> take a degree in advanced math or political science or military strategy
> to arrive at the conclusion that you are a LONG WAY from having the
> consensus of the whole project.

In general, that's correct.  In particular, if you need 30 people just
to *start* the discussion period, that's going to kill many potential
options before they have any chance of building consensus and others
will be far too entrenched by the time public discussion starts;
also, it's 30 DDs, not 30 people.

> By way of example, this proposal was not some off-the-hip idea of
> Joerg's.  It has already been discussed to the point of little (or
> rather no) objection in another forum, and has in-principle support
> from quite a few people.

Could someone link to that discussion, please?  It may contain answers
to questions being asked now.

> You'll note it was not proposed as a vote,
> even though it could easily get the required number of seconds to do
> so, but rather as a discussion point to further build that consensus
> among a wider forum, and hone some of the little (but important)
> details.

I applaud that it appeared pre-proposal[!], but I think the emphasis
is on building a majority (not consensus).  The discussion so far
seems to have consisted of Joerg[*] and others defending the proposal
as it currently stands, rather than engaging in any
consensus-building.  There was one question[+] but no follow-up on
that in a week, so I've moved from seeking amendments, to emphasising
the profound problems in the proposal, in the hope of getting
follow-up or at least avoiding that first public draft continuing.

* - http://lists.debian.org/debian-project/2008/12/msg00191.html

+ - http://lists.debian.org/debian-project/2008/12/msg00195.html

> That you seem to now be waging a 'campaign' against it, does seem to
> indicate that you have quite missed the point.  How about we drop this
> war-word 'campaign',

Fine by me: I didn't introduce 'campaign' to this aspect of the discussion.

> and you instead come up with a concise list of
> your concerns, so that we make take them to build a better proposal
> rather than load them into a vote option as ammunition to try and
> shoot it down.  I don't want this to get just enough support to
> squeak by, I want everyone to agree on the problem and give their
> best to finding a solution that they like.

Here's a summary list of concerns I mentioned in other emails:-
1. 2Q is unjustified and excessive;
2. the obvious spoiler effect may exclude consensus options
prematurely (interaction of thresholds and Condorcet voting);
3. it favours organised campaign groups who gather in secret before
springing discussion on debian lists;
4. it encourages defending proposals too early, during the discussion

> I think your comparisons to local government councils as 'similar'
> organisations is a misdirection.  You say any constituent may take
> something to the council which they must then vote on.  [...]

No, I never said that.  Any constituent may ask something of the
council which (as I understand it) we must then answer - it rarely
results in a vote because most questions are matters of fact. However,
DDs have nothing similar in the debian project - to reduce GRs, having
another way for developers to ask a question that nearly always gets
answered might help.

But if one thinks comparisons to local government councils are a
misdirection, what about company boards or the ICANN At Large?

Or what about providing some other, better comparisons or analyses
from *outside* debian, instead of just complaining about ones which
I have offered?  Patches welcome!

> A better comparison to the case we are concerned with here would
> be a referendum or plebiscite, where every constituent, informed or
> not, is called upon to cast their lot for one option or another.
> I don't know of anywhere else in the world where such a small
> minority can call for a poll of that magnitude. [...]

Well, I think England has no right for voters to call referendums, but
I believe it only takes 10 or so from 3000 to trigger a partial
election in some circumstances.

The referendums-called-by-voters I've heard about most are in
Switzerland, where 50,000 or 100,000 voters (depends on the topic) can
call a referendum[@], out of 4.8 million voters[#].  Scaling down,
that's like 11 or 21 DDs.

@ - http://www.ch.ch/ebuku/demokratie/00001/index.html?lang=en&glossarid=1

# - http://www.admin.ch/org/polit/00054/index.html?lang=en

I think Switzerland's direct democracy is fairly well-respected (as
much as most government models), isn't it?

I'd welcome other examples, particularly if the minimum is equivalent
to anything like the 30 or 60 in the original proposal.

> [...] but it is a
> conservative step in the right direction, the effect of which we
> will be able to then gauge.  If it really is a terrible mistake,
> and proves so in practice, I likewise don't think we'll have any
> trouble finding 30 seconds for a resolution to reverse or modify
> it in some more suitable way.

I don't think a 600% increase is a conservative step.

Also, if this reform doesn't work and we have trouble finding 30
seconds for necessary resolutions, then I fear we'll have trouble
finding 30 seconds for another internal-policy bugfix resolution.  I'd
feel safer if this was a limited-time trial at first, or at least the
previous SRP could be used to modify it, as a safeguard.

> As a further practical proof of practicing what I'm hoping for here,
> you may note that I say this not as a member of the self-select few
> that are regulars in these forums, but as a concerned member of a
> project who just wants to Get The Things Done That We Are Supposed
> To Be About.  Minimising the number of votes we hold where the
> options have been composed by small warring factions that are hostile
> to each other's opinions seems to be a good start to minimising the
> time I must waste being conscripted to one militia or the other,
> against my will and better judgement, for the final bloody showdown.

I did notice and this participation is welcomed (who are you outside
debian BTW?), but don't you think there's a risk of this moving the
project to votes where the options have simply been composed by
*larger* warring factions?

Alternatively, would it make the path of least resistance "ignore
everyone else whenever possible because they'll never get 30 or 60
DDs together"?

I wonder if there is a need for a more radical reform where once a
proposal reaches the required number of seconds to trigger a vote, a
ballot jury takes over and runs a vote which includes the proposal,
the status quo, and a set of alternatives where the whole population
would vote *for* at least one option, and reasonable combinations
thereof.  Isn't that how Condorcet voting works best?

> If 30 people can agree that all their concerns have been met, that
> certainly adds orders of magnitude more thought about the issue
> than if just 5 had thought about it, which greatly increases the
> chances that it might have actually covered the concerns of the
> people who will first hear about it when the call for votes goes
> out.  And isn't that what we really should be trying to do with
> every vote?  Most parliaments won't put an issue up for vote if
> it is not clear they have the numbers for it to carry.  It wastes
> time that could be used for other things people do agree on and is
> just a general embarrassment to the proposers.

Who claimed earlier that comparisons to government were misleading? ;-)
We're also dealing with non-voting and lack of encouragements to
participate in pre-vote activities (actually active discouragements of
participation IMO) in a way that isn't often a problem for elected
members of political parties.  It's quite easy for a straw poll of DDs
to suggest overwhelming support for one view and then another view to
be more strongly represented if it comes to a crunch like a GR.

> When was the last time we had a vote that didn't leave some relatively
> large proportion of developers feeling deeply unhappy about the result?

Last April's DPL vote, just two votes ago.  (I'm not sure how the
Project membership procedures vote was perceived in general, because I
was not following project news and lists very closely when it closed.)

> [...]  I'd have no problem with us
> preemptively applying the new conditions of this proposal to the vote
> for the proposal itself (it's a superset of the existing rules so we
> could easily do that by consensus with the final proposer/seconds).
> That seems like an easy first test to shake out the fear from the
> genuine uncertainty.  If it results in an even better proposal than
> what the first 5 people agree on that would certainly give me some
> early confidence that we've really done something intelligent and
> valuable here.  If it falls down under it's own rules, maybe we'll
> learn something useful from that too.

That's a great idea.  Actually, it would be interesting to see how
many DDs *in total* would second this proposal, compared to the number
who vote for it.

I welcome any feedback on the concise list of concerns above, other
examples of requiring equivalent of 30 DDs, or any other questions,
but I may take some time to reply from now on.

My Opinion Only: see http://people.debian.org/~mjr/
Please follow http://www.uk.debian.org/MailingLists/#codeofconduct

Reply to: