accumulo-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Sean Busbey <bus...@cloudera.com>
Subject Re: [DISCUSS] Majority voting without prior discussion
Date Wed, 07 May 2014 16:07:17 GMT
We should not require an explicit discussion period prior to a majority
vote, especially in our bylaws. Discussion and conflict resolution should
happen as a part of our normal community interactions. If these things are
not already happening, a mandated warm up period isn't going to fix
that. Procedures are no substitute for community.

Generally, vote callers should have an easier time if they try to lead a
discussion period first. Certainly if there isn't consensus over the
fundamental matter of the vote, a DISCUSS thread is preferable to using a
VOTE thread for discussion.

However, there's no way to ensure that all concerns get hashed out prior to
a vote. It is the responsibility of each person who votes to make a good
faith effort. For those against that means explaining their concerns and
how they can be met, especially during consensus votes. For those who are
for the proposition it means attempting to address the concerns of those
against and they must be particularly mindful in majority votes.

While reflecting on how to phrase this message, I kept coming back to the
ASF reasoning on voting[1]:

> Reasons for Votes
>
> People tend to avoid conflict and thrash around looking for something to
substitute - somebody in
> charge, a rule, a process, stagnation. None of these tend to be very good
substitutes for doing the hard
> work of resolving the conflict.

If more discussion is needed or modifications to the proposal are
necessary, we always have the option of failing the vote and trying again.
There's no need to add rules to draw the vote out either though altering
vote periods or mandating a discussion period. It should be sufficient for
concerned individuals to include the need in their vote of -1. A well
functioning community should be looking for consensus. If you can't get
enough people to join in on voting for more discussion, that discussion
isn't likely to result in consensus.

I would like to see us gain more rigor around sunsetting major releases. I
don't think adding an additional vote type is the way to go about that.
Instead, I'd like to see the discussion around how we're going to handle
things in 2.0.0+ include setting up the whole lifecycle for major release
development around release planning.

-Sean

[1]: http://www.apache.org/foundation/voting.html


On Tue, May 6, 2014 at 6:49 PM, <dlmarion@comcast.net> wrote:

>
>  Your comments suggest that we need to tailor our model to follow the U.S.
> Senate. We will need a vote to end debate, so that we can vote on the
> measure. Can I filibuster? Just kidding, I wouldn't wish that on anyone.
>
>  In all seriousness, I agree with your statements. I did the same thing
> with the blog thread, discuss, gather feedback, vote on text that was
> agreed upon. I went back and looked at the bylaws, which specify majority
> approval for ending a planned release and for an official release. What if
> they were consensus approval instead? Or, maybe a modification to the
> bylaws that states certain types of votes cannot be started without a
> discussion first.
>
> -----Original Message-----
> From: Christopher [mailto:ctubbsii@apache.org]
> Sent: Tuesday, May 06, 2014 4:59 PM
> To: Accumulo Dev List
> Subject: [DISCUSS] Majority voting without prior discussion
>
> Devs,
>
> As something that came out of the vote thread about EOL'ing 1.4, I was
> thinking:
>
> The purpose of a majority vote seems to be when we've already discussed
> and planned, and we just need things to come down to a final vote. Things
> like releasing, for example, occur after discussions, planning, and aren't
> a surprise in any way. It seems to me that there are two main points I want
> to make:
>
> 1) Prior discussion/planning should be a prerequisite for things which are
> majority vote.
> 2) The default for any ambiguous or arbitrary vote item that does not fall
> into a predetermined type, should require consensus.
>
> The problem with majority votes without discussion is that there may be
> serious concerns a minority of persons voting have about something, that
> could be resolved with compromise.... where there is plenty of room for
> gathering consensus. Coming together as a community to move forward with a
> mutually agreed upon path should always be preferred where possible. In
> some cases, differences are irreconcilable and action just needs to be
> taken to move forward (releasing, for
> instance) on a majority decision, but even here, there is up front
> discussion about those differences (code development, release planning,
> etc.) prior to such a vote.
>
> Binding actions to a majority vote that has insufficient prior discussion,
> especially when there is no mechanism to extend a vote, or sane way to
> alter the contents of the majority vote while in progress, leads to actions
> that don't have the consensus of the community, even in circumstances where
> consensus was possible to achieve.
>
> I think our bylaws should be updated to reflect the two ideas above.
> I'm not sure the exact wording needed *(please submit proposals in
> response to this), but I think it should declare that any voting that does
> not clearly fall into a vote category explicitly enumerated, or if there's
> any doubt, should default to consensus. Before we had bylaws, this appeared
> to be the precedent... as we often took great care to respond to any
> objections, delaying, canceling, or extending the vote to do so. We should
> continue to operate with that same sense of community in future decisions
> as well, and I think consensus voting whenever possible is the way to do
> that.
>
> It was also discussed that it may be helpful to enumerate end of life
> procedures in the bylaws as well. I'm not sure this is as important of an
> issue if we agree that the default should be consensus... but I'm willing
> to entertain that discussion in this thread as well.
>
> Thanks for your time and input.
>
> --
> Christopher L Tubbs II
> http://gravatar.com/ctubbsii
>
>


-- 
Sean

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message