On Mon, Feb 25, 2013 at 11:58 AM, Chip Childers <chip.child...@sungard.com> wrote: > On Fri, Feb 22, 2013 at 04:23:41PM -0500, David Nalley wrote: >> On Fri, Feb 22, 2013 at 3:29 PM, Chip Childers >> <chip.child...@sungard.com> wrote: >> > Hi all, >> > >> > Continuing the graduation discussion (you're going to see plenty of >> > threads like this as we head down the road), the PPMC had a discussion >> > around how to best select the name we would recommend to the ASF board >> > to be our PMC chair. >> > >> > Currently, our bylaws [1] state that we would use the Single >> > Transferable Vote method for this selection (see section 2.4.5). In the >> > thread on the private list, our mentors have suggested that we may want >> > to reconsider this position. >> > >> > I'll admit that we have it in our bylaws, purely because I started with >> > the Hadoop project's document as a starting point for our own. I >> > believe that the Hadoop folks probably have a very good reason for this >> > approach, but perhaps it's not right for our community. >> > >> > I'm proposing the following change: >> > >> > CURRENT: >> > >> > 2.4.5. If the current chair of the PMC resigns, the PMC votes to >> > recommend a new chair using Single Transferable Vote (STV) voting. >> > See http://wiki.apache.org/general/BoardVoting for specifics. The >> > decision must be ratified by the Apache board. >> > >> > PROPOSED: >> > >> > 2.4.5. If the current chair of the PMC resigns, the PMC votes to >> > recommend a new chair using a lazy 2/3 majority voting method. >> > This vote would be held on the cloudstack-dev mail list, after a >> > discussion is held on the cloudstack-private list to nominate a >> > candidate for the role. The decision must be ratified by the Apache >> > board. >> > >> > This is the discussion thread to see if people have any opinions. If we >> > don't have any big concerns, I'll start a new VOTE thread to change the >> > bylaws (using the rules in section 3.4.9 of the bylaws themselves). >> > >> > Thoughts? >> > >> > -chip >> > >> > [1] >> > https://cwiki.apache.org/CLOUDSTACK/apache-cloudstack-project-bylaws.html >> >> >> >> So here are my thoughts: >> 1. Chair is appointed by the board, generally upon recommendation of the PMC. > > True. Even as a TLP, that remains the case. > >> 2. There is a high likelihood that we'll continue to grow and might at >> some point have contention for the Chair, even if it is friendly. >> Discussing people (IMO) should not happen in public. There's a reason >> we already do that for committers in private. Think of some of those >> conversations that we've already had that simply have no place in >> private. And I know your proposal calls for discussion in private, and >> vote in public but here's the thing: >> >> The decision is one for the PMC to make. >> You're effectively 'voting' by nominating someone and pushing them >> forward - which means the 'vote' will be more of voting theatre than a >> real vote, and really more of a ratification of the decision made in >> private. > > Yes, Sebastien made that point as well. I fear I missed the mark with > the specifics of this proposed change. > >> Additionally, only votes by the (P)PMC are binding, and >> unlike technical discussions, it would be bad form to discuss people >> in public. So I am personally struggling to see the benefit afforded >> by 'voting' in public. You can disagree, but if there is dissent it >> being public would potentially be deleterious to the person it is said >> about. >> >> I can see doing away with STV (or only resorting to it if clear >> consensus isn't achieved). > > I'd like to modify my proposed change, but I think we still have issues > to discuss with this modification: > > 2.4.5. If the current chair of the PMC resigns, the PMC votes to > recommend a new chair using a lazy 2/3 majority voting method. > This vote would be held on the private mailing list, after a > discussion to nominate a candidate for the role. The decision must be > ratified by the Apache board. > > The problem I see with this new version, goes to the point you > made about contention: What happens if the discussion doesn't lead to > a single candidate? Do we opt for STV as a backup process? In that case, > does it even make sense to have that lazy 2/3 majority statement above? >
So STV has a lot of good things - but really seems (IMO) best suited for large groups with multiple candidates, with multiple good choices. > The issues we have to balance are (1) fairness, (2) being as open as > possible, (3) respect for privacy for individuals. > STV does a good job of 1 and 3. Lazy consensus does a decent job of 1 and a good job of 2. > I'd also point out, as one of our mentors did in the private list > discussion, that the chair role is largely an administrative one. > > And while we're at it, the private discussion brought up the idea of > re-introducing a rotation (or at least a "term"), which would allow > for a regular opportunity to recommend a change to the board. I'm not > sure that it's required, given the nature of the job... but does anyone > have an opinion? > > -chip I'm personally 'meh' on this. As you noted the role is largely an administrative role than fearful power and authority; and I am inclined to have a person, particularly if they are bearing the job well, to continue with it until such time as they no longer wish to bear it. I do understand the point that 'regular' rotation can provide the perception of the project not being 'controlled' by a single entity. >From inside the project, I don't see that, but perhaps outside the project that is an important consideration. --David
