The 2014 OpenStack Individual Member Director Elections and Red Hat

tl;dr – the affiliation limit means that at most one of the two Red Hat affiliated candidates can be elected. The cumulative voting system makes it likely that both of us running seriously damages both of our chances of being elected. A preferential voting system like Condorcet or STV would not have this problem.

At Red Hat, those of us who contribute to OpenStack take very seriously our responsibility to put what’s good for the project first and foremost in our minds – to wear our “upstream hat”, as we like to say. That’s especially true for me and Russell Bryant.

However, now that the candidate list for the 2014 OpenStack Individual Member Director Elections has been finalized, we find ourselves wrestling with the fact that Russell and I are both running as candidates. Two aspects of our election system make this a problem. First, the cumulative voting system means that those who would be happy to vote for either me or Russell are forced to choose between us – essentially, we are damaging each others’ chances of being elected. Secondly, the affiliation limit means that even if we were both lucky enough to receive enough votes to be elected, one of us would be eliminated by the limit.

The combination of these two issues means that we have to factor our affiliation into our decision. The rules place affiliation front and centre in election system, even though Individual Member Directors are not elected to represent their employer.

Now, I’m personally guilty of not pushing this election system issue hard enough over this past year. At one point I favoured experimenting with a tweak to the cumulative system over a more dramatic change because I found the prospect of getting a majority of over 25% of our enormous electorate to vote in favour of a change so daunting. I want to be completely open about this decision we now face because I want to help raise awareness about how important an issue it is.

Given that Red Hat is a Platinum Member and has a automatic seat on the board, the options we’re weighing up are:

  1. Continue with both Russell and me on the ballot, accepting the risk that we’re damaging each others chances.
  2. I or Russell remove ourselves from the ballot, giving the other of us the best possible chance of being elected.
  3. Brian Stevens steps down from the board and I or Russell takes his place, giving whichever of us remains on the ballot the best possible chance of being elected.

It’s not an easy decision. We both feel we have something to offer on the board. Both of us would be very proud to be elected to represent the Individual Members. Both of us feel that Brian Stevens (our CTO who we greatly respect) is the best possible representative for Red Hat on the board.

We will make a decision on this before the election, but right now we don’t see any of the options as being particularly better than the others. But, at the very least, I hope everyone will find this useful as a concrete example of why our election system needs to change.

6 Responses to “The 2014 OpenStack Individual Member Director Elections and Red Hat”

  1. Col says:

    Why does it need to change? Why should Red Hat (or anyone else) be allowed to have as many representatives as it likes? It’s a pretty common scenario in software community elections, and most companies manage to figure it out easily enough before letting employees put their names forward, not afterwards.

  2. john says:

    I totally agree with Col above.

    And in general when such rule is in place, you make compromise. Neither of you wants to give up to be on the director board and that’s the issue. Take turn. Respect each other. Internally you Redhat has to vote on what to do with OpenStack and have your director representative represent RedHat in the director board.

    Changing the system to allow two? What if there are more than two from the same company? I think the system is fine. I am sorry, but your argument is very weak.

  3. markmc says:

    Colin – the main thing the Board has been working on changing is the voting system. See https://wiki.openstack.org/wiki/Governance/Foundation/ElectionsCommittee

    The cumulative system sucks in this situation because my running likely takes votes away from Russell and, likewise, Russell’s running likely takes votes away from me.

    A preferential voting system like Condorcet or STV wouldn’t have this issue. We could both run without (substantially) damaging the chances of the other.

    I mention the affiliation limit because, without the limit, at least the risk of damaging each other might be worth taking – we could hold out the hope that we’d both be elected.

    So, yes – we have to figure it out between ourselves before the election. But we’d much rather the individual members had the chance to express their preference rather than us guessing their wishes.

    I think it’s unlikely we’ll change the affiliation limit any time soon, but bear in mind that the individual member directors do not (and should not) represent their employers. I do not consider myself a Red Hat representative on the board.

    And to be clear – these issues don’t only apply to our situation. I’m sure there are others who are equally as affected. We’re being open about this in order to help raise awareness about the issues with the voting system.

  4. John says:

    You just deleted my message when it clearly wasn’t a spam.

  5. markmc says:

    john – sorry, I didn’t delete your message, I just didn’t get a notification so never approved it

    Hopefully my response to Colin helps – the main issue is that the voting system makes it damaging to both of us if we both run. Preference based voting systems would mean it’s fine for both of to run and let members express their opinions more fully on us as candidates.

    Also, to repeat – the affiliation limit does tend to suggest that individual members represent their employers, but that’s absolutely not the case. We see ourselves as representatives of the individual membership.

  6. […] The 2014 OpenStack Individual Member Director Elections and Red Hat […]

    [WORDPRESS HASHCASH] The comment’s server IP (67.192.46.14) doesn’t match the comment’s URL host IP (174.143.194.225) and so is spam.