Let us know how we are doing

Take the DCI Working Group's new survey!

Page tree
Skip to end of metadata
Go to start of metadata
Status

RESOLVED 

Outcome The TSC affirms that there should be no limits or quotas on TSC membership diversity
Minutes Link 2019 10 10 TSC Minutes

During the Hyperledger startup period there was a constraint on the number of TSC members from a single organization or its affiliates to no more than 3 members.  I'd like to propose that the TSC Membership always be constrained such that no more than 25% of the members can be from one organization or its affiliates.  Hyperledger is a project to promote decentralized networks, why wouldn't we want to ensure the technical leadership is also decentralized?  Conway's law states: "organizations which design systems ... are constrained to produce designs which are copies of the communication structures of these organizations."




27 Comments

  1. Thanks Todd for contributing to the debate. Your proposal overlaps others though so I'd like you to separate the different aspects so they can be more effectively managed:

    #1 should go under TSC election - Who may vote ?

    #2 should be put into a new topic called something like "TSC membership limit per organization"

    #3 should go under TSC Size

    Thank you.

    1. Done.  For 2 I just co-opted this page.

  2. I support this initiative. No more than 25% means no more than 2 with the current 11 member structure. 

  3. The simplest approach would be amending section 4.a.ii of the charter with the verbiage dropped from section 4.a.i:

    “… provided that no company (including related companies or affiliates under common control) shall have more than three (3) votes on the TSC.”

    1. I think that wording is very poor though. There are indeed two possible interpretations: One is that no company can have more than 3 people on the TSC. Another is that companies can only have 3 votes, independently of how many people they have on the TSC. Meaning that some members have no votes.

      Admittedly, the former is probably what's being meant but it should then just say that.

  4. This would be an extraordinarily bad thing to implement.

    People are not defined by their employers.  That they may have pressures to vote a certain way by those employers can not be refuted.  But they are on the TSC as individuals, and everything we do should be in defense of their right to serve independently. 

    Implementing a rule like this hard-codes an assumption that duly elected TSC members can not be trusted to vote independently, to resist their employer pressures to vote one way or another.  It changes this separation of concerns from a first principle worth fighting for, to its opposite.

    It also creates a ton of uncomfortable edge cases:

    • how do we count subsidiaries, like Red Hat, who operate independently?  How do we even know who owns private companies?
    • what happens when a TSC member changes jobs to one who already employs 3 members?  Do they have to no longer be a member?
    • do we increase this number if we increase the size?

    I'm sure there are more.

    I can't think of a single example of something like this from other mature open source communities.

    Let's figure out approaches that enhance the ability for TSC members to act independently and be accountable to the public, not to their employers.


  5. To add to Brian's point above, I think implementing this would require companies to have some level of internal coordination, as to who should run, which we do not have today.

    Indeed, contrary to what some might think, the result of the latest election is due to the fact IBM did NOT coordinate. We let anyone decide for themselves whether to run or not and I literally discovered who was running when the nominations got disclosed, along with everyone else.

    I think this would effectively make people on the TSC company representatives, which I don't think is what we want.

  6. As much as I appreciate Brian's comments, I think his view is extremely naive at best.  To suggest that Hyperledger figure out approaches that enhance the ability for TSC members to act independently and be accountable to the public, not their employers is nearly laughable.  Employers pay people to act on their behalf.  Hyperledger doesn't pay any TSC members.  TSC members almost certainly perform their duties on their employer's time.  As such TSC members will always be accountable to their employers first and foremost and to the public to the extent their employer allows.

    As to Arnaud's comment, I never meant to imply that there was any collusion on anyone's part.  I simply point out that at the formation of the TSC it was felt necessary to have diversity, so I have to ask, what changed after the startup period that makes the limitation "extraordinarily bad"?

    1. Todd Little, I understand from your TSC nomination that you've doing software work for 40 years, but it's not clear how involved you've been in open source communities before this one.  The expectation that participants here (TSC candidates, but not just them) represent themselves and not their employers is not "naive", it is standard for this industry.  It is also normative - it is of course possible that if you had been elected, you would only vote as per Oracle's interests, and that would be well within your rights to do, as there's always the chance your opinions align with Oracle's interests.  But it would be wrong for people to assume that's the case - wrong for them to even vote for you as an Oracle employee rather than to vote for you for your contributions.  This is not just because if you left Oracle, you'd get to keep your seat on the TSC.  It's also because to whatever degree your opinions end up disagreeing with Oracle management, we have your back, and you don't have to burn your personal reputation pushing for something that you don't believe in.  If Oracle doesn't allow you to participate in the Hyperledger TSC on company time, that's between you and them.  I know many people who participate in open source projects without involving their employer. 

      1. Brian Behlendorf Sure, to some extent, but individuals with similar backgrounds (such as all working on the same project for the same company) are going to come to very similar opinions on most topics. Which is why diversity in general is so important.

        1. Diversity is highly desireable, and is something voters can take into account when casting their ballots.

  7. The issue is best left to IBM to address - to keep the their current TSC members because of the organization's rules and because the people will be effective in their elected role or decide that it's better for the community that some should step down to allow others space on the TSC.  I have seen a negative reaction from a couple of people outside the Hyperledger community to IBM's perceived role ("dominance") in Hyperledger. I don't think that is a commonly held view, but if it is, this election likely hurts Hyperledger overall more than the contributions of the individuals, and IBM might want to address that.

    If the rules are to be changed, I would definitely not want an organization to limit individuals from running, but rather limit those appointed to the TSC based on the election results. How an organization chooses who runs is up to the organization - grassroots or internally agreed upon - is up to them.

  8. I'm more concerned about the project diversity of the TSC than the company diversity.  How many TSC eyes are on Iroha or Burrow, for instance?

    If we focus on core DLTs (and not cross-project things like Ursa or Transact), I only see two TSC members who have non-Fabric contributions:  Dan and Nathan.  Is this enough?  If we are going to try to modify the TSC composition to enforce some diversity, I'd much rather do it at the project level by letting the maintainers of active, post 1.0 projects that are large enough pick a TSC member(s) according to some formula.

    1. I stand corrected:  Mark Wagner has contributed to Sawtooth.  Sorry for the mistake!

      1. Technically, but to the point you were trying to make, there is only a single person involved w/Sawtooth on the current TSC. Though, I see him more voting with the Ursa block. :) I don't think you should necessarily exclude Ursa and Transact, even for the sake of discussion. But even if you include them, the question is whether the Fabric team can make all the decisions, which is clearly the case. Which means a year (or more???) were there will be less debate while we just hope the decisions don't actively hurt the under-represented projects.

        On a more positive note, the increased developer/maintainer representation is a good thing.

        1. I have been involved with TSC meetings and have contributed to the debates since the first TSC meetings and will continue to do so, for the foreseeable future. In spite of never having been elected to the TSC.  Based on my observations, I do not think there will be "less debate".  We have to wait and watch to see whether the TSC is swayed by the technical merits of the debate rather than voting in lockstep.

        2. The people from the last TSC who are not on this are : Binh, Mic, Baohua, Silas, Kelly- All developers and maintainers. I dont see how this is proof of "increased" developer/maintainer representation.

  9. Whilst it is good to get diversity of views on the TSC, the current governance and election processes do not make that easy.

    To give Fabric its due, it is the project that has the most number of contributors; IBM has contributed mightily to Hyperledger. However, Hyperledger is more than Fabric. 

    There is a possibility of groupthink (all maintainers/developers, mostly associated with one project or company). There is common culture that further skews views (company culture etc.)-  There is also a question of public perception (i.e. outside Hyperledger) about the fairness and decentralization of the TSC.

    My contention is that 30% electoral participation amplifies this phenomenon; as committed and  organized (on the election) groups skew the electoral results; in the case of the 2019 elections, allegedly fair and square. I say allegedly as there is no transparency of results and Condorcet is remarkably opaque and quite convoluted.

    I believe the truth lies somewhere between the pure view that Brian espouses and the view that employers control individual views.

    As one of the activities for the next election (and through the year), I suggest that we need to raise this turnout. Low turnout can be a result of apathy, indifference or a sense of futility (according to studies). Futility is the most worrisome one, as it is a sense that one's vote will not count.  The 80-20 rule has been cited to defend the low turnout. I do not buy this argument, unless there is data backing this.

    There are standard techniques for raising turnout. Engagement through the year is important as we must work on the lists, the reachability etc. through the year. Already there are many who are eligible. 

    Giving enough of a leadtime to work on the lists is important as time pressure and other factors contribute to the general feeling of urgency and hence difficulty in organizing  the elections. The fact that a single person has to run this election does not make matters easier.

    In short the proposal is have techniques to raise the turnout.



  10. Enforcing diversity is not the right solution to get to a diverse community. This will alienate people from contributing and will make people wonder if they were elected for the sake of diversity or because they were actually elected. Is this really any different from reaching quotas in our goal to reach gender equality in technology? I have heard from a number of white males who feel disillusioned with their ability to advance in their careers and from a number of females who feel like they were only promoted to reach some quota (even if they rightly should have been promoted).  

    What other steps can be taken to get diversity in our community? Diversity in the community will lead to diversity in the TSC. 

    1. Tracy, these arguments reflect those happening in the broader world.  They are  valid.

      However, the question is much more focused in this proposal. Should there be a limit on how many people from the same organisation are on the TSC?  5 or 6 out of 11 may constitute a majority; what  if it becomes 11/11. Will it still be OK?

      There were limits to the number of people from one organisation on the TSC (which seems to have been in the initial charter). There are precedents: in the US and elsewhere: term limits, steps to prevent domination by a single group (like the composition of the US Senate, which may of course swing in the opposite direction- i.e. tyranny of the minority).

      The solution has to be looked at carefully for unintended consequences, which is easier said than done (the word "unintended"), let us be balanced in our appraisal of the problem and the adoption of the solution.

      The broader question of gender/geographic/racial diversity needs to be addressed, maybe we should put that up as a different topic. 


    2. Although I agree with Vipin that there are different dimensions to the diversity question, I think these are very important points and I'm glad you brought them up.

      Do you have any suggestions on what to do though? I suspect most of us if not all are very open to doing the right thing here but what does that translate to?

      1. Tracy's comment is :" Diversity in the community will lead to diversity in the TSC"...

        What are some of the concrete steps to achieve diversity in the community? I can think of a few. However, we may have to move this to another  Topic or under the DCI WG

        Geographical: Having hackfests (or bootcamps) in different areas of the world is a great start; However we need data about the continuing involvement from these locations; also to see how we could remain engaged. Are there such numbers available from Tokyo, Brasil, Other areas?
        Get involved in ID2020 or some other group which is global in scale; just showing up to the Summit is not enough; there are HL connected people at the top levels (David Treat/Alissa Worley from Accenture- Blythe Masters our previous chairman is the chair of ID2020 etc.)  

        Gender: In New York I am a member of women in blockchain, very powerful group- I know some people from Hyperledger are invoolved; but we should be evangelizing; we know that some of the leadership in this group is very focused on "Decentralization" Now that Besu is formally within HL maybe we can persuade them to have a talk on that.

        Racial: I live in Harlem, I will hook up with a group like Silicon Harlem and persuade them to host a HL talk or workshop; There are many other opportunities in different parts of the world. We should spread the word and have them contribute. 


  11. Getting back to the concern regarding over representation from a single entity, I first have to repeat that this is somewhat of a moot point because looking at the history of TSC decisions one can clearly see that this wouldn't have mattered in any way because the vast majority of the votes were unanimous.

    But for the sake of trying to find a middle ground I just thought: How about setting the limit to 49%? This would ensure no single entity has full power which would address the crux of the matter.

    Practically speaking though, this is not without some challenges either. It requires a clear definition of a "single entity" and a way to determine which entity a person is affiliated to, just to name a few.

    In W3C people are required to disclose their affiliation. As far as I know we don't have anything like that. This would then become a requirement to be able to implement anything like that. At least for those who run for the TSC.

  12. Another comparable:

    The CNCF has a rather... elaborate TOC election process that allows the projects, the "end-user" community, and the Governing Board to appoint seats:

    https://github.com/cncf/toc/blob/master/process/election-schedule.md

    I can see how such a process might be an improvement over now, at the cost of considerable complexity.  Thoughts?

  13. I understand that only about 20% of members voted.  The easiest way to influence the composition of the TSC is to ... vote!


    1. I agree, 34% voted, still low. Is your exhortation to the commentors on this page? I suspect that they have all voted.

      This is why Get Out The Vote (GOTV) activities are important. Brian Behlendorf had a set of concrete suggestions.

      Back to whether we need a limit on TSC members from a single organisation.

      To recap:

      • Brian says no since every member is a free agent and is elected purely on technical effectiveness 
      • Todd says yes, since TSc members views are circumscribed or even dictated by the company they work for, so limit to less than or equal to 3 members from a total of 11
      • In practice from data, (for example the choice of vice chair last year) company plays a role. Commercial considerations and familiarity with solutions also create platform specific views
      • A number of others agree broadly with Brian with some other dimensions,
        • there is an externally perceived bias, that could reduce the effectiveness of Hyperledger in the external world.
        • The atmosphere is collegial enough that most votes were unanimous
        • Ways to influence the debates and technical direction through open participation in debates (not just being in the TSC)
        • Limiting number of people from the same org. has advantage of good governance practices (diversity in orgs) and external perception

      Actual mechanics of how to do it (this is my suggestion):  the electoral officer looks the Condorcet ranking and chooses the winners based on a. ranking b. skip if company already fulfilled limit ; this would be completely private (as private as it is today)