Director-less issues

[danield06]

Posted at https://lists.w3.org/Archives/Public/public-w3process/2020Jan/0000.html

copied here: Hello all,

the fate of W3C without Tim has been on my mind long before leaving W3C, so I'll give my 2 cents on the recent proposal.

I'm looking at

https://services.w3.org/htmldiff?doc1=https%3A%2F%2Fw3c.github.io%2Fw3process%2F&doc2=https%3A%2F%2Fw3c.github.io%2Fw3process%2Fdirector-free

I'm concerned by two things:

• the added complexity with two new "committees"
• the power now residing in the CEO hands.

On the second point, there are now hundreds of references to the CEO and Team as decision-makers and with the Team under full control of the CEO, this effectively puts all the process decisions into one person's hands, which could become an issue in the future if this person is not able to handle all this power correctly.

I think we should try to categorize the current Director's functions along their level of Web technicality, draw a line, and give the most technical pieces to someone else than the CEO. Not sure to whom, maybe not a single person. You could argue that the CEO could delegate these (or whatever) pieces and create such a Technical Director function but I'd rather see this implemented transparently based on due process.

On the first point, new committees, if there is such a TD function, then no need for a new W3C Council providing a higher authority for some of the ex-Director/new CEO decisions. Given the AC appeal mechanism already in place, and if the main technical items are separated, this should not be necessary.

Same idea for the new TAG committee created to select the "Director" TAG seats: it should not be necessary with a appealable Technical Director function selecting them. And it would be better than a pure TAG co-optation procedure..

So IMO having ome sort of Technical Director function, or a Web Architectural Board, would effectively solve both the CEO power concentration and the added committees issue. Maybe this function could be implemented by a trio: one staff, one TAG, one AB, selected by each constituency for a given period, or maybe just one person, e.g. the chair of the TAG (since it's for Technical stuff). Or maybe by one TAG and a W3C CTO (from staff).

Anyway, I also think someone should be assigned to do a quick external study on how other relevant SDOs implement their own Technical Director function/Arbitrage, and also how organizations (non-profit or commercial) deal with their structural issues when the Creator (of the technology and the organization) leaves.

Take care.

[michaelchampion]

This is similar to #316 and #331 . Not being on the AB anymore I'm not sure how much consensus there is for the W3C Council proposal, but I urge the AB to engage the larger community in this important topic before locking down a Process proposal. The ideas of a Technical Director (elected for a limited term or hired by the Board of Directors with no expectation that it is a "benevolent dictator for life" appointment) or fairly small Architectural Board (elected or chosen by some mechanism similar to that IETF uses) deserve to be explored in some depth.

I strongly agree with @danield06 that "someone should be assigned to do a quick external study on how other relevant SDOs implement their own Technical Director function/Arbitrage". @mnot has contributed thoughts about IETF's structure in this thread that can be explored. I believe @tantek recently attended an ECMA TC39 meeting and came away with the impression that W3C could learn from their experience. Likewise there was a TPAC 2019 breakout https://www.w3.org/2019/09/18-future-minutes.html where several speakers made comparisons to similar standards organizations (and IIRC OSS projects that touch on the web). Some relevant resources include: https://tools.ietf.org/html/rfc2026#section-6.5 https://nodejs.org/en/about/governance/

IMHO, this is not a task for someone on the Team, but one that needs a broader set of participants from the AB, AC, Team, and external stakeholders. Maybe a Community Group? Or just an issue in this GitHub repo?

[jeffjaffe]

Thanks, @michaelchampion . I'll try to schedule time in the next few months for the AB and other stakeholders (e.g. you, Daniel, Mark, and Tantek to discuss these issues.

[dwsinger]

Thanks for the feedback. At a rough level,

• we were trying NOT to create new committees, so the W3C Council, while it serves a new function, is not separately formed but instead is AB+TAG. (What's the second new committee?).
• we were trying to recognize current reality in what we documented as being team/CEO, for the most part — places where the old process said Director, and the function was either routine (announcements etc.) or required a person (discipline, for example).

We clearly, in your view, went too far in the second. Most trade and standards bodies I know don't have a technical director as such; they expect the members to provide technical direction. Even a body like the TAG is unusual; though the IETF has roughly similar (IAB, Dispatch).

Are there other examples of bodies that formed around a single person and had to work out how to live without them?

[michaelchampion]

Most trade and standards bodies I know don't have a technical director as such; they expect the members to provide technical direction.

It's worth looking at other organizations that don't have a technical director to see how they deal with dissent / non-consensus. TC 39 comes to mind ... https://tc39.es/process-document/

Are there other examples of bodies that formed around a single person and had to work out how to live without them?

The most-discussed example of an organization losing a BDFL is Python after Guido van Rossum stepped away https://mail.python.org/pipermail/python-committers/2018-July/005664.html . See also https://discuss.python.org/t/comparison-of-the-7-governance-peps/392, https://medium.com/@herveberaud.pro/python-after-guido-bdfl-the-future-of-the-governance-of-python-2969bab19c7e

I'm not sure about other cases... is Linus still BDFL of Linux?

There's also https://blog.leafe.com/why-openstack-failed/ , an argument that OpenStack is less successful than it would have been if it had a BDFL

[dwsinger]

I would like this split into specific issues that have more descriptive finer-grained titles.

[frivoal]

Agreed to close in the 2022/10/28 Process-CG meeting.