w3c / process

W3C Process Document
https://www.w3.org/policies/process/drafts/
190 stars 128 forks source link

Recall procedures for TAG and AB #882

Open mnot opened 4 months ago

mnot commented 4 months ago

The Board has a removal procedure, but the TAG and AB do not, beyond the CEO's ability to discipline them.

Given the direct process role that they now play when Councils are convened, it seems appropriate to make the TAG and AB more accountable to the Membership by defining a procedure for recalling them.

Straw-man proposal: a recall can be instigated by a demand signed by 20 Member representatives, and it's executed as a requisite member vote.

frivoal commented 4 months ago

Only commenting on a side aspect of the proposal for now, rather than on the core question, but invoking Bylaws concepts like Member representatives and requisite member votes for groups like the AB or TAG which are anchored in the Process is weird, and creates inter-dependencies that I'd rather avoid. If anything, we should go with AC Reps.

mnot commented 4 months ago

Yes, that bugged me too, but those procedures / definitions seemed more appropriate. Maybe I've just had my head in the bylaws too much lately.

In the longer term, rationalising the bylaws and process should make such things easier (I hope). In the meantime, let's get agreement in principle on the parameters first, and worry about implementation language later.

dwsinger commented 4 months ago

one place to look for similarity or grounding might be the appeal procedure in the Process?

frivoal commented 4 months ago

one place to look for similarity or grounding might be the appeal procedure in the Process?

Yes, I was thinking about that too: https://www.w3.org/2023/Process-20231103/#ACAppeal seems like a workable basis. I'm happy to try and draft something based on that. TLDR: it's a two step process, where you need to convince 5% of the voter base that the question is worth taking up to have a vote, and if you do, you hold a vote during which you need to reach majority for the proposal to pass.

Incidentally, both for this recall procedure and for the general AC appeal, I think we could/should borrow the Bylaws's notion of a threshold that varies based on quorum. passing drastic things by simple majority without a quorum requirement worries e a little bit, even if there's not precedent of it being an issue. Maybe I should open a separate issue about that?

frivoal commented 4 months ago

See https://github.com/w3c/process/issues/886 for a discussion of adopting voting thresholds along the lines of what the bylaws do for requisite member votes.

chaals commented 4 months ago

Addressing the substance, with thoughts that haven't crystallised for me yet:

I'm uneasy about this in general. I understand the need for accountability, but I don't think single-seat recalls provide a very good version of it for a representative body like TAG or AB.

I'd perhaps be OK with a "nuclear spill" option resulting in the whole AB or TAG being put up for election. I'm also leaning to suggesting that we introduce effective good standing requirements for AB and TAG. (I was a proponent in removing them from working groups, where they had become rarely-used to the point that enforcing them was likely unfair in practice).

There's already a lot of asymmetry in the way W3C works, given it already institutionalises machtpolitik by putting a premium on mass adoption in an ecosystem where there are a few superpowers. I am not very concerned about those superpowers behaving badly. But I am concerned that we'll end up motivating others to do so.

Being able to recall a single individual from the fairly small AB/TAG, presumably on the basis of supporting a position that a small proportion of members strongly disagree with, seems like a tool that invites gaming, and something that makes it less attractive for good-faith actors to stand in the first place, and places a higher price on them acting as they see fit.

IMHO (and I realise people disagree with me) we mitigate some of the asymmetry through the use of STV to elect these increasingly powerful councillors, and by holding consensus as a core value for those bodies. Lowering the barriers to enable decisions to be forced more easily, instead of having to slog through more consensus-building, seems to me a step in the wrong direction.

I'm trying to think through controversial things W3C has done (or decided against), and how a recall procedure would have affected the outcome. My initial guesses suggest more political game-playing and resultant rancour, for no clear practical benefit.

I'd like to see some arguments that explain how this would work in practice and how that would be an overall improvement.

dwsinger commented 4 months ago

I tend to agree with Chaals.

Also the impact on the person concerned of even having a recall vote issued against them, let alone having it succeed, could be a large mark on their career. I have a hard time imagining it would or should ever be used.

I suppose we could have a "vote of no confidence" in the TAG or AB as a whole, which triggers a "general election", but I cannot imagine a time in the past where that would or could have been used (and the efficacy of both bodies has gone up and down over the years).

cwilso commented 4 months ago

I'm somewhat ambivalent about recall procedures; I think they are a nuclear option that are unlikely to be applied, or at least applied repeatedly or without serious reason to be applied. I wouldn't object to any of the options suggested.

My disagreement with you, Chaals, is that STV is increasingly excluding those "superpowers" (as it prioritizes diversity of opinion, those affiliated with centrist "superpowers" are unlikely to garner enough votes to get elected), and that increase in diverse representation has sadly turned into consensus-blocking in practice. (I.e., significant lack of doers, significant surplus of objectors). This has diminished the value of those bodies, in my opinion.

As for the impact on someone of having a recall vote issued against them, frankly we do not expose enough real information about participation to initiate recalls on anything but political grounds, and as someone probably more likely to have such a recall issued against me, I don't see it as any more embarrassing or career-marking than our election mechanisms already are, so I don't think it's of great concern.

I will note as I did originally that the "discipline" the CEO has at their disposal already includes the power to remove AB and TAG participants:

The CEO may take disciplinary action, including suspending or removing for cause a participant in any group (including the AB and TAG) if serious and/or repeated violations, such as failure to meet the requirements on individual behavior of (a) this process and in particular the CEPC, or (b) the membership agreement, or (c) applicable laws, occur. Refer to the Guidelines to suspend or remove participants from groups.

frivoal commented 4 months ago

Personally:

I don't have a strong view on individual vs collective recall. Collective is likely less prone to abuse, but also fairly blunt.

chaals commented 4 months ago

TL;DR: I've thought a fair bit more. I'm more opposed to individual recall, more supportive of Good Standing requirements but with caveats about how they are implemented, and ambivalent about total spill mechanisms that would re-elect an entire body.

Recall

I've spent a collective decade+ on the AB, and have been involved with W3C since before the AB came into existence (and ran for the first TAG) seen situations where I could easily imagine a recall. More importantly IMHO, I can think of many situations where someone on the AB or TAG would have been reluctant to take a principled position given the threat of a recall. Some of those positions may have been frustrating - which is how a recall would be justified - but some I think were actually important.

The way W3C seems to work, if there is someone prepared to take the opprobrium for starting an effort as divisive as a recall, then it suddenly becomes normalised behaviour - may who would never step up and put their name on the initial call will justify supporting it. There are hundreds of members, any of whom can decide to take that initial step on, for any number of reasons.

Finally, although we have a reasonable representation of the membership's diversity of views on those bodies, they are relatively finely balanced. I think the impact of having a recall procedure for individual members, whether used or not, is not helpful.

Good Standing

Recently the AB has been trying to address the problem that sometimes members of these important bodies who just don't respond.

As I noted above, the bodies are elected in a way that makes them somewhat representative of the diversity of W3C membership. A consequence of this is that the smaller the pool for any given election (whether that's from the split-cycle we still have as a hangover from the way we used to elect people, or the more extreme case of a single vacancy) the more the overall results can be expected merely to reinforce the most powerful group (or set of groups for multiple seats), and the less the results are likely to favour candidates with broad support from their "opposition" as well as deep support from those they most closely represent.

W3C used to use good standing very strictly in some groups to determine whether someone was a participant whose dissent affected the measure of "consensus".

I think it makes sense to have a good standing requirement for these elected bodies. However unfair it may seem to a "good individual", I also think there should not be a discretionary mechanism for enforcement - if it applies it should be written in such a way that it automatically applies to any member who fails to maintain good standing.

I also think that an individual vacancy should just be left open, with a special election only taking place if there are three open seats. A major reason for this is that an individual special election although run entirely in good faith can easily produce pretty undemocratic results in the relevant circumstances.

For the AB, which is already happy to have only 9 or as many as 11 members, clearly the specific number doesn't seem crucial. Likewise the Board doesn't seem to have felt a need to reach its full complement.

Full Spill

This is a potentially real scenario. There was a concerted and successful effort to replace the entire TAG, some years ago - it took well over a year to execute in part because it had to be done through the course of two election cycles. I don't know if the people involved would have tried to pull a TAG spill if that option were available, but many of them are still around. We could ask them. (It's probably instructive to see who says they were actively part of that...)

If there is a situation where one of these bodies loses the confidence of the membership, it seems reasonable to allow a complete spill. This means that all seats would be vacated, and the entire body is re-elected in one go.

A key difference with a recall is that while in practice many people will blame the problem on individuals in the relevant body, there is no formal mechanism that will historically reinforce the suggestion that it is one person's fault the body failed to fulfil its function of providing consensus-backed guidance.

It's also fairly resistant to gaming: given that the outcome would be a full election, the bodies will be reconstituted as the most broadly representative they can, in principle sacrificing the value of institutional memory that is the apparent rationale for our split election cycles. A situation a body has "gone rogue" would mean this is appreciated by many, across the various differences of technical and governance opinion that sometimes divide us.

As an effort to entrench or reinforce power for a specific group, a spill is idiotic, and very likely to fail spectacularly with real blowback. The annoyance of this taking place, meaning an extra election - which could be compensated by automatically stating the it replaces the next expected election - and the relevant group not meeting for a couple of months is probably worth the lesson.

Equally, I think we could continue to live OK without enabling this.

css-meeting-bot commented 2 months ago

The Revising W3C Process CG just discussed Recall procedures for TAG and AB, and agreed to the following:

The full IRC log of that discussion <fantasai> florian: Mark filed an issue that there doesn't exist procedures for recalling AB/TAG members
<fantasai> ... this is almost true. There is possibility of removing for disciplinary reasons (defined fairly broadly) by CEO.
<fantasai> ... Board has ability to eject its own members, and Membership has ability to recall.
<fantasai> ... I think accountability is healthy. Especially if mechanism exists and not used, it's a sign of confidence.
<fantasai> ... on the flip side, need to avoid enabling a channel to abuse
<fantasai> ... either of the entire group, or of individuals on it
<fantasai> ... Discussed in AB; balancing of accountability and potential for abuse was one concern
<fantasai> ... Another question was should this be recall of individual members or recall of the entire group
<fantasai> ... No consensus
<fantasai> ... I don't think it's our job to decide what to do, but we could design a few possibilities.
<fantasai> florian: I have a very drafty PR that re-uses AC Appeal mechanism for any identified set of directors, from single one to the whole thing
<fantasai> ... [summarizes PR]
<fantasai> ... added safeguards against too many recalls of same member or same requester
<fantasai> ... alternative could be to do the whole group
<fantasai> ... that's as far as I can go on my own
<cwilso> q+
<fantasai> fantasai: I think individual recall via AC Appeal might be not great, would prefer to have such a mechanism trigger a snap election.
<fantasai> ... for misbehaving members we have CEO ability to remove
<fantasai> ... and we could also add ability for AB to eject its own members by supermajority, similar to Board
<fantasai> florian: Note that Board has a lot of decision-making power, whereas AB only has two actual powers: as Council members wrt FOs, and participating in TAG appointment.
<fantasai> ... so this probably influences how important it is to do recalls
<fantasai> cwilso: I'm increasingly concerned about potential for Denial of Service attacks here
<fantasai> ... it would be easy for a disruptive group to cycle through and keep initiating recalls
<florian> q+
<fantasai> ... though I guess realistically it's relatively well-contained
<cwilso> ack me
<fantasai> ... since requires a high level of support
<fantasai> florian: in terms of limiting abuse, I tried to introduce 2 ways: same Member can't invoke too often; but also regardless of who it is, you can't invoke too often against a single participant
<cwilso> qq+
<JennieM> q+
<cwilso> ack me
<Zakim> cwilso, you wanted to react to florian
<fantasai> JennieM: I have concerns that this could trickle down into WGs
<fantasai> florian: do you have the same concern if it's about recalling the entire group?
<fantasai> fantasai: why concerned about WGs?
<fantasai> florian: rule only applies to elected groups; I think JennieM's concern is that if we apply it, it might tempt people to ask for it to exist on other groups as well
<fantasai> ... and use it as precedent for asking for more
<fantasai> ... I can see this if it's about individuals, but less if it's collective
<fantasai> ... because it doesn't call AB to stop existing, but removing all members and have a new election
<fantasai> ... WGs aren't elected so wouldn't apply there
<florian> q+
<fantasai> fantasai: I think if a member is misbehaving, chances that enough of the AC understands the problem well enough to remove the individual, without either the CEO or the group itself understanding the need to remove that individual well enough to act on it, seems unlikely
<fantasai> ... triggering a snap election against the entire group makes more sense at the AC level
<fantasai> florian: Maybe instead we could just document reasons why a CEO should remove a participant, to help the CEO use this power when appropriate
<fantasai> ... e.g. Board can remove members for cause, and cause is violating established policies -- so need to establish the policies
<fantasai> fantasai: 2 possible changes to Process proposed, and 3rd option to establish better guidelines
<fantasai> ... do we want to take these to the AB, or want to discuss more (or add more options)?
<fantasai> ... (I don't think we should be drafting Process text for any of these, just high-level atm)
<fantasai> ACTION: fantasai and florian to take these 3 options to the AB