w3process
w3process copied to clipboard
w3c
What to do when there's no consensus in the Council
The ability of the chair to call a vote may act as a disincentive to participants to try and reach consensus. However, removing it would mean that failing to reach any decision would be a possibility.
Alternatives suggested have included always using a 2/3 majority, or letting the problem escalate into an AC appeal if too much time passes.
See https://www.w3.org/Consortium/Process/Drafts/director-free/#addressing-fo
This probably relates to #279
It does relate to #279 IMHO, but I think the two can be dealt with separately.
I would suggest that a super-majority of AB members and a super-majority of TAG members is required to sustain a formal objection.
By supermajority I mean something like 2/3 majority of all votes - so if there are 9 people on the AB and two are recused, you still need 6 positive votes to sustain the Formal Objection. Given the relatively small numbers, a simple majority plus one additional vote might be a reasonable alternative.
@chaals your suggestion means that if around half the members are recused then the formal objection can never be sustained, because the supermajority cannot be achieved. Is that what you intended?
This is kinda like the state where the Director can't decide. Well, they keep arguing. If a decision is nonetheless needed (and the lack of a decision is a decision of sorts, at some point), then I think the question to ask is how do we force a termination? Allow them to vote? A majority vote is easy; If it's a super-majority, we have to decide which way it's biased (an FO requires a super-majority to be sustained, or the original decision needs a super-majority to be sustained and otherwise the FO is sustained)?
I agree with @nigelmegitt we have to do the arithmetic after recusals, so a supermajority of those members not recused. We don't expect recusal to be heavily used, mind you.
This also depends on how we see the participants who raise formal objections. Usually people who raise formal objections are minority in the working groups. Would 67% of votes to sustain their objection result in the perception that it is very hard to get formal objections accepted?
It's currently hard to get a formal objection sustained; one has to convince the Director that a WG consensus (as determined by the chair) is somehow wrong. There aren't many examples I recall of the Director intervening to overturn a consensus, and whatever he decides is likely to drive away some proponents of the losing side.
@michaelchampion I think @dbaron has an example of a case where the Director overturned the CSSWG in the distant past. (Presumably this was the correct action: @dbaron is usually right on things.) So it's not common, but it does happen. I don't think this case drove anyone away, either.
Not all the things that can be formally objected to are consensual decisions. For example this is objecting to a previously established consensus:
- Objection by a someone who is not a WG/IG member to a consensual WG/IG decision to publish something
But these are not:
- Objecting to a Team's decision that the CR exit criteria are / are not fulfilled
- Objecting to a Teams's decision that a Member submission should be acknowledged / rejected
- Objecting to a proposed charter for a new Working Group
- Objecting to a Chair's decision to appoint an Editor
- Objecting by a WG member to a proposed decision of the WG
I would suggest that we could require a two thirds supermajority of the Council to sustain an objection against a per-established consensus, but that a simple majority should be enough to sustain a formal objection otherwise.
If this principle seems palatable, we may need to look at details, as maybe not all cases necessarily neatly fall into one or the other category, but on a high level, that seems workable to me.
I don't think we should constrain the Council's working mode much, including when & how it votes. I'd rather the Council be able to figure out for itself how to best get its work done. If the resulting decisions seem bad, maybe it'd be worth revisiting in the future.
@hober Consider the viewpoint of those faced by an FO whose result might undo years of work and/or substantially delay work (have you ever been in that situation?): how would you react if the process was not predictable?
Also consider if you were a member of the FO council. How would you feel if rules changed for each FO? Would you spend the time to get up to speed with the myriad previous decisions and precedent?
The process is no less nor more predictable than today's decision by the Director. Yes, we have been in this situation. The rules are not changing, please don't introduce strawmen. We have requested the team that documentation of all FOs and decisions be built so that the entire community can see precedent and previous wisdom, it's under way.
Overall, we are trying to replace the Director here with a body that represents the membership. We are not trying to eliminate, weaken, or re-define what FOs are and how they work.
The rules are not changing, please don't introduce strawmen.
Maybe this is the disconnect: the rules should probably change because the conditions are changing. Instead of a single individual, who was ultimately responsible to the membership and the community, we now have a group of individuals with each their own allegiances.
Regarding the specific issue raised here, my recommendation is that, as detailed at #631, that, in the event that a voting procedure is specified and when there is no unanimity, that a simple 2/3 majority rule be applied.
Do you disagree with this proposal, and, if so, why?
Yes, I do. The playing field of formal objections is already tilted against the minority player. We should be ready to listen to a lone voice, and not set the bar so high as to deter raising questions. Supermajority requirements are also very sensitive to the phrasing/direction of questions and decisions. A chair might make an unpopular decision if they think that more than a 1/3rd of the Council would not vote to overturn, for example, even if a majority of both their WG and the Council disagree strongly.
@michaelchampion I think @dbaron has an example of a case where the Director overturned the CSSWG in the distant past. (Presumably this was the correct action: @dbaron is usually right on things.) So it's not common, but it does happen. I don't think this case drove anyone away, either.
Since the recent traffic on this issue caused me to notice the older ping to me: I think the issue @fantasai was referring to was issue 3 in CSS 2.1 issues-4 which was about defining the table model differently for HTML and XHTML, also see the w3c-css-wg thread on the directors decision (which was held during an AC meeting welcome reception). I can't find the actual formal objection.
@hober Consider the viewpoint of those faced by an FO whose result might undo years of work and/or substantially delay work (have you ever been in that situation?): how would you react if the process was not predictable?
In most cases FO are quite predictable (except in some rare cases where an AC rep would use FOs quite often for rather small issues, because somewhat misunderstood the meaning of FO but this always got resolved easily by a simple talk). WG tend to know in advance where strong disagreements lay and they can't be truly surprised by a FO if they followed the review process all the way. I'm not saying it's their fault if they ultimately get the FO in their face at the end, I'm saying that controversial matters are visible from afar. Then the fact it is going to take time to get to a final decision, be it by the director or by a council, can't be helped.
Also consider if you were a member of the FO council. How would you feel if rules changed for each FO? Would you spend the time to get up to speed with the myriad previous decisions and precedent?
I don't think that a member of an FO council necessarily has to know how previous councils resolved previous FOs. Arguments and situations can be very different. If WGs refer to precedents as argument points, they would certainly need to document those but constraining the FO council discussions too much seems rather undesirable (and impossible, in fact).
WG tend to know in advance where strong disagreements lay and they can't be truly surprised by a FO if they followed the review process all the way.
Sometimes, unfortunately, people do realize things late in the process. But indeed, if someone raises an FO 'late' I expect eyebrows to be raised, and it may weaken the case: "why didn't you say all this in the WG?" "if you object to a spec. on this subject, why didn't you object to the Charter, and instead wait to PR?". Now, it may be that something new has come to light, or there is some new realization, of course, so these are not diagnostic, but they are questions to ask.
A chair might make an unpopular decision if they think that more than a 1/3rd of the Council would not vote to overturn, for example, even if a majority of both their WG and the Council disagree strongly.
Are you saying that, under the current W3C process, a Chair can make a decision without the consensus of a group and without violating the W3C Process? It is the second time I hear this today and find it extremely surprising.
Supermajority requirements are also very sensitive to the phrasing/direction of questions and decisions.
Making decisions by super-majority tends to generate clear positions -- esp. on a topic of importance.
Asking that the majority (super or simple) support the objection avoids the situation where an objection is sustained due to the apathy of Council member -- based on the assumption that the objection is overruling the opinion of a consensus body.
The playing field of formal objections is already tilted against the minority player.
I would hardly describe the current playing field as tilted against the objector: as it stands, an objector can unilaterally, with minimal justifications and with minimal consequences, (a) slow down a project and (b) consume a large number of community resources.
The current proposal is, the Council decides
- By unanimity if possible
- By consensus otherwise
- By a simple majority vote if all else fails
I believe this is the correct outcome, and that a supermajority vote in particular would be a poor mechanism to adopt for the reasons mentioned by @frivoal in https://github.com/w3c/w3process/issues/280#issuecomment-688604751 and by @avneeshsingh and @dwsinger in https://github.com/w3c/w3process/issues/280#issuecomment-502343564 and https://github.com/w3c/w3process/issues/280#issuecomment-1254367035
Regardless I think this discussion needs to be escalated to the AB.
@fantasai The proposal at https://github.com/w3c/w3process/issues/631 had two aspects: (1) what kind of voting method: super-majority (2) what would the motion be "should the FO be sustained"
The proposal at https://github.com/w3c/w3process/issues/280#issuecomment-1255664340 tackles only (1). What about (2)?
I still agree that a 2/3 majority requirement is reasonable. A Formal Objection is asking to overturn something intended to approximate consensus, and the playing field should be heavily tilted against them.
If a chair has resolved a decision against the wishes of a majority of a WG, that's a pretty obvious violation of the decision policy. A council that does not understand and support an objection to that had better have some pretty clear reasoning if they want to be trusted. (There are such cases, e.g. the group has alienated everyone who was not "toeing the line", but if they reach the council we have issues with the Team Contact and chair, CEPC, and the clear reasoning seems feasible to establish).
@nigelmegitt asked
@chaals your suggestion means that if around half the members are recused then the formal objection can never be sustained, because the supermajority cannot be achieved. Is that what you intended?
If the council reaches consensus, it isn't an issue. If they are too small to be viable, the current draft says they should be dissolved and a new council formed. I would consider a council that can only resolve with a vote, and can't get enough votes to make a decision either way, is too small to be viable.
Reminder: not all decisions are required to be backed by consensus. Indeed, one possible ground for an FO is that there should have been consensus and it was not, in fact established.
Any vote needs to be of those not dismissed or renouncing, i.e. of those qualified to serve, and I suspect that it should simply compare yes vs. no and not consider abstains.
A situation we were keen to avoid was requiring super-majority to make any decision, as that could leave us decision-free indefinitely. I agree, that "in the case of a vote, the FO is sustained by a super-majority, otherwise it is overruled" avoids that problem.
If the council reaches consensus, it isn't an issue. If they are too small to be viable, the current draft says they should be dissolved and a new council formed. I would consider a council that can only resolve with a vote, and can't get enough votes to make a decision either way, is too small to be viable.
This seems like a recipe for endless forming and re-forming without ever being able to resolve.
They can always get a vote; the requirement is based on those not dismissed or renouncing, and excludes abstains. Whether they'd like to issue a decision if many also abstain is another question. I prefer, as you know, a "majority of those voting yes/no" and maybe can live with "supermajority of those voting yes/no to sustain the objection, otherwise it is overruled".
An obvious problem with the super-majority is that we could be asking the Council to write supporting material for a decision that a majority (but not super-majority) disagree with. In that case, the "minority opinion" is actually the majority opinion.
I am opposed to any rule that could leave the Council without any decision. That includes, for example "A majority of those serving is required to make a decision to sustain or overrule" as then if a majority abstains, no decision is made. A super-majority to make either decision also leaves area in between the two super-majorities. Lack of a decision can be the deadliest thing.
I'm a bit torn: in general I believe supermajority voting is a way to break impasses when there is dissent but an issue SHOULD be decided by some semblance of consensus. If we were talking about a situation where there was't a concept of a Formal Objection and the AC voted on whether to approve a Charter or Proposed Recommendation, a 2/3 supermajority requirement seems a workable substitute for "consensus."
But assuming FOs and Councils, if a simple majority of a group of experts who have spent time investigating decide a FO has merit, the FO has enough credibility to request that the proponents address the issues before getting W3C's imprimatur.
But assuming FOs and Councils, if a simple majority of a group of experts who have spent time investigating decide a FO has merit,
@michaelchampion By "group of experts" do you mean the FO Council? If so, in the case where the FO is technical in nature, have you considered the possibility that the group of experts might not in fact be experts in the technical topic and the domain it is intended to address, e.g. they spent orders of magnitude less time studying the topic than the group whose subject to the FO. I do not the FO Council can be expected to become technical experts in the time they are provided.
Yes I mean "experts" == "FO Council"
Here's a scenario:
- A WG that does NOT REALLY have sufficient technical expertise gets agreement on a spec. Actual experts did not join the WG, perhaps because of the patent commitments, perhaps they suspected the charter was flawed but didn't know/care enough to formally object to the charter. So there is no dissent in the WG to advancing to Proposed Recommendation.
- in the AC ballot, some number of actual technical experts (or people concerned about the social impact of standardizing the technology, whatever) formally object.
- A FO council is formed from AB and TAG members (elected on the basis of their expertise/experience/understanding of web technologies)
- A simple majority of the Council concludes that the concerns raised in the FO are warranted
I am very comfortable with the FO being sustained in that scenario.
I do not the FO Council can be expected to become technical experts in the time they are provided.
TimBL was not a technical expert in every issue for which he adjudicated a Formal Objection. But he did have a broad and deep general understanding of web technology and invested sufficient time to make reasonable, informed analyses of FOs. That's what we can ask of an FO Council: they use their knowledge and information to assess an FO, and if there is sufficient evidence it is warranted, request the proposers of the spec or charter to go back and address the concerns.
@michaelchampion
I am very comfortable with the FO being sustained in that scenario.
Here's the same scenario as above recast in a different light:
- members A form a WG
- members B choose to not join the WG because (i) they are not paying attention, (ii) the WG does not fit the product plans of their respective companies, (iii) their boss does not believe the standard will succeed, (iv) their companies do not wish to commit patents, etc.
- the WG advances to Proposed Recommendation with a specification that meets all Process criteria, including editorial quality, multiple implementations, etc.
- members B wake up and formally object to the specification on a technical basis, e.g. a technical choice made in the specification would require modifications in X millions of their devices
- the FO Council decides by simple majority to sustain the objection, undoing many years of work from members A and potentially jeopardizing the adoption of the specification
In that scenario, the bar to sustaining the FO should be extremely high.
It should be difficult for a minority group of members to ignore work and then object to it at the last minute on the basis of input they could have provided earlier.
The incentive should be for members to provide input/raise concerns from the beginning, and not to sit on the sidelines and delay their input until the bitter end (this makes for a haphazard standards process, at best).
Recall that (a) W3C standard are voluntary and (b) members B can always develop their own specification.
that's what we can ask of an FO Council: they use their knowledge and information to assess an FO,
There is a fundamental difference between the FO Council and TimBL:
- TimBL was a single individual who was completely invested in the W3C in terms of reputation, network, knowledge, etc.
- the council is made of individuals that come and go (sometimes at the whim of their employers), some are employees of corporation with interests that go beyond the specific FO, some do not represent members (invited experts), some only have a small amount of time dedicated to the council, etc.
As a result, decisions will be made differently by a council than they would have been made by TimBL.
The bar for the FO Council to sustain an FO should be high and the FO Council should not be encouraged to revisit WG technical consensus but in the most egregious cases.
P.S.: Thanks for engaging in the discussion. Happy to continue here and/or privately, whatever you prefer.