Chairs to actively ensure both sides of an argument are expressed

[jwrosewell]

Chairs must make equal time to hear both sides of an argument and work with participants to encourage consensus via such a route.

Given the lack of representation of smaller organizations that can afford the time to participate in the wide range of W3C groups and other trade organizations, it is incumbent upon the W3C to ensure that lack of counter arguments is a testament to widespread consensus, rather than an indication of wider participation.

This is important to further ensure that the W3C is not used as a channel for anticompetitive conduct or collusion.

[chaals]

I believe that is obvious in the statement "groups should seek to make decisions by consensus".

What concrete change to you propose?

"Equal time for both sides of an argument" is a far too restrictive measure. Some reasons for that opinion:

• it presupposes there are precisely 2 positions, whereas in practice there are often a range of possibilities, and open discussion leads to convergence.
• it fails to take into account that different proponents are not in an equal position to express arguments, for example in language ability, initial knowledge of a proposal, and ability to consolidate opinions of multiple stakeholders
• it works very badly for asynchronous discussion since it presupposes limiting the amount of input someone can provide to match that of others, and we don't have a functional mechanism to achieve that even if it made sense.

[dwsinger]

We require that groups listen to all considerations; and if that doesn't happen, we have robust mechanisms to "stop the train" (notably formal objections). I don't think "both" is a useful word here; many technical design questions have a wealth of factors to bring into consideration and balance, and indeed we expect working groups to consider, and balance. Ideally, that process results in satisfying all the concerns; sometimes, we have to find a balance, and that means that some will feel that the balance is off. In cases of serious dissent, we have the processes of formal objection, AC discussion, and appeal, and we have had cases where important questions are scheduled at an AC or even all-member (TPAC) meeting.

I am having a hard time envisaging what we could improve on here; generally, the complaint I hear about the W3C is that it takes time (and consequent delay) to hear every point, and debate it, and thus moves too slowly.

[jwrosewell]

Comments have been made about the use of both sides of an argument. I withdraw that and accept it should be “all" sides, or all perspectives or all considerations from every angle. This should be uncontroversial and suspect we are not far apart. It appears to accord with W3C values.

However, my suggestion is that Chairs formally provide time for contributions from the smaller, quieter or less extrovert members of groups to ensure their contribution is understood. Where stakeholder groups are absent they would need to seek them out to ensure representation. It is not the same thing as groups seeking to “make decisions by consensus”. Consensus means agreement. This point is about how that consensus is discovered and achieved. Where there is no agreement the silence of the assembled group could for example be taken for assent to a proposal. I am sure that is not the right way for W3C. Positive agreement should be required otherwise agreement should not be assumed.

Don’t you agree that silence in the face of a proposal being taken as “agreement” or “consensus” is a bit too totalitarian? Don't we agree that all sides should be solicited and heard from before a decision is taken?

[mnot]

There is a tricky balancing exercise to engage in here. On the one hand, you want to assure that the consensus is genuine and takes all views into account. On the other, making frivolous and obstructive contributions -- with malice or not -- puts the success of the standards effort at risk. That's largely because standards are an exercise in design-by-committee, with all the negative connotations that has.

Barring someone from contributing (in e-mail or as a draft proposal) is something that should happen very rarely, and only because they are judged to be disruptive. I believe the W3C has a process to guide that.

How meeting time is allocated is often more contentious. A good chair will use the time well, to assure that different viewpoints are heard, while trying to avoid fostering group-think by only allowing a subset of contributors to present. This is why the IETF prefers discussion time over presentation time: if anyone can get into the queue to make a point, that helps to ensure fairness (and a good chair will make sure that someone who reinserts themselves into queue too much will get deprioritised).

Of course, it's widely noted that speaking in front of a room of technical experts is hardly attractive to 'the smaller, quieter or less extrovert members of groups.' That's one reason e-mail and draft contributions are allowed, to accommodate other means of participation. If their arguments are convincing, they'll gain confidence to speak about it, and/or find advocates who can speak on their behalf.

Silence being considered assent is established as common practice in many (most?) SDOs, provided that appropriate notice is given and any objections raised are dealt with. In some circumstances, more positive support is desirable (e.g., starting new work), but that doesn't extend to asking each participant individually. The reality is that when 100 people are working on something, only perhaps 20 of them are actively paying attention, and even fewer are proactively participating. Any of those 100 people can object, but if you hold up progress until you get the explicit assent of each one, you'll get nothing done.

Regarding your proposal specifically: allowing anyone who has proposal to consume meeting time would create perverse incentives for those who wish to stall or stop an effort or for those who do not bring helpful proposals to the table. The resulting delays and distractions in turn create a disincentive for implementers and other practitioners to participate in the standards process.

It's the chair's job to allocate the group's time efficiently, so that it's able to ship a product that is interoperable, secure, represents consensus, and gets real-world implementation. In balancing these goals, they need to perform some qualitative filtering, rather than blindly giving all comers equal time. If they get that balance wrong, there should be robust appeal processes to correct any issues.

[dwsinger]

There are several things we ask chairs to consider and notice, and we ask participants to assist in noticing.

One is that indeed, some people are more shy, or culturally attuned not to speak unless asked. Indeed, it helps if chairs ask them for input in real-time (face to face or telecon) meetings; "Naoko, you've indicated familiarity with this — what do you think?"

This also extends to liaisons or requests for input; it's quite common for someone to speak up and say "that would have styling implications, we should alert CSS" or "ATSC has adopted this spec., and this security fix has backward-compatibility issues; we should make sure that we alert them to the question" or "I'm concerned we have no-one from the ePub community here, and this change might affect offline use". Again, we hope that the chair is tuned to these, but realistically don't expect the chair to be all-seeing or omniscient, and encourage anyone to speak up.

On the question of consent; generally, the pattern over the last few years has been that when there is a formal decision to be made (e.g. "is this good enough to be our First Public Working Draft?"), there is both a discussion in a real-time meeting and then the chair sends out a call for consensus. And here again, I think it's fine for anyone in the group to point out that a decision might be 'significant' for some reason, and ask for an explicit call for consensus. However, though we can make such questions explicitly visible we cannot, of course, force people to answer; and indeed many questions go unaddressed by many members, and we have to assume that abstention (and consequent silence) means at worst a lack of dissent, if not actual assent. (And often CfCs are expressed explicitly saying that silence will be taken as assent, but we'd like at least some explicit indications of support).

[dwsinger]

Transfer to the team to improve chair training, to make sure chairs understand the need, and the mechanisms, to ensure that all voices are heard and all points raised and considered. Chairs can (and sometimes do) easily lapse into hearing only what they expect or want to hear, for example. Perhaps less "equal time" than "all viewpoints and considerations are explored and understood", as we can get distracted by the side-line of "measuring time".