Recording council disagreements

[frivoal]

Forking off https://github.com/w3c/w3process/issues/278#issuecomment-986043959, from a comment @msporny made:

Last I heard, the final tallies for/against would be made available, but not how individuals on the FO Council voted. I understand the reasoning to hide the votes -- so no one gets "bullied" into voting a certain way. That said, I don't think that concern is outweighed by the transparency of the process.

From this part of the draft process:

Decision of the W3C Council should be unanimous. If despite careful deliberation the W3C Council is unable to reach consensus, the Chair of the W3C Council may instead resort to voting. In that case, the decision is made by simple majority, with the Chair of the W3C Council breaking ties. […] A rationale supporting the decision must be provided as well. In the case of non-unanimous decisions, members of the W3C Council who disagree with the decision may write a Minority Opinion explaining the reason for their disagreement. Minority Opinions must be included in the announcement of the decision.

As long as the decision is unanimous, you don't need the names of people who voted, since they all support the decision. In case of a majority-only decision, the minority opinion gives you insights into what the the disagreement was. It may be worth stating explicitly that the authors of the minority opinion should/must identify themselves. It's also interesting to wonder: If people didn't support the decision, are they required to write a minority opinion, or can we just have a non-unanimous decisions, but with no further insights on who disagreed and why?

[msporny]

As long as the decision is unanimous, you don't need the names of people who voted, since they all support the decision. In case of a majority-only decision, the minority opinion gives you insights into what the the disagreement was. It may be worth stating explicitly that the authors of the minority opinion should/must identify themselves. It's also interesting to wonder: If people didn't support the decision, are they required to write a minority opinion, or can we just have a non-unanimous decisions, but with no further insights on who disagreed and why?

Historically, IIRC, FO decisions have often been provided with an explanation of the logic that led up to the decision. I've found these write-ups insightful and compelling. I've often changed my position (e.g., EME) after reading the FO decision prose. I imagine we'll keep this mechanism going forward.

The thing that changes in this new director-free approach is that it is possible for there to be dissent on the decision, and it seems as if that dissent needs to be documented as well. "We came to a decision through a simple majority vote, here's the rationale on that... and here is the rationale from those that dissented." Documenting dissent is an important exercise in any governing body... it helps the organization remember why some decisions were made, and the documented problems with that particular decision.

I will also note that simple majority is a terrible outcome... For example a 49%-51% vote signals almost perfect disagreement -- it's the worst sort of divisiveness in a body -- maximized disagreement (in other words, you literally couldn't do worse than the outcome you came to!) :). If an FO comes to that -- it signals a larger problem. We might want to entertain the possibility of "failure to resolve an FO" reconvening a different group if we end up in a situation such as that, or that the FO finding is a set of recommendations that would make the decision less controversial... or backing off to support for REC? What I'm trying to say is that we might want to have a better exit than to use the result of a simple majority vote.

Perhaps we should consider a 75%+ supermajority vote in the case where there is a failure to come to consensus? ... and if that fails, there is a proposal from the FO Council on what could reach a supermajority.

[dwsinger]

@msporny all good thoughts. but it's important to realize that in the case of FOs, there is no "no decision" – a failure of the council to decide leaves the status quo standing. The chair needs to do all they can to find consensus and indeed should see a nearly-balanced vote as a poor decision, and revisit and re-craft as much as they can to get a better result: but we decided we can't afford a process that might lead to a lack of decision.

I completely agree that we need rationale and, if they exist, dissenting opinions.

[msporny]

a failure of the council to decide leaves the status quo standing.

Good. I thought we were exploring options where that wasn't the case. If it is the case that a failure for the FO Council to decide leaves the status quo standing, then many of my concerns evaporate. I think the only one that remains is that I'd like us to consider a >75% supermajority if it comes down to voting instead of a simple majority (>50%).

To apply this to a concrete case, for the DID Core formal objections, the question was whether or not to publish DID Core as a REC. If the FO Council were to meet, and fail to reach consensus, and then vote and fail to reach majority (for whatever definition of majority we end up on), then the outcome would be to publish DID Core as a REC? Regardless of the outcome, a majority/minority opinion would be available to the membership and general public? Am I understanding that correctly?

[dwsinger]

@msporny whether a bias is a good idea depends on the direction of the original decision. so, for example, if the chair of a WG had decided that the WG did not have consensus to publish, and you wanted to object to that decision, the bias would have been the other way. In other words, a bias towards leaving the original decision standing is not always the right way. The council needs to make a decision, just as the Director does. We do not want automated processes making decisions on our behalf.

[frivoal]

Regardless of the outcome, a majority/minority opinion would be available to the membership and general public?

Yes. I base this on:

A record of each Formal Objection must be publicly available.

Decisions of the W3C Council must have the same level of confidenntiality as the object of the Formal Objection.

pushing a REC or adopting a charter are public things, so council decisions must be public as well. Including minority opinions, if any. Source for that is:

Minority Opinions must be included in the announcement of the decision.

Process also says:

A record of each Formal Objection must be publicly available.

So maybe we can even simplify this must have the same level of confidentiality, and simply go with must be public.

I will also note that simple majority is a terrible outcome... For example a 49%-51% vote signals almost perfect disagreement […] Perhaps we should consider a 75%+ supermajority vote

Agree that 51/49 is close to a worst case scenario. But we do have:

Decision of the W3C Council should be unanimous.

So the council is not meant to be happy that it reached 51/49, and the chairs are expect to try and get the group to unanimity if possible, or near unanimity if not. I think that's deeply ingrained in the w3c, so I am not too worried that we'll go for 51/49 when 99/1 would be achievable.

But it remains an interesting question about what to do in the cases where we cannot do better than simple majority. So far, the idea is that decision by simple majority, with an explained rationale, and a documented minority opinion, and the possibility of an AC appeal, is better than no decision. But it is a hard question, in particular because it's not typical, so it's hard to reason about given that we don't have a lot of experience of that sort of situation to go by.

[frivoal]

Most of this is covered in the https://www.w3.org/Consortium/Process/Drafts/director-free/ branch, but the AB needs to decide on whether:

• vote totals must/may/must not be reported
• individual positions must/may/must not be reported by name

[plehegar]

"resolved: Individual positions MUST NOT be attributed in the Council Report" "resolved: "May" "

• AB - October 20, 2022

[frivoal]

"resolved: "May"

This lacks the context of the full AB minutes to be understandable. That "may" is about the council reporting vote totals (when there is a vote rather than a decision by consensus).

[css-meeting-bot]

The Revising W3C Process CG just discussed The council may report vote totals, and agreed to the following:

• RESOLVED: Merge PR 665 with s/point/position/
• RESOLVED: Close issue 605

The full IRC log of that discussion

<fantasai> subtopic: The council may report vote totals
<fantasai> github: https://github.com/w3c/w3process/issues/605
<fantasai> plh: Allowing the Council to report votes
<fantasai> plh: if there is a vote, AB decided not to impose whether to report or not the vote totals
<jeff> q+
<fantasai> fantasai: Individual positions cannot be reported, but the vote totals can be
<fantasai> florian: Minority report still exists (just can't be signed)
<plh> ack jeff
<fantasai> jeff: Pull request strikes out "must not report individual votes"
<fantasai> ... maybe forbidden by the next line?
<fantasai> florian: That's how I read it
<fantasai> ... I think it was redundant previously
<fantasai> ... I can also keep the "must not report individual votes"
<fantasai> jeff: agree implied, but whole thing is confused
<fantasai> ... but what about minority report?
<fantasai> florian: It states an opinion, but not who states it
<fantasai> jeff: I think you're correct it's handled by the next line...
<fantasai> ... maybe [wording tweak]
<fantasai> fantasai: My suggestion is s/point/position/
<fantasai> florian: suggestion taking the PR with that change
<TallTed> +s position
<jeff> +1 Elika
<fantasai> RESOLVED: Merge PR 665 with s/point/position/
<TallTed> s/+s/+1/
<fantasai> RESOLVED: Close issue 605