W3C home > Mailing lists > Public > public-w3process@w3.org > October 2017

Re: Working group voting procedures in Process 2018

From: Chaals McCathie Nevile <chaals@yandex.ru>
Date: Tue, 24 Oct 2017 23:33:56 +0200
To: "Jeff Jaffe" <jeff@w3.org>, "L. David Baron" <dbaron@dbaron.org>, "Michael Champion" <Michael.Champion@microsoft.com>
Cc: "W3C AC-Forum" <w3c-ac-forum@w3.org>, "chairs@w3.org" <chairs@w3.org>, "ab@w3.org" <ab@w3.org>, "public-w3process@w3.org" <public-w3process@w3.org>
Message-ID: <op.y8mt6uqdftbnq3@desktop-kurf4r9>
On Mon, 23 Oct 2017 22:55:17 +0200, Michael Champion  
<Michael.Champion@microsoft.com> wrote:

> I also  have reservations about the changes to 3.4.  I'm embarrassed to  
> see that I seem to have proposed the "one person one vote" language in  
> https://github.com/w3c/w3process/issues/24 because I have no  
> recollection of what I was thinking, and David Baron is right about the  
> risk of having organizations appoint members to a WG solely to weight  
> the vote tallies.

I think that point is being stretched.

>  Chaals has a point that it seems likely that any obvious abuse of this  
> voting rule would trigger a formal objection, but I disagree with

>> I'm not terribly concerned at what the default looks like because I  
>> think it represents an attempt to save discussion on an issue rather  
>> than a soundly justifiable basis for claiming the answer is *right*.

> There's a lot of recent research/policy indicating that defaults are  
> taken very seriously and it's important to set them appropriately, e.g.  
> https://en.wikipedia.org/wiki/Nudge_theory,  
> https://en.wikipedia.org/wiki/Default_effect_(psychology)  .  If we are  
> going to change the defaults for voting, it's critical that there is  
> broad consensus on what the default policies should be.

OK, I may have been a little unclear here. I agree that the default should  
be sound.

I happen to think that the "one person, one vote" approach is sounder than  
what we had, given that we are less defensive than we were 20 years ago  
about whether organisations were trying to stack groups to get outcomes  
that favoured thme over others. Which is not to ignore that as an issue,  
just clarify that the assumption these days is that individuals are  
generally expected to present their best understanding rather than a  
corporate policy designed to achieve a competitive edge.

> I stand by my comment on that issue that I'd prefer not to do anything  
> that undermines the definition of consensus as "lack of sustained  
> opposition",

I don't see that we undermine consensus on that basis in the process, and  
I would like to *strengthen* the point that any voting represents a  
certain level of failure of the group to get there.

> but if necessary I could live with supermajority voting rules to resolve  
> deadlocks.   The Process CG seems to have concluded that discussing  
> supermajority or quorum thresholds would  take too long to fit it into  
> Process.

As well as the concern that it takes a long time to get people to agree, I  
think it is a bad choice of default, since IMHO it serves to further  
legitimise voting to resolve issues. Combined with that concern is that in  
my experience questions are often framed for votes in such a way as to  
prejudice the outcome, and waving some air of supermajority authority over  
that is "putting lipstick on a pig" - offering far more respectability  
than I think is a good idea.

In my understanding, voting is generally a way to move past an issue with  
a solution that is probably workable, and unlikely to be great. This is  
why a formal dissent is carried alongside the work until it can be  
examined at a higher level than the Working Group.

>  That being as it may, I prefer the Process 2017 language to the current  
> draft language about default voting rules.

I don't. I think in a world where participation is far more fluid, a  
default of one vote per participant and most votes wins provides a  
reasonably straightforward way to move on past a deadlock, without giving  
undue support to a claim of "we already fixed this". In bad cases - and I  
am reminded of Dave Singer's reminder that they are the reason to have a  
proces, as well as the fact that votes only occur in bad cases - there is  
already a reasonable likelihood of disillusioning people so that instead  
of sustaining their formal objection and getting it carefully addressed  
later, they are harassed out of the group.

Sadly, I have seen that happen enough to want to protect against it more  
than to protect the sanctity of some quick-and-dirty resolution of a  
problem.

cheers

Chaals

> -----Original Message-----
> From: Chaals McCathie Nevile [mailto:chaals@yandex.ru]
> Sent: Monday, October 23, 2017 12:53 PM
> To: Jeff Jaffe <jeff@w3.org>; L. David Baron <dbaron@dbaron.org>
> Cc: W3C AC-Forum <w3c-ac-forum@w3.org>; chairs@w3.org; ab@w3.org;  
> public-w3process@w3.org
> Subject: Re: Working group voting procedures in Process 2018
>
> On Mon, 23 Oct 2017 18:49:14 +0200, L. David Baron <dbaron@dbaron.org>
> wrote:
>
>> I'm curious about the rationale behind one of the changes within #24,
>> which covers voting *in working groups* (which is described in both
>> the new and old process as a rare procedure that should only be used
>> when consensus cannot be reached).
>>
>> In the current process, votes in a working group MUST be taken
>> per-organization (or group of related members).  In the revised
>> process, the default voting process (which can be overridden by
>> charters) is that votes in a working group default to one vote per
>> participant.
>>
>> This change seems to introduce the risk that, if a working group is
>> facing issues contentious enough to lead to a vote, it allows
>> organizations to add new members to the group in order to change the
>> results.  This seems undesirable to me.
>
>  From my perspective it is true that some organisation might try to fill  
> the group to win a vote. In the unlikely event that an important issue  
> really got determined this way and left people unhappy at the outcome, I  
> would expect a formal objection. I expect part of the director's  
> analysis of such an objection to include looking at any such attempt at  
> "distorting the outcome" with about as much contempt as the particular  
> case merits.
>
> Voting is a suboptimal approach for most important decisions anyway. It  
> is potentially useful to stop a bikeshed discussion (not because it gets  
> a good answer, but because there isn't one apparent and it stops the  
> time being sucked into different ways to make a bad decision...).
>
> An alternative perspective is the old HTML Working Group, which had far  
> more invited experts - each given one vote - than organisational members  
> who were thus a small minority in any official vote. While I hope that  
> was an historic anomaly, in a group where one large organisation has 4  
> times as many people as anyone else doing 75% of the work, while I  
> suspect there will be other problems it seems reasonable to let them  
> have more than 1 vote, in the broken case that this is the only way  
> forward on an issue.
>
> So yes, there is a power shift in the "default" model. Between Arrow's  
> theorem, a sense that very many questions are badly put to vote in my  
> experience, and the sense that this is already a case that should have  
> been avoided, I'm not terribly concerned at what the default looks like  
> because I think it represents an attempt to save discussion on an issue  
> rather than a soundly justifiable basis for claiming the answer is  
> *right*.
>
> cheers
>
> Chaals
>
>> (I'm coming to this from the perspective of a member of the CSS
>> working group, which officially has 19 participants from Google, 11
>> from Apple, 11 from Microsoft, 8 from Mozilla, 6 from Vivliostyle, 5
>> from Adobe, 5 from BPS, etc., but has also never held a vote.  But I'm
>> under the impression that there have been a small number of working
>> groups where voting was used a good bit.)
>>
>> -David
>>
>> On Wednesday 2017-09-27 20:36 -0400, Jeff Jaffe wrote:
>>> Dear AC representative, WG Chair, or member of the public,
>>>
>>> The W3C Advisory Board is forwarding a proposed Process 2018 draft
>>> [1] to the Advisory Committee for consideration and comment. The plan
>>> is that, based on the received comments, a revised draft will be sent
>>> to the Advisory Committee for formal Review prior to the November
>>> TPAC meeting and that there will be time for questions and comments
>>> on the proposed Review document at the TPAC meeting.
>>>
>>> [1]https://w3c.github.io/w3process/
>>>
>>> The major changes in this document and their rationale, with links to
>>> the current process and a diff from it, are provided in a
>>> backgrounder [2].
>>> [2]https://www.w3.org/wiki/Process2018
>>>
>>> We call special attention to issue #5 - designed to increase agility
>>> for errata management moving us closer to a living standard model and
>>> issue #52 which updates participation and election rules for the TAG.
>>>
>>> Please send comments as soon as possible (to facilitate response
>>> preparation) and prior to October 26th (a 4 week comment period).
>>> Specific comments on the text are best filed as Github issues or even
>>> pull requests at the Process CG github
>>> site<https://github.com/w3c/w3process/issues>.
>>>
>>> More general discussion and comments should be sent
>>> topublic-w3process@w3.org  (Mailing list archive, publicly available)
>>> or toprocess-issues@w3.org  (Member-only archive).  You may discuss
>>> your comments on any other list, such asw3c-ac-forum@w3.org, as long
>>> as you send the comments to one of the W3process lists above and copy
>>> that list in the discussion.
>>>
>>> Jeff Jaffe, Chair, W3C Advisory Board Charles McCathie Nevile,
>>> Editor, W3C Process Document David Singer, Chair, W3C Process
>>> Document Task Force
>>>
>>
>
>
> --
> Chaals is Charles McCathie Nevile
> find more at http://yandex.com
>


-- 
Chaals is Charles McCathie Nevile
find more at http://yandex.com
Received on Tuesday, 24 October 2017 21:34:29 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:51:45 UTC