RE: Decision process in the HTML working group

(fwds reduced to >> and > for simplicity)

Maciej Stachowiak [mailto:mjs@apple.com] wrote:
>Dan Connolly wrote:
>> This business where the FO is a follow-on to a decision seems
>> broken, to me. The point at which to object, formally, is when
>> the question is put, not after the decision is made.
>
> I've asked around, and that doesn't seem to be the way other W3C
> Working Groups do it.

I agree, in my experience it's run differently in different groups.

>I've heard from representatives from the Web
> API, WAF, SVG, CSS, CDF, Web Security Context, Mobile Web Best
> Practices and Device Description WGs, in all cases they decide by
> simple majority after sufficient discussion, and Formal Objections
> have to be registered separately. I encourage you to ask other
> chairs about this.
>
>> I'm well aware that other chairs in other groups do it differently.
>> I still think it's wierd/broken.

The most effective system I have experience with is Lauren Wood when she ran the DOM WG.  There are three states of response, to me, that are interesting in making a decision:

1) I like this solution
2) I don't like this solution, but I can live with it (you can lump abstains in this category)
3) I can't live with this solution

(there is a four state that I think is largely untinteresting - 4) abstain.)

Lauren would use this tool frequently, and do a straw poll of the WG to see how people felt.  It requires absolute honesty, of course - #3 is essentially forewarning: "if the WG chooses this solution, I will have to vote no on the specification and make a formal objection to the director."  If there were a few #1s, lots of #2s, and no #3s, it was still a good decision.  Even a single #3 could sink a proposal, depending on their actual objection.  I don't believe simple majority is a sane way to make decisions, if only because I know that some objections are hard stops for implementers, and in those cases those objects are blocking.  I will use the Ogg Theora example - I believe that Apple and Microsoft would have to object to this being mandated in the specification.  Everyone else in the WG might vote for it.  Given the strength of the objection, I would have to say that it would be a dire mistake to require it in the specification, because two of the major browser vendors would have to be incompatible with the specification.  That seems stupid.

> From experience, it seems much more productive than the way we do it
>now, in that you don't end up with every vote putting the project it
>risk. I'm not sure why you think it's broken. The point of Formal
>Objections, as I understand it, is to appeal a decision you strongly
>disagree with, not to prevent a decision from being made.

Every vote - heck, every technical point in the spec - does put the project at risk, if it causes one or more of the major implementers to back out of the WG (or of implementing the spec).  I agree with what I think is the point you're making, though - we shouldn't require Formal Objections to the director to state #3, "I can't live with this solution".

>Maybe we could stick to the Process until after you go through the
>process of getting your changes adopted.

Play nice.  :)

>> I have no use for "I disagree with the majority" data.

I do, FWIW.

>It seems that these are the most extreme positions possible on any
>issue. And trying to push people towards the most extreme positions
>is likely to polarize the group more, rather than leading to
>compromise. It's hard to become tolerant of differences when every
>disagreement puts the project at risk

I agree.

>I know that in past cases of WG decisions I've disagreed with,
>whether as a commentor or group member, I've been happy to express my
>disagreement, have it recorded, and move on with no further appeal,
>but I would *not* have been happy to abstain or have my vote look
>like an abstention.

Indeed.  I would like to acknowledge and note objections that are #2.  I also think it's critical to acknowledge the difference between #2 and #3.

-Chris

Received on Thursday, 10 May 2007 18:31:12 UTC