Outcomes for the Working Group

Colleagues,  

Since suggesting a contingency wind-up process for the group, I've received a deluge of off-list messages.  While I'd like to be able to respond individually, post-graduation commitments are commanding my schedule this week.  In the interest of providing timely detail to my thinking, as well as returning to inbox sanity, here's an exposition on the most common query: What might happen next for the working group?

In my view, there are roughly five possibilities.


1) We publish a consensus Last Call text by July.  (Within the longstanding meaning of consensus; more on that below.)  This is by far my most preferred outcome, and as I noted, I plan to work for it right up until our deadline.


2) We wind up the working group if we fail to arrive at a consensus Last Call text in July.  My previous note discussed this sort of contingency plan.


3) We push back our July deadline.  We have, however, already delayed Last Call about four times (totaling a year and a half).


4) We publish a vague Last Call text that does not reflect actual agreement on the major issues.  For example, we might rely extensively on imprecise terminology and nix clarifying non-normative examples.  The practical effect would be to merely postpone our disagreements, compounded by subsequent confusion, frustration, and loss of context.  (Some participants have referred to these phenomena as "false consensus" or "fake consensus" in prior meetings.)


5) We publish a Last Call text by redefining our rules of decision, either relaxing the threshold for consensus or making extensive use of the chair reconciliation procedure.

Since its beginning this group has operated with a consistent understanding of consensus: general agreement among the various stakeholder perspectives.  Two previous decisions are illustrative.  First, we decided on our 6/6/12 call that Internet Explorer 10 Release Candidate would be noncompliant—despite opposition from both Microsoft and me.  Being on the wrong end of consensus sure is unpleasant.  But neither of us challenged the decision-making procedure.  Consensus does not require unanimity.  Second, we decided at our 4/11/12 meeting in Washington to continue with the Compliance document.  A substantial majority of the room appeared to favor dropping the Compliance text.  But we persisted, since some stakeholders—in particular, the privacy advocates—were opposed to the move.  That decision also generated little procedural controversy.  Consensus is not just a majority vote.

As for the chair reconciliation procedure, it's been used only once in the group's history: deciding whether a tri-state user interface would be required.  On that single narrow issue, the chairs devoted substantial group and staff time to hosting focused discussions, carefully framing a poll, soliciting extensive written input, and authoring a detailed opinion.  Nevertheless, some participants filed a formal objection afterwards, and stakeholders from various perspectives expressed significant concerns about the process.  The chair reconciliation procedure hasn't been used since.

The W3C rules give our co-chairs de jure discretion in both setting a consensus threshold and resorting to the reconciliation procedure.  But there are good reasons we have de facto maintained our decision criteria: we formed this group with the aspiration of reconciling diverse viewpoints, and the group's very legitimacy flows from agreement among the various stakeholder perspectives.  Moreover, with respect to the chair reconciliation procedure, we additionally don't have the time or personnel to shepherd any sizable share of our open issues through that intensive process.  Certainly not in the six weeks before our deadline.


So, those are the possible outcomes that I see.  Again, I greatly hope we can arrive at a consensus document by July.  That's my most preferred outcome.  I'll be working hard on it.  But in the unfortunate event we don't agree, we would be irresponsible to kick the can down the road (#3), whitewash our disagreements (#4), or move the goalposts (#5).  That's why I think we should establish a contingency plan.

Sincerely,
Jonathan