formal vs. prose and PR [was: Re: Final minutes QA WG Teleconference May 16]

At 9:34 AM +0200 5/23/05, Dominique Hazaël-Massieux wrote:
>Hi,
>
>The minutes of the meeting the QA WG held on May 16 are available at:
>http://www.w3.org/2005/05/16-qa-minutes.html
>
>The main topics of discussions where the resolution to request the
>transition of SpecGL to Proposed Recommendation,

Which was resolved in the affirmative.

http://www.w3.org/2005/05/16-qa-irc#T15-13-33

Despite the outstanding issue

"Formal vs prose language normativity"

.. for which the QA Working Group has failed to do due diligence to
resolve the issue.

That is to say a phalanx of consistent comment from customers has been
ignored, and the Working Group has left an ill-considered requirement
in the document.

That by itself should make the Director "just Say No" to a PR request
at this time.  This is not executing the W3C Process with acceptable
quality.

Trying to make specifications repair their own errors by requiring a
tie-breaker rule is futile. They can't. It is not going to work. Thus
doesn't assure the quality of the Web.

When a specification is found to contain an internal inconsistency
after it is developed, published, and implemented, one can't just
look inside the specification for how to resolve the error. It is a
failure in encapsulation, of separation of concerns. To decide the
appropriate action to take, one has to consider a wider horizon than
just the specified technology and its governing document.

One has to look outside to the load, the actual traffic, that the
services following this specification are carrying in the operational
configuration of the Web. Consider the actual cost of
miscommunication (because of disparate interpretations of the
specification) with the actual cost of mis-features (because the
tie-breaker imposes an inferior design).

When theory fails, we have to get pragmatic.

To pretend otherwise discredits our effort.

Asserting a precedence that the development group genuinely believes
in is fine. Requiring a precedence in the absence of that collective
opinion is counter-productive.

W3C, whose line of business depends on selling the value of open
written specifications, must be modest as to our claims for what this
accomplishes. Spurious claims, given the self-proclaimed monopoly
position of the W3C in discerning the common good, will be magnified
in the perception of our customers.

Every futile provision we leave in this document detracts from the
traction the W3C holds with Web implementers. Don't go there.

Al

Received on Monday, 23 May 2005 16:03:32 UTC