- From: Bijan Parsia <bparsia@cs.man.ac.uk>
- Date: Wed, 24 Oct 2007 14:57:29 +0100
- To: public-owl-wg@w3.org
This is my summary of the discussion instigated by my following email: <http://lists.w3.org/Archives/Public/public-owl-wg/2007Oct/0079.html> I provide it to help with disposing (whether affirmatively or negatively, in whole or in part) of my proposal in today's telecon. I've aimed it to be narrow rather than comprehensive, evidence based, brief, neutral, and fair (sometimes neutrality isn't fair, for example, but I've made no prioritization). Occasionally, I may interject an editorial aside with my position, though I'll strive to keep these to a minimum, preferring toward none. So lack of comment or analysis does not indicate that I agree or disagree with a point. It ended up being a lot longer than I'd hoped :( THE PROPOSAL: In the last telecon, I proposed that we take the core formal documents of the OWL 1.1 specifications, as appearing on webont.org, and publish them as first public working drafts as soon as technically feasible, that is, within a few weeks. These are: Structural Specification Formal Semantics RDF Mapping Together, they form an update, albeit incomplete, to the OWL 1.0 Semantics and Abstract Syntax document. They are the most worked out, albiet incomplete, specification of a possible OWL 1.1 language. There is also considerable implementor and user experience, as well as nearly a year of feedback. In the telecon, the proposal was split into two parts: 1) whether we should commit to publishing the core trio by our first heartbeat requirement, and 2) whether we should publish as fast as possible. The group made a decision on the first part (slightly modified to allow dropping the RDF Mapping). PROCEDURAL ISSUE: This decision was challenged on procedural grounds. Regardless of the merits of those challenges, the chairs are considering a procedural proposal to 1) only put questions to the group that have been more or less (I'm unclear how much) explicitly mentioned in the agenda 24 hours before the telecon, and 2) to allow a "cooling off" period of a week on any decision. REASONS IN FAVOR: 1) WD drafts are widely advertised and very natural targets for review. Thus, publishing the input design as FPWDs has the potential to attract the most possible feedback from the widest group of reviewers. 2) Getting through a first publication gets us through all the technical hurdles. The earlier that's done the better 3) It establishes a publish often ethos 4) It helps establish good momentum. It seems clear that the proposal is correlated with some more aggressive reviewing than might have otherwise happened. For example, I believe between the first and second telecon, I was the only person who posted any document reviews to the list. Now there have been several and issues I did not *imagine* existed were raised. 5)It triggers early IP consideration. A key basis of the proposal is that WDs, even FPWD, do not entail or suggest commitment by the group --- de facto or de jure --- to either the design or the presentation of those drafts. See: <http://www.w3.org/2005/10/Process-20051014/tr#first-wd> REASONS AGAINST: I count at least 4 people explicitly voicing opposition against various aspect of the proposal (part 2, but many apply to part 1). A) People feel that WD ought to be "worthy", reflect group consensus on the design (this one is unclear to me; it's seems to be signaled by the phrase, "consensus document"), that they fossilize design, they should reflect work the group has done, or they need to be "ready". It was also insinuated that publishing some of the drafts would be publishing a non-"honest" WD. EDITORIAL: I would prefer that people take more care in their choice of language. For example, there's no reason to say that my proposal involves some level of dishonesty. One can just say express a preference, or aversion, or point to *material* consequences in a neutral way (e.g., "I think enough people will take the WD as reflecting fixed design to cause more trouble than the likely benefit of wide review is worth.") It helps to try to take into account expressed differences of fact. B) People expressed the desire for more group review before publishing anything. The consequential benefits haven't been clearly enumerated to my ken. C) At least one person said in private email to me that they didn't think wide review was necessary or desirable. D) Several people expressed discomfort with the general idea. E) It was claimed that WGs are not supposed to comment on WDs after they've been published. F) It was claimed that no significant additional review would be triggered. G) Early IP consideration might force IP actions which are unnecessary as the design may change. This was presented as a pure hypothetical. I.e., no specific IP claim was motivating this. H) I got another email stating that WDs should either reflect the state of consensus (which doesn't me we all endorse the design or documents) or be clearly labeled as lacking that. The main difference from my proposal is that they preferred finer grained indication of what is actively being disputed. However, they also wrote that my proposal was non-harmful and they wouldn't block consensus to publish on this basis. (I made a similar proposal as "next steps": <http://www.w3.org/mid/BA6A0FF1- AEA1-42C7-924F-7000CA97202E@cs.man.ac.uk> Jim weakly endorse this without the Oct first publish, i.e., as a goal for the F2F). PARTIAL AGREEMENT? Several strong opposing voices stated that they felt that the Formal Semantics document was OK to publish. This has, afaict, consensus among the people opposing as something OK to publish, pretty much anytime. * Jeremy strongly opposed the RDF Mapping unless it was essentially stripped of all content. That suggests to me that he is ok with a separate RDF Mapping document.: <http://lists.w3.org/Archives/Public/public-owl-wg/2007Oct/0174.html> * Elisa express opposition to publish the Structural Specifiction (or the MOF metamodel, although that's not in the proposal) on the grounds of lack of alignment with ODM. <http://lists.w3.org/Archives/Public/public-owl-wg/2007Oct/0161.html> I thought, but cannot find, that she said that merely dropping the UML diagrams from the structural spec would leave it in a publishable state from her POV. * Jim expressed opposition to the structural specification *per se*: <http://www.w3.org/mid/F9A27E86-A5E8-4D40-8E77-FE17CF9F0247@cs.rpi.edu> I'm unclear whether this means he objects to its publication for review. He also expressed some criticism about the RDF Mapping: <http://www.w3.org/mid/A98D4B0B-4CC3-4FC8-9B56-41B641DF047D@cs.rpi.edu> But I'm also unclear whether this means he objects to publication for review. MY DISCUSSION: I still believe that there is considerable benefit and no downside to my proposal. So I continue to put it forward.However, even just publishing the Formal Semantics gets reasons 1-5 *for it*, and I think it completely settles 2, and goes a long way to 2 and 4. As a subset of my proposal, I obviously support it. I think I'd rather not publish the Mapping or the Structural Specification as WD at this time, than publish the redacted versions of the Mapping or the Structural Specification. The problem with redacted versions is that it shields from wider review precisely that which is controversial. There has been strong criticism of those documents, but I, for one, do not agree with most of it. There is also strong community uptake as witnessed by the several implementations and very positive implementor and user feedback (e.g., at OWLED 2007). We should take this into account, not because these things are immutable, but that they are very good evidence of good and successful design and presentation. In fact, to the degree that it is controversial, I think we should publish. If we don't publish, I think it would be helpful if group participants made some efforts to solicit significant feedback. I, obviously, shall do anyway, but if opponents agree with the goal of wide review, but disagree with the proposed means, I would certainly take it very positively if they helped reach that goal by other means. Corrections or friendly amendments are most welcome. Cheers, Bijan.
Received on Wednesday, 24 October 2007 13:56:04 UTC