Sr. Director, Product Research, Epicor Software

Software Architecture

This blog is not associated with my employer.

Tuesday, March 27, 2007

WS-I and Transparency: WTF?

Over 300 people from 100 companies came to San Francisco in early April 2002 for the first ever WS-I community meeting. I was there. Like most of the attendees, I had never worked on a standards effort of any kind. But the "founding board" members of the WS-I – IBM, Microsoft, Oracle, SAP, HP, Intel, etc. – had each reached out to their customer bases to drum up participation. The carrot hung out the window for us smaller companies to sniff was that we could hang out with the big guys to work on web services standards and, importantly, we could get a WS-I conformance logo to grace our product.

It's exactly five years later and I'm in Delray Beach Florida for the WS-I Spring 2007 Community Meeting. The event's web page (for members) list says I'll be one of 32 people attending. Only 18 companies are being represented. The WS-I Board is made up of 11 companies, who, one would think, are obligated to participate. So the "community" beyond the leadership team will be me plus 6 other companies. Over 70 other companies pay dues to keep their up their membership, but I'll wager that many (like my company) have fallen out of good standing.


Marketing – There are test programs that determine whether an application conforms to a WS-I profile. You don't need to belong to the WS-I to run them – or even claim you've run them successfully. You can just say your app conforms to a WS-I profile. It's up to someone else to call your bluff. With membership optional and conformance claims free for the taking, it's no wonder the membership is practically extinct. My company's dues are paid out the marketing budget. Hmmm.

Resources – A tenet behind the WS-I charter says that, for any profile, a set of sample applications and a test suite identifying test assertions will be developed on multiple platforms. The idea was to show best practices and demonstrate the potential interoperability benefits behind the effort. But developing this work is enormously expensive. Microsoft, IBM, SAP, Fujitsu, and others have deep pockets. But over time the contributions have declined to the point where the organization can't "afford" to charter new profile work. But there is a bigger challenge to the long-term solvency of the WS-I…

Transparency – This is the big one. To be fairer than necessary, the WS-I was conceived at a time when web services hype was still growing and the founding members were deeply suspicious of each other's motives (the players, I think: IBM, Microsoft, Oracle, HP, Intel, SAP, Accenture, BEA, and webMethods). The hype issue has almost disappeared. And the mutual suspicion problem is in a little better shape except for a lasting grudge within the WS-I Board over the fate of an effort 3 years ago to incorporate SOAP with attachments into the WS-I Basic Profile.

But there is a bigger problem. There are 2 WS-I organizations: the working groups and the WS-I Board. Of course, all organizations have boards that are independent and need to be able to deliberate privately. But the WS-I Board – and I mean the institution, not the individuals or their companies – is frankly an anathema in the face of the WS-I's public mission.

WS-I working groups deliberate within narrowly defined charters, but the discussions are open. Members are free to discuss and describe deliberations wherever they like. The only caveat – and it's significant – is that no "intellectual property" developed by working group members can be published as WS-I material without Board approval and (depending on the nature of the work) as membership-wide vote. Most members can live within those constraints. But the Board is another issue.

The WS-I Board was intentionally designed to be a vendor-driven black-box that controls what is advertised (with a bit of mockery) as a membership community. This isn't hyperbole (despite my usual tendencies). Here are the rules that are driving the WS-I straight into the ground:

- ALL WS-I Board proceedings are confidential – even to the membership. There was an effort by IBM a year ago to prove that Board members could discuss what happened in meetings even if the minutes were secret. But that theory was rejected by others in the Board as a reckless interpretation of the bylaws.

- Virtually all proposed Board actions – like approving a document for publication or launching profile work – requires 9 of the 11 Board members to vote "yes". Abstentions count as "no" votes (a rule which is potentially being addressed)

- Of the 11 Board positions, 9 are allocated on a permanent basis to "founding members". The other 2 are elected. Of those 2, one is effectively (and justifiably) locked up by Sun Microsystems, who was obviously blackballed from participating when the WS-I was formed. But the point is that there is no real opportunity for WS-I members to become part of the leadership.

- The Board determines what work is in scope for the WS-I, but has been utterly unable to declare the criteria behind making this determination. Part of the reason is that the Board is trapped by a wacky bylaw treatise inured with a declaration of unambiguous consensus. In other words, the board can't give its own membership, even internally, any feedback without 9 of 11 votes affirming both the act and the message.

How in the hell can members have any involvement in determining future profiles and other work for the WS-I when (a) there are precious (and highly subjective) rules about what work is in scope and (b) Board members can't agree about what, the bylaws allow them to say to their own membership? The WS-I is the epitome of dysfunction and the reason is mainly the misguided invention of that whacked-out cocoon in which the Board comfortably resides.

So, I'm participating in a Working Group chartered to come up with requirements for future WS-I profile work. To be fair, I've been away from the WS-I for a year, so I feel bad about making waves on the first day back. But I know the workings having chaired their XML Schema Planning WG in 2004-2005. Anyway, the Requirements WG is only allowed to submit draft Working Group charters for WS-I Board approval. Astonishingly, our Working Group can't even give feedback to OASIS or ask the W3C a question without a 60-day approval processes and WS-I Board vote. We can't contact the organizations that own the material the WS-I is considering for potential profiling efforts

I'm going to propose that the WS-I Requirements working group take several actions: (1) Questions and feedback to other organizations are deemed "non-material work", which means the whole WS-I membership does not need to vote before publication. (2) That approval to make materials developed by working groups or committees public requires a simple majority (I can be talked into a 2/3 margin) to be approved for publication. (3) That votes by the WS-I Board on work submitted by a working group or committee are made public to the membership (and by extension – the world). Of course the Board and the bylaws will have to yield and be amended, respectively.

Without these reforms (and others), the WS-I will continue its utter collapse into a small club of vendors seeking to satisfy a few constituents that, for now, are propping up the justification behind the entire organization. The rest of the world (spot the pun) will overtake the WS-I like callous covers a splinter on your heel. And while I strongly think there are much better programming models on the web for many situations, I also think there is a large place in the economy for WS-*. Large organizations will need the mechanisms behind WS-* as provable standards for to bind their touch points. And large IT companies will wave WS-I profiles, and the like, as bespoke competencies fulfill them.

Maybe the WS-I should be disbanded for the shear arrogance of its founders who squander the enthusiasm and contributions of the membership. But starting over also sucks.