Late last month CNET's news.com published an interview with Robert Glushko, one of the earliest movers in the use of XML in e-commerce. Glushko is no stranger to the technical standards game, having been involved with both OASIS and the UN's CEFACT, as well as with ebXML.
In the prelude to the interview itself, the writer, Martin LaMonica, suggests that "Web Services", the creation of Microsoft and IBM, were promoted at the expense of ebXML, primarily in order to please Microsoft and IBM. That Microsoft paid travel expenses for some UN officials as part of a CEFACT mission is offered as proof that the standards process is "very politicized".
Well, of course it is! Standards are, after all, views of the world, and I've yet to see a view of any part of the world as highly contested as "commerce" be anything but political. What else could it be? I suppose, however, that Glushko really means by political is "unfair", which is to say, arranged so as to favor large technology companies. Well, of course it is! This general move away from standards bodies like the ISO, rooted in democratic institutions and democratic national governments, is part of the general contraction, at least in the United States, of the public sphere along with an aggressively deregulatory legal and political environment, and it goes back to at least the early 1980s.
And, more to the point, that contraction of public space and creation of a deregulatory environment predates not only XML, and the commercial rise of the Internet, but nearly predates the creation of the personal computing market itself. That some tech companies are benefiting from it should surprise no one who's been paying any attention whatever to politics and business in the past 25 years. The remarkable thing about some of Glushko's claims, but more about the overall tone of the article itself, is that the tech giants are singled out for criticism that's equally due just about every multinational corporation.
Glushko also blames "people like you (in the media)" for misusing the term "standard". In his view, the ISO, ANSI, and the UN create standards, while OASIS, the W3C, and WS-I create "specifications that occasionally have some amount of consensus". How do we know this? Apparently, ISO-ANSI-UN create standards because they are, in Glushko's view, "standards organizations". What, in Glushko's view, is a standards organization?
a standards organization is something that is chartered to be a standards body that has some standard procedures...To me, a standard, in the best case, is some kind of specification which is developed by consensus of all the serious players and for stakeholders of some domain. (Standards development) has some open process and (the standards) are freely available and implementable.
That definition applies perfectly well to the W3C and to OASIS. What Glushko's definition leaves out, of course, is any notion of working in the public interest, of democratic accountability or oversight, of regulatory compliance or creation. But public interest groups are as free as any other organization to participate in W3C Working Groups and OASIS Technical Committees. In fact, I suggested that they start doing so as early as 2001, in the first thing ("The Politics of Schemas") I ever published on XML.com.
Web Services Versus ebXML?
But the most remarkable bit of the Glushko interview is the establishment of a competing, antagonistic relationship between something Glushko calls "Web services", which are, tout court, "proprietary specifications" and ebXML. In fact, Glushko takes a page from Richard Stallman's book and corrects LaMonica when he refers to "Web services standards".
When pressed to justify this extraordinary set of claims -- first, that ebXML is a competitor to "Web services" per se; second, that "Web services are proprietary specifications" -- Glushko points to WS-I's intellectual property policies! So, let's make sure, for those of you following along at home, that we understand Glushko's claim: All "Web services" technologies -- including, for example, SOAP, which is indisputably the most foundational web service technology -- are "proprietary specifications" because there is an organization, WS-I, which has a bad, or so Glushko claims, IP policy.
What Glushko hopes you don't notice or know, of course, is that SOAP is implemented everywhere, by everyone; that the W3C has an excellent set of IP policies; and that many, even if not all, web service technologies are standards according to Glushko's definition.
Let me also point out, though I have neither space nor time to develop this much, that Glushko's other set of claims about why the ISO-ANSI-UN stratum is superior to the W3C-OASIS stratum (that Glushko wants to include WS-I is no reason that we should think of it as allied with the W3C or OASIS) is that it's too expensive to participate in more than one of these organizations, if you're a small company, and that travel to the face-to-face meetings is expensive. Apparently I've missed out on all the free travel to ISO-ANSI-UN meetings. And Glushko's claim that the exotic locales of these meetings -- he mentions Tokyo, Berlin, Vienna, Vancouver, and Washington -- makes participation too expensive is patently absurd. That these meetings are not always in the US can be equally used to bolster the claims of openness and even globalist perspective.
Frankly, I read this interview with Glushko and found it totally useless ("web services versus ebXML" is not worth discussing) where not trivially obvious ("standards are political"). So why am I writing about it today? Because it spawned a very long, detailed discussion among XML developers, which I will try to summarize in the remainder of this column.
XML Developer Reaction
Rich Salz, one of my fellow XML.com columnists, pointed out that the ISO-ANSI-UN stratum is no less political than any other stratum:
...the acronym ISO had to be deliberately chosen to not stand for anything in English so that AFNOR (the French national standards body) wouldn't walk away? Claiming ISO is any more or less political than W3C, IETF, ANSI, IEEE, OASIS, et al, is ridiculous.
While Salz is right, the difficulty is that "political" means a lot of different things, and there are some kinds of politics that really are preferable to other kinds. But Salz also skewers Glushko's other claim, about the IP policies of WS-I:
WS-I is not a competitor to OASIS. WS-I does not add any IP claims to other's standards. Yes, there are things in the IBM/MSFT web services stack that are proprietary...but they're not part of WS-I...did he read the WS-I IP document? It binds everyone who joins to [a] don't-sue, cross-license agreement. You cannot get a stronger RF policy!
David Megginson pointed out, and rightly so, that Glushko fails to mention the Internet Engineering Task Force, which isn't a standards body by Glushko's definition, but which has successfully guided the Internet for a very long time.
Tim Bray, now employed by Sun, offered some history in response to the claim that the ISO was "predictable":
...the notion that ISO is "predictable" really can't be allowed to go unchallenged. Some may have forgotten those days in the early nineties when any whimsical HyTime-related theorizing could get ISOfied in weeks, in one case with insiders sitting up the night before ratification, long after the votes had been counted, hacking the query language... then there's the way that IBM used to get proprietary extensions into SQL by having small-country IBM field offices arranging to cast the votes for their countries.
I'll say for my own part that I find the ISO one of the most interesting institutions around, but mostly from the point of view of sociology. The problem with the ISO from my point of view as an open source developer is that its standards are all but inaccessible to me, since I refuse to pay hundreds of dollars to own copies, and since the ISO refuses to make them available on the Web for free. I would happily use ISO standards more often, and I might even implement one or two, if the standards themselves were easily accessible. That's just basic. Further, the ISO may well be undergoing some kind of institutional change for the worse. How else to explain the stupid idea to charge royalties to use ISO country, language, and currency codes? (See my "ISO to Require Royalties?" for that story.)
Megginson offered another characterization of the ISO, pointing out that it has never really meshed well with computer technology:
The fact is that ISO never did well with computer technology, either before or after XML. We use the four-layer DoD networking stack, not the seven-layer ISO/OSI stack, and we look to the IETF, not ISO, for our protocols. Even modest computer tech successes like SGML have been rare for ISO.
Glushko Responds
Glushko himself joined the discussion on the XML-DEV mailing list, providing some crucial bits of the story of his work with Veo and Commerce One:
But now that I look back at it, I have to conclude that we vastly overinvested in standards activities, spending gobs of money to participate in the W3C or OASIS or ebxml or UBL to do "good work" whose direct ROI for us was minuscule. We might have been better off investing our time and talent in products rather than standards because a lot of them got co-opted or undermined.
He also praised the CNET writer, Martin LaMonica, who "got more things right with this interview than any one I've ever done". However, Glushko clarified his actual position:
I do think it is fair to say that neither OASIS, W3C, nor WS-I are standards organizations according to the definition I advanced here. But I don't think it is fair to lump the W3C in with WS-I on either openness or IP terms, and I'd hate for people to make that inference. The W3C worked very hard to put a royalty-free policy in place while OASIS and WS-I have aggressively resisted one.
Glushko glossed his claim about face-to-face meetings, using his experience with ebXML as a guide:
Holding the meetings all over the world undermined our progress, because at each new city there would be a sizable group of new participants who could come because it was local or nearly so. We'd spend a lot of the time just getting them oriented. So while I can appreciate Paul Downey's argument for geographical diversity it comes with both larger economic costs and at a tax on productivity.
There is far more to be said about these issues than can be summarized in this article. Standards aim at many different purposes, including market regulation, creation and consolidation. I think it's reasonable that these different purposes might be served best by different kinds of standards organizations; but all of them need more input, in my opinion, from people and institutions that work explicitly in the public interest. But, even with that input, there is no magic bullet for making good -- that is, fair, balanced, and technically elegant -- standards. There is no magic bullet precisely because, at bottom, making standards just is politics by a different name. And no one should be surprised that, at bottom, politics is a messy, messy affair.
What's your opinion on the Glushko interview?
(* You must be a member of XML.com to use this feature.)
Comment on this Article
| Titles Only | Titles Only | Oldest First |
- Your Own Politics Are Showing
2004-05-13 07:24:51 Len Bullard [Reply]
Since I started that thread, let me reiterate the points that you are skirting:
1. Because of the rise of consortia and the proliferation of immature specifications labeled standards by an inexperienced press, the technical landscape is now litered with overlapping competing "standards". The term is now meaningless.
2. The role of standards traditionally supported procurement based on citation of international standards. This worked reasonably well with some exceptions.
The time has come to give 'standard' a meaning that supports both technical excellence and reasonable procurement policies. In my opinion, there are three legs, an organization must stand on to create legitimate standards:
1. Works through a legitimate and recognized standards organization with transparent processes. This means that while a consortium can create and vette the technology, the actual work of editing and shepherding the standard should be done by professionals. It isn't always apparent from the point of view of the technically trained how much of that work is legal work.
2. Must insist on participation agreements that clarify the status of all Intellectual Property constraints. Ideally, this is royalty-free implementation and indemnification for all members against lawsuits. This stops the SCO nonsense up front. This stops the Microsoft and Sun patents from having any more meaning than they filed the paperwork without due dilligence. We can't fix spilt milk but we can get smarter by learning how to stop the IP wars before they start.
3. Must offer conformance testing with a service mark for products which pass the conformance tests. This proves legally and technically that a product is what it says it is
with regards to the standard.
With these three legs, one knows:
a) What a standard is. It isn't a specification. It isn't proprietary and encumbered technology with a patina of 'standard wording' on it.
b) What one must do to comply and how to prove it.
c) What to expect. It is safe to implement the technology without keeping a stash of money for the lawsuits.
Without such conditions, the grand experiment in open systems, open source, and equal access will grind to a halt.
To quote again,
http://www.tdan.com/i016hy01.htm
"What Makes a Standard a Standard?
"Simple. Not implementation, but conformance. And, conformance is "known" only after
conformance testing. Then and only then can users know with any degree of certainty that a vendor's product conforms to a standard."
It is time to take the personalities out of this process, and to balance the role of the consortia that create technical specifications with the role of the international standards organizations to provide international standards.
Len Bullard
