By Eric Knorr
Quick—how do most e-commerce dollars change hands?
No, not through big B2B marketplaces such as Covisint. And not through B2C sites such as Amazon. The fat pipe for e-commerce dollars remains electronic data interchange (EDI), a standard born in the 1970s that enables businesses to order and bill electronically, typically for raw materials used in manufacturing. People don't think of EDI as e-commerce because most of its connections are still proprietary rather than Internet-based. But whatever you call it, EDI continues to transfer many times the dollars conveyed by Internet-based commerce, according to most estimates.
Most of those EDI dollars flowed among a few hundred big corporations, mainly because EDI links are expensive to establish and maintain. For years, companies that relied on EDI lamented its limited reach and high cost. So in 1999, corporate EDI users, along with IBM, Sun, Ariba, Commerce One, and various other members of the OASIS standards group—plus the United Nations—decided to establish a rich common denominator that would make electronic trade relationships easier and cheaper using XML over the Internet.
In May 2001, after 18 months of grueling work, ebXML was born. My question of the week is: Why is Microsoft opposing—and why is IBM all but ignoring—ebXML?
Obviously, ebXML's corporate proponents have money to spend on technology. And they drew on decades of electronic trading experience to develop a common standard for companies to engage in machine-to-machine commerce. Moreover, according to Jon Bosak, a distinguished engineer at Sun (who led the original W3C working group that created the XML 1.0 specification), ebXML is gaining traction among governments all over the world, particularly in Asia. But mention ebXML to Microsoft, and you'll get your head bitten off—while the only IBM product that supports ebXML is the obscure Trading Partner Interchange software.
Instead, by all appearances, Microsoft and IBM appear to be in cahoots to build a Web services infrastructure that will eventually subsume the functionality of ebXML. (The first collaborative effort, WS-Security, was announced in March.) IBM's director of e-business standards strategy, Bob Sutor, was a lead player in the development of ebXML—which he recently told me was merely a first step. "The ebXML work we have now is the first generation to move what has been done in EDI to XML and the Internet," he says. "When we as an industry are done, the infrastructure will contain far more than what either ebXML or Web services contains today."
David Burdette, director of Web services product management for Commerce One, acknowledges that ebXML focuses pretty much exclusively on B2B, a narrower scope than Web services. But he also disputes any implication that ebXML (particularly version 2.0, released last month) is not ready for prime time, asserting that it is "completely fit for the purpose of B2B messaging in its current form."
Three points argue in favor of the adoption of ebXML: It exists, dozens of governments have launched initiatives that support it, and the EDI people behind it move more dollars over the wire in one year than Internet e-commerce industry has in its history. Sun is the only big technology company still pushing ebXML. That, and the fact that Microsoft's BizTalk standards competes with ebXML, could easily account for Redmond's hostility. IBM's motives are more mysterious. Could it be that the company wants to develop standards with Microsoft over which it can exercise intellectual property rights, which it recently relinquished for ebXML?
Sun's Bosak acknowledges that if Microsoft and IBM want to, they can crush all chances for ebXML's success. At this point, I think we need to hear a very good explanation of why all the work that was done on ebXML should be passed over. The ebXML standard is anything but moribund—and a related OASIS initiative for a Universal Business Language is well on its way to developing a standard library of XML business documents.
The global economy needs all the help it can get right now—and it needs open B2B standards, not proprietary ones.