XML Hype Down But Not Out In New York

April 11, 2001

Edd Dumbill

Signs of reality were setting in this week at XML DevCon 2001 in New York City. As vendors and professionals were feeling the pinch of the economic conditions, the cloud of dust raised by recent overmarketing was starting to settle.

Web Services: "Evolution Not Revolution"

While the vendor evangelists were still busy running around proclaiming web services from the podiums, the top dogs of the industry were more realistic about where this latest trend stood. Norbert Mikula, CTO of DataChannel, reminded the audience at a keynote panel about the lessons of recent history, giving 4GL as an example. Ten years ago we were told that programming was dead, managers could glue together components to create systems, and 4GLs would solve all our problems. That never happened, and programming is still very much with us.

Likewise, Mikula warned the audience about overselling web services. Recent vendor keynotes have seemed a far cry from the not-even W3C-recommended SOAP -- audiences have learned of the car that will automatically reschedule their day and lunch appointments because it and they got stuck in traffic, order up a pair of pants from the cleaners, and so on. Instead, web services are an "evolution, not a revolution", said Bob Sutor of IBM and OASIS, "It's an attempt to clean things up a bit, get the lower level of our stack uniform, and free ourselves up to develop the high layers."

This cleaning up is a valuable and important move, but developers and managers should recognise it for what it is. There are plenty of higher layers that SOAP, WSDL, and UDDI leave unaddressed. In particular, panelists noted that there's a lot of out-of-band human communication involved with business transactions. If it ever arrives, at best the fully automatic conduct of business using web services is a long way away. In a later talk, Uche Ogbuji of Fourthought Inc. observed that, if anything, web services emphasized the need for human oversight of a system due to the vastly expanded (compared to traditional application architectures) range of communication partners and transaction types.

XML DevCon Panel
Keynote panelists Jon Udell, Tim Bray, Norbert Mikula, Peter Chen, David Orchard and Bob Sutor.

SOAP Interoperability

Microsoft was the one large vendor steering away from the inflated visions of automated personal environments. Instead, its keynote from David Reed focused on ongoing SOAP interoperability work. Being so large, Microsoft practically has its own industry within itself, and it's run up hard against the problems of non-interoperability of SOAP implementations among the three SOAP stacks that its developed internally. Add to that the varying degrees of completeness of implementation in SOAP stacks external to the company, and there's an obvious problem to solve before SOAP can be proclaimed lingua franca of the web services world.

Although conference attendees might be unused to seeing the DOS prompt up and command line executions of programs during keynote speeches, it was a useful reminder that there's a lot of hard work to do yet, and that the payoff of glitzy promises may be some years off.


XML is no longer the pure, world-uniting force for openness we all believed it to be a few years ago. Its very success, and the consequent influx of dollars, has changed the nature of the information that developers are being given at conferences. None of the vendors quite knows where we're running to, but they're running faster and harder than before to get there before their rivals. IBM, Microsoft, and Sun's latest mantra is to "cooperate on specifications, but compete on implementations"; and competing they are. A panel on web services saw speakers keen to proclaim their implementations as more complete, more advanced than the competition. Never mind that in a show of hands, under 50% of the attendees had ever touched SOAP, and even less so the upper levels of the emergent infrastructure.

Despite vendor assurances, it seems plain that the race to compete on implementation puts the viability of multivendor institutions like OASIS at risk. The increased rivalry cannot but help spill over into these environments. One example is the size of the newly formed XML Protocol working group at the W3C. With over forty participants, the size of the Protocol WG exceeds even that of last years' political football, the Schema WG. Everybody wants in, and the most likely victim will be the protocol specification itself.

Innovation Still the Highlight

The XML world is still a young one in terms of practical development experience, and the most interesting sessions at XML conferences are ones where individual developers share the techniques and tricks they've discovered for processing XML. The conference had a great team of speakers, and conversations and debates in the corridors and bars proved more instructive to me than a thousand keynotes ever could.

However, it will be fall before the next onslaught of XML developer conferences, and I hope the rest will do the industry good. Some of the best developers and speakers are looking tired after eighteen months of practically non-stop conferences, education and debate; and too many of the presentations have changed but little since the middle of 2000.

The relentless flood of specification development and marketing endangers the discovery and communication of solid XML programming experience. It's time to slow the hype machine a little, take a rest, and get on with building practical systems.