Nick Carr's essay blog
January 15, 2005
It may be the ultimate machine dream: Connected through the Internet, computers do business together, swapping complex data on production and demand, reconfiguring processes in response to market signals, tying together elaborate supply chains on the fly. It’s clean, it’s precise, it’s automatic. It replaces the messiness of human relationships with the cold logic of code.
You might think that the dream had died with the hype about “B2B exchanges”—those electronic marketplaces that were supposed to render traditional supply arrangements obsolete. But it hasn’t. It continues to influence perceptions of IT trends ranging from real-time operations to business process management to the evolution of the Internet. Nowhere, though, is it so evident as in discussions about web services.
Web services, in simple terms, comprise a set of software tools and standards that allow diverse applications to talk to each other over the Internet. At a technical level, they’re an exciting and important development. As with middleware and read-anywhere programming languages like Java, web services help managers escape the maddening incompatibilities of legacy systems. They promise to allow companies to weave together their own information systems—and connect them with the systems of partners—without having to resort to expensive, time-consuming, and risky code rewrites. To resort to the common jargon, web services replace rigid “tight couplings” between applications with flexible “loose couplings.”
Things get dicey, though, when the discussion shifts from the technical level to the business level, when web services start to be promoted as the glue for a new type of enterprise. In this model of future commerce, software is used to automatically and seamlessly connect the processes of different companies, allowing the quick assembly and disassembly of entire value chains reaching from the supply of components to the delivery of finished goods. In turn, businesses evolve into increasingly specialized modules that can be easily plugged into these flexible new enterprise networks. Joined in a “service-oriented architecture,” companies themselves begin to look and act like loosely coupled software applications.
So what’s real, and what’s fantasy? The answer may lie in an article by Andrew McAfee, a Harvard Business School professor, in the new issue of the MIT Sloan Management Review. In Will Web Services Really Transform Collaboration?, McAfee takes a scalpel to the common wisdom about web services. “If information systems could look for and find each other, share data and execute business processes, all with no (or very little) human involvement, we would probably find ourselves in a very different business world,” he writes. “Web services, however, will not create this world, nor will any technology on the horizon.”
McAfee explains that computer-mediated collaboration between companies requires three very different kinds of agreements. The first and simplest are agreements on “transport”—the networking protocols that allow applications to connect. The second are agreements on “payload”—the data standards that allow applications to share information. The third and most complicated are agreements on “process”—the sequence of activities in a work flow and the allocation of responsibility for them. The problem, McAfee points out, is that web services only really automate transport agreements: “They make it possible for two applications to talk to each other[i.e., transport], but they don’t specify what conversations they should have [process] or what words they should use [payload].”
It’s possible for two or more companies to negotiate agreements about payload and process but, as always, that takes a lot of human interaction and a lot of time. McAfee describes how it took more than a year for IBM and one of its distributors to hash out the shared protocols necessary to create an automatic, computer-to-computer ordering system. Such efforts will be worthwhile for automating stable, high-value connections between the processes of large companies—the kinds of processes that used to be linked through electronic data interchange, for instance—but otherwise they’ll rarely be worth the hassle. And they fall well short of the kind of fast, flexible connections that the promoters of web services often promise. Web services are unlikely, in other words, to usher in an era of modular, plug-and-play value chains.
I see McAfee’s argument as providing further evidence of the existence of what I’ve come to call “the technologist’s fallacy.” As I explain in my book Does IT Matter?, computing professionals and pundits sometimes exhibit a tendency "to confuse business with information processing, to want to see companies as, in essence, computers. They overlook, or give short shrift to, the physical and human characteristics of commercial organizations—to all the things that can’t be reduced to digital code, that can’t be ‘exposed’ or ‘made transparent’ through networks. This skewed perception leads them to conclude that companies, like computers, can and should become components, or modules, in broad and flexible networks.”
Web services will play a key role in overcoming incompatibilities between information systems, helping create a more standardized and productive IT infrastructure for business. But they’re not going to usher in an age of protean value chains. And they're not going to usurp the place of managers in negotiating, overseeing and modifying complex partnerships. Machines matter, but people matter more.