Why web services will unravel Microsoft's dominance

Microsoft's testimony in the current anti-trust proceedings contradicts its position on web services, points out CBDi Forum in a characteristically perceptive article. The dissenting states want Microsoft to unbundle middleware functions from the Windows operating system, in response to which Bill Gates has testified that that would make Windows less well integrated. But as CBDi points out, "Today we understand that 'integrated' systems are actually very bad news. They are hard to build, maintain and change ... a sensible design heuristic is to use loose coupling techniques ..."

Furthermore, CBDi goes on, Microsoft has led the way in developing loosely coupled protocols architectures in the context of web services, and suggests that, "If Microsoft is recommending its customers engage in this form of loose collaboration we should expect [it] to do the same."

Here's where it gets really interesting, because this is one instance where Microsoft simply can't afford to eat its own dogfood, says the article. "Microsoft is almost certainly between a rock and a hard place ... Sure [it] could publish certain services, but in all probability it would be prohibitively expensive to "fireproof" all the necessary services." Forget all the technical arguments; it has a compelling business reason not to modularize its middleware.

I find this fascinating, because I believe it perfectly illustrates a theory that I'm a great fan of, which is Christiansen's concept of disruptive technologies, as outlined in The Innovator's Dilemma. Subtitled "When New Technologies Cause Great Firms to Fail," this book explains why established companies cannot help but fail to adapt to new technologies. I have every respect for Microsoft's leadership; but equally I believe they are predestined to fail in their efforts to steer this great corporation into the web services age. It simply can't be done.
posted by Phil Wainewright 1:44 PM (GMT) | comments | link

Web services mean fatter clients, remote-managed

It's wrong to believe that web services mean everything will be done on the server and every client will be dumb, as this article from SD Times points out. But there is an extra twist I would like to add. All the excitement about thin clients came because people were fed up with the many headaches involved in managing complex clients, and centralizing everything seemed to be an answer to that. In fact it wasn't, but it gave a strong clue to what users really want — they want all the benefits of rich clients with none of the management overhead of fat clients. The only way to achieve that is to deliver rich functionality to fat clients, but to manage them from the center. Fortunately, web services (combined with P2P and grid computing concepts) enable that.
posted by Phil Wainewright 3:40 AM (GMT) | comments | link

Intelligent agents for dummies

Wouldn't it be nice if you could send an email or text message saying, "Give me last month's sales," and get an answer straight back? Next month, that becomes possible for salesforce.com users, thanks to a new natural language query service announced this week by strongly-backed startup Dejima. Users will send their information requests by email to Dejima's servers, where its intelligent agent technology will translate the natural-language message into an XML query to be run against the salesforce.com database. The Dejima system interprets the response and instantly generates an email reply to the user. There's more about how this intelligent agent technology works on Dejima's website, including a short white paper (PDF), which gives one of the most intelligible explanations of this technology that I've yet seen.

But here's what impresses me. The service runs on Dejima's servers, so installing it is just a matter of setting up an account; and it costs just $10 per month per user. Nor has salesforce.com had to change its application; it simply had to work with Dejima to configure the XML interface to the backend database. This is a great example of the power of on-demand web-based services to increase the capabilities of existing applications simply by plugging in new functionality. I think Dejima could do more to open up that back-end interface, and presumably the company will in the future ... I can see some interesting applications for instance using RSS feeds. Imagine being able to just send an email or text message asking "What's the latest news from the markets?" or "What's the latest news on that earthquake in Turkey?".

This ease of configuration using standard interfaces reminds me powerfully of the last great revolution in computing, when microcomputer manufacturers like Apple and later IBM produced a standard backplane so users could plug in third-party products that extended the capabilities of the machine. To me, that's what XML web services are — the 'PC backplane' of business applications.
posted by Phil Wainewright 2:42 AM (GMT) | comments | link

Business definitions of web services

Two competing definitions of web services, one founded in technology, the other focussing on the business angle, appended to an article in April's Business 2.0:

• "Web services access programmatic functionality over open protocols using XML and HTTP ..."


• "A Web service is a piece of functionality taken from a company's business processes or infrastructure ... made accessible over the Internet ..."

I much prefer the second, even though it's less formally precise than the first. Unfortunately, it's the first type which we hear the most often, prescribing specific implementations of web services architecture before anyone's even finalised the business case for the technology. So all credit then to Phillip Merrick, CEO of WebMethods, for using language that moves beyond the underlying technologies to the nub of the business case.

I'll adapt his language a little more, to arrive at the following definition:

Web services are slices of a business that it publishes to others using the Web.

The body of the Business 2.0 article is full of examples of businesses that have done just that. I think it brings home something that I mentioned previously in the context of writing about Grand Central. These early web services adopters aren't truly doing systems integration or application integration. They're doing business integration. Web services enables them to achieve this by publishing elements of their business in real-time to partners or customers.
posted by Phil Wainewright 2:53 AM (GMT) | comments | link

WebServicesThatSuck?

Vincent Flanders, author of Web Pages That Suck, is looking forward to a whole new generation of bad site design to write about. "SOAP. Look it up. It's the future. We're headed for the World Wide Weblog. The goal is to turn the world into Lotus Notes for the Web — without any Lotus Notes software," he says in this interview on SitePoint. What does that change? "Web design is becoming almost irrelevant. It's all about programming now," he says, which opens a whole new universe where site design can go badly wrong.
posted by Phil Wainewright 6:35 AM (GMT) | comments | link