Loosely coupled invention

To glimpse the awesome potential of loosely coupled systems, take a look at BookWatch Plus. Thousands of discrete web sites and services interact to make BookWatch Plus possible, and yet none of them were designed with the intention of creating it. Here's a rundown:

• Thousands of webloggers each day post to their weblogs, and as part of the routine of posting they alert weblogs.com. This is a service that's been operated by Userland to promote blogging since 1999.


• Weblogs.com checks the weblog, and if it finds the page has been updated, it adds it to its Recently Changed list, which it publishes both in HTML and as an RSS-format XML feed.


• For the past several months, OnFocus.com has operated a service that reads the Recently Changed RSS feed and searches each weblog listed there for new links to books at Amazon.com. It uses the results to publish a list of the most frequently mentioned books on its Book Watch service. It too offers an RSS feed.


• In April, Google released its API, which allows developers to access search results and other information using web services.


• This week, Amazon followed suit with an API of its own, and shortly afterwards Amazon employee Erik Benson added the BookWatch Plus service to his Mockerybird site. This reads the RSS information from Book Watch, adds more information about the book using Amazon's API, and fetches related news information for each title using the Google API, before publishing the aggregated information as a web page.

Why is this awesome? As Tim O'Reilly writes, "Innovation will come from APIs that support 'unintended consequences'." By exposing their output via APIs or as RSS feeds, all of these participants have plugged into a network of innovation, where the potential outcomes are beyond the imagination of any individual participant. And remember, this is just what people have done for fun, as an experiment. Think what they'll be capable of doing for money, or for power.

Tim O'Reilly's article provides an excellent rundown of why more businesses should start publishing APIs, and also provides some background on how and why he personally encouraged Google and Amazon to take a lead:

"Inspired by the ever-prescient Jon Udell, I started talking up web services back in my March 2000 JavaOne keynote. I was fascinated with Jon's idea that web services could do for the internet what the shell and pipes did for the Unix command line — create a loosely coupled architecture in which people could build new functionality out of small, independent tools. But I was disappointed to see that web services seemed to go off into an enterprise black hole (what Clay Shirky calls EDI++), rather than becoming the freewheeling next generation internet programming and power user environment that Jon and I had imagined. So I started lobbying some of the interesting web sites to start offering SOAP APIs."

So hats off to Tim for helping to make this happen. My thanks also to Julian Bond, whose email to me this morning made my break my usual rule of never posting on a weekend. This one wouldn't wait. Apart from anything else, it finally gives me a context for quoting from Small Pieces Loosely Joined, a book that I'm surprised not to see on the Book Watch list (but which, through the magic of loose coupling, will get a nudge up the rankings when I post this item). It's by Dave Weinberger, who is a co-author of cult blogger's handbook The Cluetrain Manifesto, and the quote is from the preface, which you can read online:

"The Web is changing our understanding of what puts things together in the first place ... most important, the Web is binding not just pages but us human beings in new ways. We are the true 'small pieces' of the Web, and we are loosely joining ourselves in ways that we’re still inventing."

posted by Phil Wainewright 9:32 AM (GMT) | comments | link

Owning the Universal Canvas

The next release of Office moves closer to Microsoft's oft-repeated vision of the Universal Canvas — which, as I recalled last week, promises to enable users to "access all of the rich function of the collaborative Internet without ever having to leave the Microsoft Office environment."

Jon Udell reported with some excitement last weekend the keynote at Microsoft's partner conference by group VP of productivity and business services Jeff Raikes: "The part ... which brought me to the edge of my seat was a preview of a genuinely XML-capable version of Word." Several Microsoft insiders have since independently confirmed that this future version is already in existence, among them Joshua Allen:

"I've been using Office 11 on my machine for about 6 months now (dog food is the best food), and I can testify that what Jon saw is really there and pretty stable, and not just 'demoware' ... The whole Office suite has married semistructured data and document authoring in a natural and unobtrusive manner, the way it's supposed to be ... Now I am just crossing my fingers that in another year, most people will see this functionality, not as a hot new feature, but as a basic level of support that they would expect of any document processing software."

This appears to be a somewhat more open vision than my earlier cynicism had feared, more in tune with the optimistic interpretation made by The .NET Guy, aka Brad Wilson: "Being able to consume something easily interchangeable like XML is a huge, huge win." As Jon Udell explained in a Byte article a year ago, the Universal Canvas provides a single, browser-based environment in which all forms of data and functionality are seamlessly available, rather than stovepiping data and functions into discrete desktop applications.

But does Microsoft — or more specifically, its Office division — have the determination to become as open as this implies? My cynicism still tells me that Microsoft's concept of the Universal Canvas has always assumed its fabric was going to be manufactured in Redmond. If it were based instead on totally open standards, then the dominant market advantage currently enjoyed by Office would be shattered. I don't see how Microsoft can be seen to allow that to happen, even if the architects of the company's server and tools products understand why it has to happen.
posted by Phil Wainewright 2:56 PM (GMT) | comments | link

Nurturing an SOA mindset

Adopting a service-oriented architecture means rethinking traditional approaches to application development:

"The service oriented environment is very similar to a game of tennis. There's a clear divide between the service provision and consumption that allows each participant to concentrate exclusively on their own concerns. And it's this separation that makes service oriented architectures different to everything we've done before ... it is a major mistake to manage both endpoints collectively, because the clear separation may be compromised ... like having tennis players that can only play with opponents they are accustomed to playing. "

The quote comes from the latest CBDi Forum newsletter, whose analysts are a team of specialists with probably more experience of researching and analysing service oriented architectures than any other group worldwide. The company is one of three analyst groups whose recent output is worth singling out as useful material to assist in reorientation for services-based architectures:

• The Laws of Evolution: A Pragmatic Analysis of the Emerging Web Services Market (April 2002) — this free download from The Stencil Group introduces service-oriented architectures and reports on web services adoption patterns


• Service-Oriented Integration - Using Web Services and XML to Integrate Systems (June 2002) — This $795 report from Zapthink explains the impact of applying a service-oriented architecture to solve application integration challenges


• Service Oriented Approach Workshop — not a report, but a set of more than 200 scripted PowerPoint presentation slides used by CBDi in its workshops outlining the scope and practice of service based architecture. Available as part of the group's Corporate (Gold) subscription, which also includes access to regular reports and a monthly journal and starts from $1500 for 25 subscribers.

posted by Phil Wainewright 3:04 AM (GMT) | comments | link

Apple satirizes .Net

In what I can only hope is a deliberate attempt at satire, Apple Computer today launched .Mac, a bundle of Internet-based services reminiscent of those once popular during the dot-com boom. Available for an annual subscription of $99.95 — and for a limited time only at a special offer price of $49.95 — the .Mac bundle comprises email, Internet-based file storage, a website builder, online scheduling, remotely managed anti-virus software and backup utilities. Slyly describing this characteristically overpriced collection of tired ideas as "innovative", Apple CEO Steve Jobs carefully avoided using the term "web services", which these days is generally recognized as referring to a service-oriented, modular architecture for distributed computing, rather than any old service delivered via the Web.

Arch-rival Microsoft two years ago staked its future on rebranding its entire infrastructure product range under the .NET moniker, in an attempt to persuade customers of its total commitment to the entirely open architecture implied by adherence to a service-oriented vision of distributed computing based on web services standards. Presumably .Mac is intended to persuade customers that Apple is equally serious about the Internet.
posted by Phil Wainewright 1:15 PM (GMT) | comments | link

Lessons in herding cats

Guaranteeing autonomy is the key to winning agreement on web services standards. That was the message that resonated for me when I read Phil Becker's account of Liberty 1.0 on Monday. At the same time, I was recalling the circumstances when I'd last investigated SAML, the specification for exchanging data on user identity and access rights that has been in the news this week alongside the Liberty Alliance announcement.

The notion of allowing users to access services from multiple providers was an aspiration of early pioneers of online application services, and in particular of a group of Internet-native application developers back in 1999 who called themselves Internet Business Service Providers. Had the term existed at the time, they would probably have called themselves Web Services Providers, because that's exactly what they were — providers of modular web-native applications delivered as component services — and much of the motivation behind forming the group was to differentiate themselves from the mainstream ASPs of the time, who for reasons best known to themselves were attempting to shoehorn top-heavy client-server applications into an Internet-hosted model.

A second motive in forming the group was to begin to agree specifications for combining services from multiple providers, as I've described in my column this week on ASPnews. Their efforts — which ended up being folded into work on SAML and a couple other standards — could almost be seen as a dress rehearsal for present-day attempts to formulate web services standards.

Revisiting those discussions was a reminder of the pitfalls of trying to create a comprehensive set of standards. Someone at the time despaired of ever reaching agreement, describing the process as "like trying to herd cats". In retrospect, that aptly described exactly what did need to be achieved. Coercion is not an option in a universe as diverse as the web itself — quite literally, the whole wide world. The only standards that will ever be accepted are those that permit the greatest autonomy among participants, which is why the concept of federation is gaining ground over other more prescriptive styles of integration. Success in achieving web services interoperability absolutely depends on perfecting the skill of herding cats.
posted by Phil Wainewright 10:57 AM (GMT) | comments | link

Novell chooses its Destiny well

Of all the announcements relating to digital identity this week, the most impressive so far has been Novell's, which augurs well for the vendor's prospects as it repositions its product family to take a leading role in web services. Of course the SAML and Liberty Alliance announcements are also important — we've already highlighted Phil Becker's excellent essay on the true significance of Liberty 1.0 in the Loosely Coupled news aggregator (see right), and I've been working on a separate article that will explore this and the SAML announcement — but specifications and standards are only any good if vendors make imaginative use of them. That's what Novell looks set to do.

Novell has given the codename Destiny to its product roadmap for directory-related products over the next eighteen months. It's a codename pregnant with significance, recognizing the reality that this roadmap is literally make or break for Novell. If it can't capitalize on its directory assets now, it never will.

The prognosis is good, though, because Novell has designed Destiny with deep understanding of the key issues in digital identity, dividing it into four areas of focus: Web services, Dynamic Identity, Intelligent Infrastructure and Federated Trust, according to this article on ASPnews. What particularly impressed me was the emphasis on building in technologies that can handle complex profile and policy information about when and how to deploy identity data. It shouldn't be a surprise to see that Novell really knows what it's talking about when it comes to directory — but it's gratifying to see that apparently it does.
posted by Phil Wainewright 3:25 PM (GMT) | comments | link