Neo-Network Computing & The Service-Oriented Client

Google’s Eric Schmidt made the perfect analogy for webware and web-based applications yesterday at O’Reilly’s Web 2.0 conference:

“Fundamentally, it’s better to keep your money in the bank than in your pocket.”

Schmidt was commenting on the notion that architectural recent developments had now enabled the vision of Ellison’s Network Computer to finally be realised; his contribution to this, both at Sun and Google is of course to be noted. However, banks are very available – we don’t have to worry about disappearing and re-appearing roads or ATMs!

Though we’ve seen innovative leaps forward in web-based applications, clients assume persistent connectivity to a server and do little to accomodate sporadic or lost connectivity in a consistent manner. Despite the rise of AJAX, web apps are still very much architected around requesting pages rather than requesting remote services. Earlier this year, I heard Adobe’s Kevin Lynch speak of the need for service-oriented clients that can utilise native client services much more intelligently and include consistent support for remote services. This can help usher in richer user experiences as well as applications that can continue to function, or degrade elegantly when connectivity is sporadic or absent.

Ironically, Microsoft is perhaps leading the way for richer service-oriented clients, such as Live Writer and Live Mail Desktop. I’d like to see Google embrace this paradigm more fully, but there’s a potential secondary industry here for software developers to harness open APIs and create service-oriented clients for leading applications such as Gmail and others.

Personally, I’d love to have a Dreamweaver+Typepad service-oriented client to manage my blogging :)