First, watch Independence Day to see how bad it is to centralize all computation to a central server (mothership). The aliens can infect the whole system with a virus 🙂
Seriously though, the so called “future” of pulling every computational complexity to already crowded Servers and keep the desktops (which match or exceed computational capabilities of servers) away from participating in any of these efforts is a very old paradigm. It is almost always promoted as a solution to the IT management issue. The pain of deploying clients, pains of providing updates, lack of good cross-platform client solutions etc. etc.
We are still infatuated with browser and legacy protocols like http and trying to cramp in capabilities that http never intended to handle.
lets be honest. It is nice to go around and talk about web 2.0 and all other stuff, but a lot of the so called 2.0 applications either run as installed client (Second life, Messenger), or uses downloaded applets that uses client capabilities. The computational complexity is happening in the client.
When I have a handheld with a Core duo 1.1 Ghz, I would rather run applications in my handheld instead of getting tied to a browser.
As I have written earlier, I hate a browser to be my primary workspace. I like reading wikipedia in the browser, but will never want to run my chat client from a browser window. Browser’s primary aim is to curtail users’ and developers’ power and options. This results in a poor user experience, not (as the author of the above article claims) less complex ones…
It is ok to go out of the browser. Internet is where my data is, not where my applications are.