In a Slashdot discussion of vendors’ service-model initiatives, Alwin Henseler has observed that the mainstream end-user community has already adopted software as a service in everything but name:

And nowadays, >90% of desktop users run a closed source OS on their desktop, that automatically downloads and installs updates with unknown contents, whenever the user goes online. And extend it by clicking ‘download plugin’ whenever something appears to be missing or not working. And keep their mail online on their ISP’s servers. And share their family pics online using a photo sharing site that popped up 2 months ago. That is in practice different from software-as-a-service, ehm… how?
…
Why software-as-a-service is not the norm yet? Bandwidth limitations? Because no company did a solid execution of the idea so far? Copyright issues with 3rd party software? Because people are used to buying install CD’s or computers with preloaded OS? As opposed to a bare minimal software install, and downloading the rest after hooking up the broadband connection? Hey wait, aren’t folks already doing that anyway, sort of?

When you think about it, pervasive broadband in both enterprise and consumer settings has enabled a seamless transition to something that in practice is software as a service. Users already depend on having a fast, high-uptime connection to do most of the things that people actually find interesting about IT today; almost as an afterthought, they trust a service provider (even if it doesn’t use that label for itself) to maintain their code base (on local hardware that the user kindly provides).

The major drawback, therefore, of current mainstream practice is that it retains old-fashioned downsides:

• cumbersome and fragile state on the client device
• too many ways for the user to be burdened by redundant complexity
• the nuisance and hazard of having data stored on a user-managed device at the edge of the network instead of in a professionally maintained piece of the cloud

Users are already making the tradeoff, even if by default, that service-model delivery and maintenance of the software stack is the better way to go. Finishing the process of on-demand adoption merely means enjoying the rest of the upsides: that is, moving more of the burdens of non-productive effort off each individual user’s to-do list.

The momentum is in the on-demand direction.