Also, I think it’s important to define what we mean by “Software as a Service.” Is SaaS an installation/update methodology, or has the term become a generalized mandate for running every client application inside the browser? This is an important question, because browsers simply aren’t up to running complex apps. Even Ajax technology is raw and bad — just try to do something GUI-ish, and see how ugly it gets. Building a large app with Ajax is not feasible at all.

So, for fill-in-the-form RFx and simple database apps like contract management, browser-based thin clients work OK, and I’m fine with slapping the sloppy “they ought to be SaaS” label on them.

But for heavy-duty analysis, or heavily interactive applications, browser-based clients don’t work well at all (when’s the last time you built a spreadsheet model over the Web?). One solution to this is the so-called “stateless thick client.” For an interesting discussion, see Tolia, Andersen and Satyanarayanan, “Quantifying Interactive User Experience on Thin Clients”, IEEE Computer March 2006.

The good news is that thick clients can be SaaS, too; in the precise meaning of SaaS, that is — i.e. automatic update and invisible temporary installation. But there’s absolutely no reason to mandate that everything run in the browser. That would be a terribly limiting outcome, given the poor state of current browser technology.