What's exciting about cloud computing is not that it will make today's applications run faster, or cheaper. Incrementally improving what already exists is useful, no doubt. But it misses the point. What makes cloud computing exciting is how it makes possible a new class -- a new type -- of application that requires cloud computing's characteristics to exist. A type of application that would be impossible in traditional enterprise applications, designed with limited resources and user populations as constraining assumptions.

These new applications will leverage the unique characteristics of cloud environments, just as previous generations of new technology platforms in the past enabled new types of applications. For example, networked PCs made it possible to deliver applications to workgroups or departments cost-effectively. Nobody would have written a departmental time management system to run on a mainframe -- the costs of the platform would have far overshadowed any benefit the application provided. In a networked PC environment, though, an application like this one could make perfect sense.

This means that one of the main benefits we'll achieve from cloud computing is -- literally -- unexpected and unknown. Understanding -- and evaluating -- cloud computing on the basis of how well it supports and cost justifies today's applications is to vastly underestimate its true benefits. To dismiss cloud computing because one fails to understand how dramatically it will change the nature of computing is short-sighted.

This myopia is understandable. Humans cannot predict the future, of course, and it's far more comfortable to evaluate anything new in the light of what already exists. But it's a mistake. As the famous anecdote goes, Henry Ford said that if he had asked anyone what they wanted, they would have said "a faster horse." In other words, more of the same, just a little better. He created mass automobile ownership, and that transportation platform is still transforming our society and economy.

David Pogue's New York Times column of July 28, 2010 offers a glimpse of the new kind of apps cloud computing makes possible. His column is about the latest release of Dragon NaturallySpeaking, which adds a few bells and whistles to the existing version.

But how Nuance added improved speech recognition is eyeopening, and illustrates the kinds of apps that cloud computing will make common. Nuance created a free iPhone application that allowed users to have their speech translated into text. Very cool.

The ingenious thing was what Nuance did with the dictation of hundreds of thousands of users. It forwarded each translation to a centralized location, where the collection was used as a testbed for modified speech recognition algorithms.

And that's the kind of application cloud computing makes possible. Enormous amounts of data, collected from widely dispersed devices, creating a large enough analytical database to enable incremental improvements. Insight from massive data gathered from an application distributed across huge numbers of devices. These are the enterprise applications of tomorrow.

A long way from ERP, eh?