With the avalanche of data in almost every enterprise, distractions abound. There's so much potential learning to be done, and so much of it fascinating, that it can be easy to lose sight of the business--or even to find one.

That thought came back to me as I read a NY Times blog post on the data analytics possibilities for the Amazon Kindle. Think of everything Amazon can discover: Which books are most often left unfinished, and at what point do readers give up? Which books keep people up late at night? Which have the most re-read passages? In which ones do readers most often skip to the last chapter?

I love this kind of data. But sometimes it's hard to build a business case around it. One of the companies I've been following the last couple of years, Sense Networks, has had to struggle with this. They can uncover fascinating patterns of cell-phone users. They found recently, for example, that people who have pre-paid cell phones and seem to be in the market for used cars are eight times as likely to switch cell-phone carriers as the general population.

Interesting, but not necessarily a financial windfall. In fact, after Intel's venture arm put $6 million into Sense last year, they moved out the founder and CEO, Greg Skibisky, and looked to replace him with someone more familiar with the business needs of phone companies.

This makes certain sense. But the fact is that the unusual studies and correlations might be more likely to lead to surprises--and breakthroughs. That's why I find myself looking more to academics and large research outfits to come up with startling advances. They're not facing the same life-or-death financial pressure as venture-backed start-ups. But is the lack of pressure itself a problem?

I'm thinking about that as I pack my bag for Seattle. As I research my book on Tuesday, I'll be meeting with people at Vulcan Inc. in Seattle to discuss Project Halo, an artificial intelligence lab financed by Microsoft co-founder Paul Allen. They're teaching computers to understand spoken and written language, and to pass academic tests prepared for humans. This kind of research could transform search, and even education. But the business case, at least in the near term, is a foggier. After 23 years at BusinessWeek, where we were told not to stray far from the profit and loss picture, it's a welcome change.