The result of such pattern or network analysis might be to reorganize departments, projects or activities around those hubs, Huberman says.

HP Labs is now prototyping a tool called Knowledge Navigator that's based on those principles. It applies text mining, clustering algorithms and statistical analyses to employee e-mails and presentations stored on HP's servers. It could handle a query such as, "Who are the top five experts on topic x?" Huberman says, even when such expertise is not explicitly noted in org charts or personnel records.

Huberman says this kind of knowledge harvesting will be used by companies internally on their employees and externally on customers, resulting in the ability to generate messages and pitches aimed at both groups. "What we will see in the next few years is a very targeted way of placing information in the hands of relevant people," Huberman says. "Sure, it can be annoying, but it's better than getting spam on things you don't care about."

Despite the benefits, he acknowledges that mining messages raises ethical and potential legal issues. "In the next few years, we will see a blurring of the boundaries between what is considered private and public," Huberman says.

Mining employee e-mails is "something the company has an interest in, and we are starting to see that interest grow," says Carl Jones, director of collaboration services at The Boeing Co. in Chicago. He says the company has a knowledge management pilot project that, among other things, examines e-mail messages.

"If you have a business problem, you may be able to mine across the e-mail spectrum and find out, hey, there are people out in the field who are subject- matter experts that can help you," says Jones. But, he adds, "we'll have to be very careful about policies on privacy and so on."

Jon Kleinberg, a professor of computer science at Cornell University , says much can be learned from the networks created by people's activities on the Internet.

"How can you infer that someone is influential?" he says. "Is it the obvious things, like they send and receive the most messages, or is it more subtle things, like they operate at the periphery [of a group] but pull together groups that are otherwise weakly connected?"

Kleinberg says answers to such questions may have profound importance for companies that sell online and rely on word-of-mouth recommendations via customers' e-mail. He's looking into two competing theories as to why that kind of e-mail sometimes leads to snowballing sales and other times fizzles.

"Is it the attractiveness of the product, or is it something about the community of people who are into those kinds of products?" Kleinberg wonders.

He says e-mail pattern analysis could help a company answer questions such as, "Who are the key people to influence?" and "For which products is it worth it, and for which is it not?"

"Social network analysis is one of the great tools for productivity going forward, and very few people understand it," says Thornton A. May, a Computerworld columnist and dean at the IT Leadership Academy at Florida Community College at Jacksonville. "People tend to think of social network analysis as a list of people -- an address book. But it should tell you not just who knows who, but who knows what as well."

Users should see social network analysis as more than a way to find dates or customers. It can "solve problems, create teams or recombine organizations," May says.