"There is something fascinating about collections of people," says Bernardo Huberman, the director of the Information Dynamics Lab at Hewlett Packard Laboratories. Huberman boasts a résumé that stretches back to the famous Xerox PARC and includes groundbreaking work in a multitude of disciplines. Of late, one of the things he and his team of researchers have been studying is what you can learn about people from their interactions on a computer network.

In 2003, Huberman, Joshua R. Tyler and Dennis M. Wilkinson published a paper titled "E-mail as Spectroscopy: Automated Discovery of Community Structure Within Organizations." In it, they reported on their analysis of a year's worth of e-mail sent back and forth between the 500 or so employees of HP Labs. They paid no attention to content; they concerned themselves only with the "to:" and "from:" sections of the e-mail header. But from that data set they were able to create an intriguing map of relationships -- clusters of HP workers who e-mailed among themselves at particularly high rates.

Many of these clusters mapped directly to explicit HP Labs work groups or departments, as one might expect. But others jumped across groups and transcended departments, revealing communities that did not exist on any official corporate map.

"Discovering that you and a great bunch of other people that are not obviously in the same organization seem to be exchanging a lot of messages and have some things in common might be very interesting," says Huberman. In the paper recounting their research, the authors note that one potential application could be to identify terrorist networks.

"The Internet is a fantastic opportunity to learn about the behavior of very very large groups of users," says Huberman. "There is, wafting around all of us, a cloud of information -- we have cellphones, we have PDAs, we have the e-mail that we read ... There is information that is implicit there ... and we have been very interested in seeing whether we could do something with it."

A growing number of companies are already applying similar research to their social networking services. If a corporation signs up with Spoke.com, Spoke will analyze corporate e-mail traffic to see who is connecting to whom and map the networks that result. So rather than have you assemble your own network, as one might do on LinkedIn or Ryze, Spoke derives your actual network from an automated analysis of your behavior.

"The real value to businesses in the current groundswell of social networking services is in the potential utilization of the conceptual underpinnings of these services to help uncover and connect to, as appropriate, the human nodes in their organizations," says Judith Meskill, a technology consultant who maintains the Social Software Weblog.

Increasingly, pundits like to tell us, we live in a "network society." Our most important relationships and communities are no longer primarily determined by family and geography. Particularly in the developed world, our atomized, alienated, transient lifestyles have resulted in our seeking community from those who share the same interests, or workplace, or some other kind of tribal loyalty.

"Networks," writes Duncan Watts, "are the signature of social identity -- the pattern of relations between individuals is a mapping of the underlying preferences and characteristics of the individuals themselves."

"One of the main questions I start asking myself is about constructions of identity," says Danah Boyd, a graduate student at the University of California at Berkeley who is a ubiquitous pundit on the topic of social software. "How do people make sense of their identity? And over and over again it kept coming back to the fact that people make sense of their identity by the people around them."

In other words, to quote one of the panelists at the Commonwealth Club: "You are who your network is." You are who you know.

If all this is true, then online social networks are God's gift to sociology. As late as the mid-'90s, notes Watts, sociologists who wanted to research social networks -- how people related to each other, who became friends with whom, how information traveled through a social network -- found their job very difficult. Information had to be gathered by hand, by passing out surveys, and the data was always suspect, because people might not answer truthfully, or even if they attempted to be truthful, might not be accurate. "A much better approach is to record what it is that people actually do, who they interact with and how they interact," writes Watts.

And that is exactly what an online social network enables. When we sign up on a social networking site, we are diving into the petri dish, and gladdening the heart of every scientist with a key to the lab. If the network can figure out what groups you are part of simply by the patterns of e-mail sent back and forth, imagine what it can learn when it knows every last bit of data you have input into a five-page profile, which might include everything from your favorite breed of dog, your sexual orientation and marital status, to your turn-ons, bedroom accessories, and tastes in music, movies and books?

But that's only the tip of the data iceberg. What if, in addition to that, it knows everything you've ever searched for on the world's most popular search engine, it has access to your blog and it has been scanning the content of your e-mails so as to better target ads to you? Researchers with access to that network -- to that online neighborhood where modern men and women spend ever greater amounts of their disposable time -- will know more about you than you do about yourself.

Read Part 2 Orkut, the satisfaction of lust, and a promising antidote to spam: Where social software is headed

Recent Stories