A study of some 500,000 emails from inside Enron reveals patterns of information sharing around an office.
It may look like a modern office building, but inside it's a gossip mill. (°Florian/Flickr)
When we think about the social media that dominate online, we tend to come up with brand-name sites: Facebook, of course, followed by Twitter and LinkedIn, with maybe Google+ and Pinterest rounding out the list. But all of these, none stands up to the nameless giant that underpins our daily online interactions: email. It connects nearly everyone online: 92 percent of online adults use email, and the remaining eight percent, I think it's safe to say, probably aren't really "online" all that much.
As a social network, email allows unstructured poking ("Call me! EOM"), photo sending, and link sharing. And perhaps more than any other technology we use online, captures our actual social network; we use it to communicate with everyone from the babysitter to the boss. Surely, the billions of messages sent each day are a source of information about every aspect of human interaction, if only some researchers somewhere could get their hands on them.
Well, of course, there are good reasons why they can't. But there is at least one instance where a solid and comprehensive set of emails has become available to researchers, and they are making the most of it. The Federal Energy Regulatory Commission posted more than 500,000 emails from inside Enron in the years leading up to and including its collapse. These emails, sent by 151 people between 1997 and 2002, contain clues to how a modern (okay, modern-ish) office place functions, the way people communicate, and the paths along which information flows. A new study of these missives (pdf) from two researchers at the Georgia Institute of Technology, picks apart these emails to learn about how gossip is used and spread within a company.
Or really, I should say "gossip," because the definition of gossip the paper uses is so loose as to render it nearly meaningless. To the authors -- and, really, to their computer program that's parsing the emails -- gossip is the mention of any person who is not receiving the email. So, for example, if Alexis emailed me and said, "Hey, could you look over this piece that came in from so-and-so," well, that would be gossip under this definition, and that really doesn't make very much sense.
Nevertheless, they tease out a picture of how this gossip -- let's call it information -- flowed around Enron during those halcyon (and post-halcyon) days.
What does this show? This is a picture of how information moved across Enron's hierarchy, as indicated by the thickness of the tie. The authors of the study, Tanushree Mitra and Eric Gilbert of Georgia Tech, have divvied Enron's employees into seven levels, zero being the lowest ("employees") and six being the highest (president and CEO). Level five includes all the vice presidents and directors; level four are the in-house lawyers. In the graphic, you can see that the plurality of the information circulates among the level-zero employees (the thick gray bar connecting the two zeroes). "Employees at the lowest level play a prime role in circulating gossip throughout the hierarchy," the authors conclude. Additionally, a substantial amount of the information that flows up goes straight to the very top, and a substantial amount that flows down goes straight to the very bottom. None of the lines seems particularly mutual: For every combination of rank, there is an imbalance in who is doing the talking and who is doing the listening.
The paper also includes a breakdown of what percent of a person's email includes "gossip," or the mention of another person's name:
What's interesting here is the very top of the graph. The vast majority of people mentioned other people in about one-fifth of their emails or fewer. But at the very top, a small group of people talked about other people in almost every email, and the majority of those people are at the very top. Now, does this mean that those who ran Enron were huge gossips? No, because as I've said, this study really isn't about gossip. But it does show that being at the very top of a company changes the nature of how you relate to other people: You are privy to much more information, and the majority of your work is in dealing with people: the delegation of work, the development of strategies, and the handling of the complex relationships of a big organization. That all falls to the very top, and you can see it in the graph above.
This study doesn't give the picture of office gossip it sets out to -- those intricacies are still buried in that pile of emails, beyond the reach of software analysis for now. But eventually, it will out, one way or another, because in putting so much of our social interactions online, we are making the invisible dynamics that tie together society visible. The patterns within them aren't fully legible yet, but they're there, awaiting their discovery.
We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.
