Social Network Analysis
Eighty to a hundred people attended the inaugural meeting of the Institute for Social Network Analysis of the Economy (ISNAE: is-nay) yesterday evening at PARC. The crowd included social network software people, venture capitalists, sociologists, deal-makers, economists, and a handful of people looking for advice on how to leverage their own personal networks.
I’ll share some of my notes and use this as a social network analysis resource page for a while.
Mark Granovetter, a professor of sociology at Stanford, did a great job of covering the waterfront of Social Network Analysis (henceforth SNA).
We’re all involved in social networking every day. It’s like the character in Moliere’s Bourgeois Gentilhomme who is astounded to discover that he has been speaking in prose all his life.
The typical person knows 500 people. That gives 125,000 possible configurations of relationships. SNA represents the relationships as points and lines. Software to sort through that many lines has only been around for the last five years, so deep SNA is relatively new.
Mark’s working on two major projects. The first is an SNA of the early electricity industry. Conclusion: A tight social network of Edison and others shifted electricity generation from a distributed model with local generators to a grid fed by massive turbines. The entrepreneurs drove the technology rather than the best technology rising to the top on its own.
Is this happening in SIlicon Valley, Mark’s second area of study? Other areas and nations send people to Stanford and to the Valley to study the ecosystem here. They are trying to find the formula that makes it happen. Mark and his students are years into the project but the completion date is always moving out faster than the project itself.
Mark introduced some of the classic concepts of SNA:
The strength of weak ties, his own serendipitous discovery. Looking for a new job? You’re more likely to get useful leads from acquaintances than from close friends. The acquaintances travel in different circles. For the most part, you already know most of the people your friends know.
The small world problem, famously studied by Stanley Milgram. Short-hand: six degrees, Kevin Bacon.
Scale-free networks, a confusingly named situation where mosts nodes only have a single connection but a few hubs have loads of connections. This is robust if you’re killing nodes because the odds are that you’ll hit one with few connections and the network won’t feel it. If, on the other hands, you understand where things are, the scale-free network is fragile; wiping out a hub does major damage.
The three authors popularizing SNA at the moment have different views on the impact of scale-free networks (think of it as a measure of density or clustering) on network resilience. The books Mark suggests are:
Six Degrees: The Science of a Connected Age by Duncan J. Watts
Linked: The New Science of Networks by Albert-László Barabási
Nexus: Small Worlds and the Groundbreaking Science of Networks by Mark Buchanan
I found Linked much more enjoyable, and useful, than Six Degrees. have yet to read Nexus. Mark did not suggest the two that follow, perhaps because their authors lack PhD’s and tenure. Gladwell is an enlightening read; you can do it on a transcontinental flight with time to spare. Johnson’s book covers lots of things besides networks, and for me it was read/contemplate/read/contemplate/etc., one of those books you savor over weeks.
The Tipping Point: How Little Things Can Make a Big Difference by Malcolm Gladwell
Emergence: The Connected Lives of Ants, Brains, Cities, and Software by Steven Johnson
I asked Mark about the capacity of the current software. Points and lines don’t convey much information. I ponder the importance of the value and quantity of content flowing through the pipe and what’s going on in the nodes. Is this someone I respect? Or hate? Or don’t know well?
Mark said the software today is ahead of our ability to use it. I take it many SNA people are still relying on dots and lines. Apparently, the most commonly used package can take as many variables as you can throw at it. It’s named Pajek. That’s Slovenian for “spider.” It’s a free download. It produces dots and lines diagrams like this. Unfortunately, the documentation is a broken translation from the original Slovenian, so it’s difficult to learn. ISNAE will be sponsoring a class in Pajek (pronounced Pie-Ack) soon. Other software for SNA is here.
The solid state physicist sitting next to me was looking for direct application of SNA. Sick of designing chips, he’s looking for networking to take him into a new field. Upon hearing of the strength of weak ties, he deadpanned, “Does this mean I should shit-can all my friends and begin hanging out with mere acquaintances?” Mark said friends have their place.
I had a small-world experience yesterday myself. Over lunch in Emeryville, a friend had mentioned Bob Karr, a guy who is investigating the interlocking network of board memebers in Silicon Valley. The next question in the evening session came from the fellow behind me, who turned out to be the same Bob Karr. You can look at the Silicon Valley network for free at www.linksv.com.
For those who are interested in SNA, Mark recommends joining the International Network for Social Network Analysis. dues are $40. Annual conferenes, the Sunbelt Conferences, have taken place on the Costa del Sol, San Diego, New Orleans, Cancun, Charleston, and Budapest. Lucky you — the 2004 Sunbelt convenes mid-May in Portorož, Slovenia. Maybe you can pick up some hints on using Pajek while you’re there.
The INSNA site has a wealth of resources: back issues of Connections, the Journal of Social Structure, a listserv, social capital links, and lots more.
Posted by Jay Cross at September 11, 2003 07:49 PM | TrackBack
