Applied social network analysis

Somehow Eric Promislow's Amazing Baconizer escaped my attention until Eric mentioned it to me recently. Eric was co-creator of OmniMark, an ahead-of-its-time XML-oriented programming language, and is a senior developer at ActiveState. "The Baconizer," he says, "is where I go to play in your basic LABP world (I'm too lazy to replace Berkeley DB with My SQL)." I've seen a few other applications that automate the traversal of Amazon's "Customers who bought this book also bought..." links, but Eric's does so in a goal-directed way. Here, for example, is the 12-hop path from my book to my wife's book:


You can reverse the connection, yielding a 14-hop path:


You can also walk a random path to my book:


If you do that a few times, you'll notice that all paths are roughly the same length, rarely fewer than 10 hops or more than 14. You'll also notice that the final hops to my book are almost always: Peer-to-Peer -> P2P -> Get in the Groove -> Using Groove 2.0 -> Practical Internet Groupware. This is emphatically not a good thing, and partly explains why my book is out of print: it failed to appeal to a critical mass of overlapping interest groups. Pick almost any other book, and you'll that there are at least several paths leading to it.

There are lots of these affinity browsers kicking around nowadays, for lots of different kinds of networks. Yesterday, for example, Phil Windley posted a nice write-up on Jo Walsh's FOAF (Friend of a Friend) session at ETCON, and included a pointer to a path-finding demo that connects two people by way of a sequence of linked FOAF files.

When you encounter one of these affinity browsers, it usually takes about 5 minutes to traverse the hype cycle from the peak of inflated expectations to the trough of disillusionment. This stuff is fun, but what's it really good for?

social network analysis

For book publishers, second-order analysis of these affinities could prove to be a powerful weapon in what has become a brutal war of attrition. Here's how social network analysis pioneer Valdis Krebs envisions it:

If you follow these links a few steps out, says Krebs, clusters emerge, and sometimes those clusters represent disjoint interests connected only through one book. He offers Thomas Petzinger's The New Pioneers as an example. It connected two different groups -- one reading books on business and strategy, the other reading books on complexity science and chaos theory. Now there are a number of books that broker that connection, but Petzinger's was one of the first popular books to do so, according to Krebs. [Seeing and Tuning Social Networks, O'Reilly Network]

Visible Path There isn't yet a tool that solves this problem for publishers, at least not that I've heard of. But a company called Visible Path has big plans to use social network analysis to turbocharge sales cycles. The company's founder, Antony Brydon (formerly general manager of IUMA), recently walked me through a demo. For salesfolk, it's all about access -- getting to the right people at the right levels in target organizations. The Visible Path software mines relationship data from contact databases, builds a network map by scanning email headers, and says to the salesperson: "You need to get to person X at company Y? Here are the paths that link you to X. Would you like to request an introduction via intermediaries?"

A delicate email protocol then ensues, because to safeguard privacy the system will not reveal the identity of intermediaries until they agree to participate in the referral. The salesperson can pursue multiple paths in parallel; activity is co-ordinated with's CRM system. Powerful Superconductive stuff.

If you're the kind of person who prefers not to think about how sausage gets made, you might find this all somewhat creepy -- particularly when you're approached by an automated relationship manager asking you to make an introduction. Personally, I'm fascinated to see how this will unfold.

Can it really work? Well, Antony demoed his own system to me, with real prospects, so he is clearly eating the dogfood. How Visible Path fares, during this enterprise software sales drought, will be one way to measure the validity of the concept.

Former URL: