How The NSA Uses Social Network Analysis To Map Terrorist Networks
Ever since The Guardian reported that the National Security Agency (NSA) has been collecting the phone record metadata of millions of Americans, the cable talk circuit has been ablaze with pundits demanding answers to what should be obvious questions.
Who knew about the program to collect data? (Apparently, all three branches of government). Who else has been supplying data? (Just about everybody, according to the Washington Post). What is metadata? (It’s data about data).
The question that nobody seems to be asking is probably the most important one: What is the NSA doing with the data and why do they need so much of it? The answer is a relatively new field called social network analysis and, while it may make people uneasy, the benefits far outweigh the risks, so it is probably something we will just have to accept.
The New Science of Networks
The story of networks starts in 1736, long before the United States became a country, when Leonhard Euler set out to conquer a famous math problem concerning the Seven Bridges of Königsberg. To solve it, he created a new form of mathematics called graph theory, which concerned itself with links and nodes in a network.
In the 1950’s, interest renewed in Euler’s networks. First, Anatol Rapoport introduced the concept of triadic closure, which asserted that networks grow when people meet through a central friend that they both know. Later, Erdős and Rényi showed that as networks got bigger, communication among the people in the network got much more efficient.
In the 1970’s a sociologist named Mark Granovetter argued that we get most of our information not through close friends, but through weak ties and in the 1990’s Watts and Strogatz built on Granovetter’s work by showing that small clusters of people naturally organize themselves into far flung networks.
So by the late 1990’s, the small field of network analysis had built into a full fledged science and it was about to be applied to an increasingly important problem: Terrorist networks.
Mapping Terrorist Networks
Valdis Krebs of Orgnet is a network scientist who in 2002 published a widely praised paper on mapping terrorist networks and has since consulted with the Defense Department on methods and approaches of evaluating and dismantling terrorist organizations.
While he isn’t privy to classified information, he describes on his website how an entire network can be mapped using commercially available software by identifying two initial suspects:
It used to be that law enforcement officers would simply watch the two men closely, but in the era of global jihad, that’s much too slow to save lives. The two might be peripheral to the conspiracy and it could take years before you could connect them to the leadership of the network, if ever.
Here’s where the data from Verizon and other companies comes in. If you can analyze communication records, you can move much more quickly. However, you don’t want to look at everyone the suspects talk to because you’ll end up with mostly incidental contacts, like friendly neighbors and delivery men.
But if you kept Rapoport’s concept of triadic closure in mind and had full access to communication records, you could look for contacts the two suspects have in common and start to build out a map of the conspiracy.
The next step would be to analyze the contacts of the suspects’ connections, again looking for closed triads within the existing network. As you progress from link to link, a fuller picture begins to form (click to enlarge).
Once you have the network mapped, you can begin to mathematically analyze it, which is how important insights can be gleaned even before wiretapping and surveillance warrants have been issued.
You can, for example, assess who is well integrated into the network by calculating who is most central; who has the widest reach by counting how many people in the network are within two connections from them and who in the network provides a crucial role as a bridge between otherwise unconnected people (as Mohamed Atta, the uppermost green node, does in the 9-11 network above).
The result is an almost uncannily accurate picture of the leadership, who can then be targeted to dismantle the network. (It has been estimated that the 9-11 network could have been broken up if just three central nodes had been taken out).
It should be clear by now why the government regards access to communication records as so crucial to national security. If the system had been in place in 2001, there is a high probability that the 9-11 network would have been broken up, saving thousands of lives and trillions of dollars.
It would be impractical, to say the least, to get court orders for each and every connection a suspect has, most of whom would not even be investigated. Without a full data set, the social network analysis could not be done and more intrusive, but less efficient methods, would need to be employed.
So whatever you might think of the program, it is most probably here to stay. What is perhaps of greater concern is that this type of analysis is not unique to antiterrorism, but is increasingly becoming a basic fact of commercial life.
Beyond the cell phone companies, social networks like Facebook, Google+ and Twitter can analyze the communications of hundreds of millions of people. Retail giants like Amazon, Walmart and Target are sifting through our purchases in order to predict our future behavior.
Wherever we go, our movements, faces and actions are being analyzed and, more often than not, it is not the government.
The truth is that there is a dark side to technology and our privacy is being breached every day by someone, somewhere. That’s just a fact of modern day life. It seems to me that if we’re willing to accept it from marketers who are trying to to sell us goods and services, we should be able to tolerate it from those who are trying to protect us.