Solving the Terror Matrix

One of the real problems faced by the U.S. military in Iraq and Afghanistan and by our Home Security mavens here in the United States is how to eliminate the effectiveness of the worldwide terror network, al Qaeda.

It is not feasible to round up every individual suspected of being a member, because there are way too many of them. Moreover, we don’t have sufficient facilities to house them, and we don’t have any idea of who they are in the first place.

What, then, can we do to increase our safety and security both here and abroad?

This same question was asked in a different context by Reuven Cohen and Shlomo Havlin of the Minerva Center and Department of Physics at Bar-Ilan University in Israel, and their American colleague, Daniel ben-Avraham from the Department of Physics at Clarkson University in Potsdam, N.Y.

Although physicists, these researchers were investigating how infectious diseases spread, and how effective various immunizing schemes actually are at stopping the spread of an epidemic. They published their results in the Dec. 12 issue of Physical Review Letters.

Recall for a moment the parlor game, Degrees of Separation. The host presents two celebrity names from any genre, and two teams then look for the smallest number of connections between them. A variant of this game can be played at the University of Virginia website, Oracle of Bacon, where a computer program calculates the degrees of separation between any arbitrary actor or actress and Kevin Bacon. For example, Elvis Presley has a Bacon number of 2: Presley was in Live a Little, Love a Little (1968) with John Wheeler, and Wheeler was in Apollo 13 (1995) with Bacon.

It turns out that any person on Earth can be connected to any other with no more than six degrees of separation, and frequently as few as three. Of course, much depends upon exactly how you define a “degree of separation.” Once you define the term, however, the rest falls into place.

Cohen and his colleagues examined this problem within the context of how to stop an epidemic from spreading, or how to immunize a population is such a way so as to prevent the epidemic in the first place.

On even a cursory examination, it is obvious that short of immunizing an entire population, it would be good to immunize at least those individuals who have the most contacts with other members of the population. Finding out who these individuals are, however, is a formidable task that requires intimate knowledge about nearly every aspect of a population.

How, then, do you accomplish this in the real world?

It turns out – and Cohen and his pals have the numbers and graphs to prove this – that all you need do is select a random sample from the population and ask them who they know. Then you immunize the people on that list. You can go one step better, by asking the members of this list who they know, and then only immunizing people on the second list. But do the numbers: If each person from the first sample of 1,000 people gives you ten names, the second list will contain 10,000 names. If each of those gives you ten, you have 100,000.

If, as is more likely, the 1,000 people from the original selection give you twenty-five people each, you get 25,000 people on the first list and 625,000 on the second. Obviously immunizing 25,000 people is far easier than immunizing 625,000.

Nevertheless, this principle is demonstrably more effective at preventing an epidemic from starting, and stopping an epidemic already underway.

Cohen and colleagues point out that the same principle is active when analyzing ecological networks of predator/prey, metabolic networks, cellular protein networks, and terrorist networks.

When you are trying to prevent an epidemic from starting or trying to stop one in its tracks, you want to locate and isolate those members with the greatest number of contacts with other members of the population. With these persons neutralized by inoculation or removal, the epidemic is halted, or never gets underway.

When you are trying to halt terrorism that is driven from cells with essentially no knowledge of each other, where only an isolated number of individuals know anyone beyond the limits of any particular cell, you can apply the same procedure. Randomly select a number of known or suspected terrorists. By one means or another, determine who they know. This can be by direct questions such as: “Give me a list of the people you know.” Or by indirect questioning where the interrogator obtains a growing list of acquaintances by chatting with the individual over a period of time. Then, simply remove those individuals from circulation.

When dealing with terrorism, you can add a couple of additional steps, where you try to determine which persons on the first obtained list are known by more than one individual, and especially which persons on the second list are known by more than one individual. These “nodes,” as they are called, represent important links within the whole network which – if taken out of action – will dramatically reduce the effectiveness of the entire network.

The most significant thing to remember about this concept, however, is that you can drastically reduce the effectiveness of any terror network simply by removing those individuals identified as people they know by a random group of suspected terrorists. Even more significant is the startling fact that if you have no knowledge at all about a terrorist network, except that it is functioning in a given population, simply by selecting a representative sample of that population, and isolating the people they say they know, you will significantly impact the terror network’s ability to function.

I do not know if our intelligence people currently working to eradicate terrorism are applying this principle. Its formal application to this problem is new, although the intuitive understanding of its principles has long been known, both to our experts and to the terrorist opponents.

The significance of this concept lies in our ability to impact terrorist operations in a meaningful way with absolutely no knowledge of how the network functions. So far as I can tell, this is a brand new insight with far-reaching implications.

Applying this principle may lead us to significant improvements in the level of safety for our people everywhere.

Robert G. Williscroft is DefenseWatch Navy Editor