Directed Scale-Free Networks
The real insight in Mitch's paper comes from the realization that terrorist networks aren't merely generic scale-free networks, but more likely an important subset: directional scale-free networks. Scale-free networks are typically depicted by a set of nodes that are symmetrically connected (I link to you, you link to me). The dynamic flows that travel through those networks, whether they be information/fluids/electricity/contagion, can travel in both directions across symmetric links. However, that doesn't actually happen in many real world networks. In these networks, links have direction (I link to a major hub, and it doesn't link back to me). Directional networks, in contrast, have links that are asymmetric and offer only unidirectional flows. A good visualization of this can be seen in Albert-Laszlo Barabasi's diagram (inset, from "Linked"). It depicts the directional flow of connections (from the left to the right):
- a central core of highly connected (via bidirectional links) nodes,
- an IN continent (links in), an OUT continent (links out),
- and various other structures (a Tube of connections between IN/OUT continents, Tendrils that feed into each continent, and islands that are clusters of affiliated but unconnected nodes).
AlQaeda.netWhen you apply the directional scale-free network model to al Qaeda, you see a fairly good fit, particularly when you assume that the al Qaeda of today is more of a movement than a cohesive organization (Mitch provides some historical analysis to back this up). Here's how it works. Al Qaeda's flow starts in connections from the feeder networks within the IN continent that instill a common animating narrative (Madrasahs, etc.). This common narrative drives social clusters to seek connections with the central core (bin Laden and associates) which will eventually transition them to become operational assets (terrorist cells) in the OUT continent. Here's the likely path of al Qaeda's evolution, some of which has already been seen, given this 'movement' model:
- al Qaeda's leadership will increasingly ask groups to act on their own, without seeking direct connections to the central leadership. This will be accomplished through the production of global media messages that contain targeting recommendation (which is essentially a low bandwidth command link). If this works, recruits within the IN continent can transition (FLOW) quickly to the OUT continent without ever directly connecting to the central core.
- IF this transition can be made, al Qaeda's central leadership would become relatively immune to disruption. Nearly all of the central core could be knocked out without damaging its ability to message those groups in the operational OUT continent. In the words of Valdis Krebs, al Qaeda could look very much like a doughnut and still be able to operate.
- Finally, local groups that enjoy a level of operational success within the OUT continent can and will go international autonomously, in that they will create/distribute media messaging and operationally manage attacks on a global scale. Zarqawi's efforts and the recent plea by al Qaeda in Saudi Arabia for attacks on global oil infrastructure are good examples of this. Within the network model, these groups would be seen as clusters on the periphery that can catalyze the operation of the entire network (by acting not just as feeders and operators, but as mirrors of the central core).