Sunday, July 10, 2005

Network Theory with Emphasis on Al Qaeda

A relatively new field that has permeated its way into numerous other scientific, economic, and social science fields is network theory. As researchers have been focusing on complexity in our world, and how seemingly random connections between different entities can self-organize into something with structure, within the last decade or so some breakthroughs have been made that have helped in our understanding of such strange bedfellows as the Internet, World Wide Web, food webs, the spread of disease through social networks, how businesses live and die in the marketplace, and biochemical networks within our cells. Such completely different systems are linked together through the same mathematical organization, which is the basis of network theory.

A great introduction into the field is Albert-Laszlo Barabasi’s book “Linked.” Barabasi and his team discovered that the systems mentioned above (and more) are so-called scale-free networks. The idea is built around the concept of a hub, which is a node or component of a system that is linked to other nodes of the system. The World Wide Web, for instance, has as its nodes websites. There are, of course, billions of websites currently. But the vast majority of individual websites may only have a couple other websites that link to it. Only a very small number have large numbers of other websites that link to it. Some examples of popular websites are Yahoo, Google, CNN, and so on. These popular sites are hubs. Hubs essentially can be thought of as holding large portions of the network together. Problems can conceivably arise if a hub fails. The Internet is a similar sort of network with servers, personal computers, and routers as its hubs. A good summary article is in the May, 2003 Scientific American.

One can understand from this model how computer viruses are so effective and can spread so quickly around the world. If a single, typical node (perhaps my computer) is ‘infected’ by a virus sent out (normally via email) to the Internet, and if that node were to somehow contain it and not send an email out to anyone else, nothing would happen to the rest of the network. But if I were to send emails to a number of friends and acquaintances, or if the virus were sent to numerous nodes in the network, as each node sent out emails the virus spreads to each of those nodes, and before long it is possible to have a cascading effect. Now, if a hub is infected, and it is linked to many, many other nodes of the Internet network, large numbers of nodes of the Internet may be infected almost simultaneously, perhaps before an ‘antidote’ piece of software can be made available. On the other hand, let’s suppose hubs are ‘cured’ of a virus relatively quickly, but some of the small, peripheral nodes are infected. The virus is not extinguished just because hubs are cured. In fact, computer viruses can persist for long periods of time (6 months to as long as 14 months after antidote software is available) even without infected hubs because they can still move around, albeit slowly, between the poorly connected nodes of the network.

One interesting aspect of this is to think of a social network such as Al Qaeda. This organization follows scale-free network structures, rather than some other types of social network structures. Other types may be a hub-and-spoke structure like a dictatorship, where a central hub runs the entire network. A tree structure is also possible, with a set chain of command (much like a typical military organization, where more minor decisions can be made locally in parts of the network, and major, global decisions made by someone like the President or a top general, and this decision cascades down to lower parts of the network simultaneously. Al Qaeda is neither of these, but rather a ‘web without a spider.’ An analysis was done after the 9/11 attacks to determine the structure of the cell responsible for the attacks (this is outlined in Barabasi’s book). Mohamed Atta was indeed a hub of the suspected 34-person cell. But, he only had links with 16 other nodes. Marwan Al-Shehhi was linked with only 14 other members of the cell. Most of the other members only may have had one or two known links to other nodes, which is a signature structure for a scale-free network. If we assume the entire Al Qaeda organization is similarly structured, what we are fighting is indeed a network that is very flexible and can tolerate some number of internal failures. If Atta was captured prior to 9/11, that event would not have necessarily crippled the cell, and the attacks may have very well gone as planned. The reason is that the other nodes still had the links to hold together the cell, so even taking out the leader of the cell does not disconnect all the other links. Killing Bin Laden will not in itself cripple Al Qaeda for the same reason. In fact we have seen this because a number of Bin Laden’s very well-connected lieutenants (i.e. other hubs) have been killed or captured, and yet the attacks and killing by the periphery of Al Qaeda continue unabated. This is completely analogous to viruses in the Internet, where if a hub is ‘cured’ (i.e. Bin Laden removed) the virus may still be able to do damage through poorly connected nodes (small cells on the periphery of Al Qaeda’s network, acting almost independently). This is part of the power of a scale-free network structure.

If Al Qaeda had more of a military structure, taking out Bin Laden and some key lieutenants would likely cripple and possibly defeat the organization, much like taking out key generals in a war can have devastating consequences for the outcome of battles and perhaps the war itself. Unfortunately, it is not so simple with the terrorist organizations we are facing.

Thanks to Zenpundit for asking me to write up something on this. I am not an expert, but hopefully this helps in some small way for understanding bare basics of a network like Al Qaeda.

9 comments:

Dan tdaxp said...

What is the best way to deform al Qaeda into a network that is easier to defeat?

-Dan tdaxp

Dan tdaxp said...

James,

Some comments.

The move away from dumb soldiers and towards super-empowered low-level leaders sounds like the strategic corporal or sheik system.

However, when you say

Vietnam showed how expensive and ineffective it was to try to take down an opposition whose organization behaved like a scale-free network

Expensive, yes. Ineffective, no.

The scale-free protion of the communist fighting force -- the Viet Cong -- was eliminated as a threat shortly after Tet. Its bitter, alienated, and weak remnants were finally mopped up by NVA regulars after Saigon fell. While Vietnam was an important 4GW, it was not a 4GWarrior, but an NVA regular, that took Saigon's Presidential Palace.

-Dan tdaxp

Mark Vondracek said...

james,

One of the conclusions of Barabasi is that scale-free networks are in many ways a natural structural conclusion when a complex system self-organizes. Scale-free networks show two basic properties which are growth (networks grow one link at a time) and preferential attachment (new nodes are more likely to attach to existing nodes that already have many links). One can think of this as a natural 'the rich get richer' phenomenon.

Certainly this all gets extremely complicated in something like a military structure. It is dominated not by a scale-free structure, but a tree-like, top-down structure with a chain of command. The exceptions are found with the elite groups, as you suggest, who are actually trained to be independent. This does make them dangerous for the enemy because even if the command structure goes down, the Delta-type forces are still extremely capable.


dan -

Thanks for your comments. A major difference between al Qaeda and the Viet Cong is the VC were localized territorally. Al Qaeda is spread globally more or less, and that is a severe problem when it comes to defeating them. There is no easy way to do this. In my opinion the best and quickest way (and most obvious) to do the most damage to the network is to eliminate the major hubs. This will destroy the most links and have the best chance to create internal chaos/disorganization as anything. Of course, this does not eliminate the network because of the peripheral links that go around the hubs and do not depend on the hubs, just like the Internet structure. Destroying hubs is also the quickest way to cut off the (main) links to money and resources needed to carry out attacks, as well as slow growth of the network because of preferential attachment. Unfortunately the 9/11 attacks also provide evidence that it does not take huge sums of money to enact devastatng results, but every little bit we can remove helps.

Even if the hubs are destroyed the nature of scale-free networks is to grow. As Barabasi points out, al Qaeda and other terror networks will not be fully defeated unless the driving forces for forming new links and adding new nodes are eliminated. This goes to the core ideas of what drives one to hate the West and the U.S. in particular. Military forces in the Mideast, support for Israel, occupation and poltical influence in the region, 'Crusade' against Islam, and so on, but these are not going to go away. There will always be some small number of people who will fall into the 'extremist' mentality and follow a Bin Laden type figure, so if al Qaeda falls apart some day, another organization will almost certainly try to replace it.

Dan tdaxp said...

As Barabasi points out, al Qaeda and other terror networks will not be fully defeated unless the driving forces for forming new links and adding new nodes are eliminated.

...

There will always be some small number of people who will fall into the 'extremist' mentality and follow a Bin Laden type figure, so if al Qaeda falls apart some day, another organization will almost certainly try to replace it.


I guess what I'm asking is:

How can encourage al Qaeda's evolution into a less scale-free network?

If I may self-promote for a second...

On my blog, I currently have a 2.5 post series about early Christianity

4GW Tactic: Love Your Enemy...
Caiaphas and Diocletian Did Know Better
... Netfaith Subversion and Deformation

It is leading up (in my mind, at least! :) ) to discussing Christianity's deformation into a non-4G non-netfaith non-scale-free movement during the Dark Ages in significant parts of the Europe.

So how can we similarly deform al Qaeda into a non-netwar organization?

Dan tdaxp said...

actically speaking, the most effective way to kill the snake is to cut off the head, but as you've pointed out, these organizations have no heads. Another approach, however, is the intelligence approach: infiltrate and obliterate from within.

Third approach: change the environment so they no longer wish to be scale-free.

Dan tdaxp

Mark Vondracek said...

Thank you for all the interesting comments and observations.

there is hope -
I agree with you and james that the fact individuals have been trained to some degree makes it difficult to defeat the network. Another problem is the sleeper cell design, as the Brits apparently just found out.

dan -
Because scale-free networks are technically built around a hub concept, destroying hubs in al Qaeda (bin Laden, et al) may in fact make it a more random network than scale-free. While the periphery can still cause problems, internally there would be a destruction of large-scale attacks that need to be coordinated, funded, and so on. Without many of the smaller nodes having links to other smaller nodes, ideally smaller attacks would continue, but large, 9/11 style attacks certainly become less likely as resources are diminished for the remaining members of the network.

Would other al Qaedas have a chance to develop and replace an ineffective original? I don't think we would see anything on the same global scale as the original, for that took many years to develop and the US and others would not allow training camps to be built again. I suppose it is inevitable that terror groups and suicide bombers will turn up as long as we have military units on Islamic land and support Israel, but a twofold attack of (1) eliminating al Qaeda hubs and (2) preventing new networks (certainly no true national military will be able to stand up to the US) from developing will save many lives and lead to a more stable Mideast. If we eliminate the hubs in al Qaeda, we also have time as an ally in the sense that it takes time for new hubs to develop, since they would need to make numerous contacts over some presumably longer period of time.

Dan tdaxp said...

Very good answer. Thank you. I must think about it.

-Dan tdaxp

Unknown said...

So many blogs and only 10 numbers to rate them. I'll have to give you a 8 because you have good content.

Free Access To More Information Aboutpricing

Dan Nexon said...

"A relatively new field that has permeated its way into numerous other scientific, economic, and social science fields is network theory"

By "relatively new" I hope you mean something on the order of decades old. Sociologists have been doing network theory for quite some time. Indeed, many of the "breakthroughs" championed by physicists and other natural scientists who have applied network approaches to social relations have been around for quite some time. There was a good series of posts about this on Crooked Timber a while back.

http://crookedtimber.org/2005/05/19/isolated-social-networkers/

http://crookedtimber.org/2005/05/25/virtual-seminar/

http://crookedtimber.org/2005/05/19/reinventing-the-wheel/

Oh, and you need to implement higher-level spam protection :-)