Saturday, April 29, 2006

Analyzing the War on Terror in the Context of Network Theory

In a post dating back to last July, there was a good discussion about how we need to think about various organizations such as al Qaeda in terms of a network structure. Attacking networks effectively will involve understanding the topological structure in order to develop strategies to dismantle the network. I want to expand on this with some new thoughts and after thinking about what and how we should learn from the mistakes that have been made during the war on terror.

In a post from John Robb at Global Guerillas, he refers to a paper that investigates the effect of noise on, ultimately, the decision making/consensus building process in small world networks. This is all related to how information spreads throughout a network, and how rule sets evolve in order for global consensus can be reached when individual agents within the network only have access to local information. Perhaps the best known rule set is the majority rule, which says that an agent will conform to the state of the majority of its nearest neighbors. What research has shown, though, is that in a perfect, noise-free environment, the majority rule will fail to reach global consensus, even if there is some amount of longer-range crosslinks within the network (meaning that some small number of individual agents not only see some number of nearest neighbors, but an occasional link to another agent outside the local boundary defined in the initial conditions). However, the boundaries between local pockets of differing viewpoints does breakdown when noise is introduced into the environment. This is, of course, a better simulation of what the real world is like anyhow. Noise, in the context of simulation work, means that there can be miscommunication between neighbors. This allows for incorrect information to be passed along and will influence agents to switch their state.

As always, theory is one thing and a real-world environments is quite another, particularly when dealing with social networks. If we think of human characteristics such as emotion, peer pressure, greed, deceit, psychological abnormalities, and so on, it becomes clear there is a great deal of 'noise' in actual environments that will need to be considered when an individual makes a decision. Here are some issues to consider for a network such as al Qaeda, as well as some lessons we should make sure to learn from from this time forward.


Lesson 1: Network formation takes time. Time can be an enemy or an ally, depending on circumstances.

I think just about anyone would agree, at least in an intuitive sense, that network growth is effective and efficient when environmental noise is low. It takes time to form real networks, especially social networks, and the world essentially allowed al Qaeda to grow unabated (i.e. the world made no noise) even though everyone new they posed a threat and had actually carried out terrorist attacks. Al Qaeda did not form overnight, but instead took years to form a global network structure. We need to learn from this, and as the current network is dismantled, we will, from now on, need to search for evidence of other networks forming or peripheral branches of al Qaeda trying to expand, and do something about it quickly. This will require police work, and not massive military action, if we want to control the spread.

Time is an enemy in noiseless environments that allow networks to grow. Time is an ally if you catch network formation early on, knowing that it takes time to form more extensive link topologies. Creating a 'noisy' environment that disrupts the initial network formation (such as through surveillance; apprehending leaders, who are hubs in the network, early on; disruption of communications between known agents; making financial and hardware accumulation difficult, etc.) will cause disruptions in new links being added to the network. This would also have the advantage of allowing us to work locally rather than globally, which is much more difficult, as we now know. Having said that, the reality is that we allowed this massive terror network to form. Then came 9/11/01.

One must remember that in large populations, typically the vast majority of people want to live their lives and are nonviolent. In the Mideast, there is a relatively small percentage of the population that would fall into the 'extremist' category, or who would actively assist extremist groups unless forced to do so. After 9/11, we had about as much world support as is possible in this day and age to go after al Qaeda and their main supporters, the Taliban. Fortunately, the bulk of those groups were in Afghanistan. This was all to our advantage, since they were also in the mountainous regions with low civilian populations. There was a low risk of any significant collateral damage or casualities. We had a great opportunity to strike at multiple hubs in the network (which would geatly weaken the network) and avoid making too noisy of an environment that would break down the boundaries between the two decisions that are possible in this scenario (tryng to connect to the simulation research mentioned earlier): either you decide to actively support al Qaeda in some way, or you don't support them in any way.

Our first mistake was to allow Afghan forces and, worse, warlords, take the lead when al Qaeda and Taliban members were on the run. It should have been large groups of special forces going in for the kill. I don't think anyone would have objected to this right after the attacks on the U.S. But then we took the pressure off to a large degree and put the focus on Iraq. What is especially disturbing is that we did this when we had weapons inspectors on the ground. Sites that were suspected of housing WMDs were given to the inspectors, including some from the CIA, and all were turning up clean. I personally refuse to believe the President and the war planners did not know our intelligence was faulty at best, because they had evidence on the ground (from the inspectors) that it was faulty, if not simply bad intelligence. Regardless, the war started.

This is where numerous mistakes were made that seem to fit into our understanding of network structure and theory. War is the ultimate source for 'environmental noise.' As studies suggest for common small world networks, noise is what is needed to break down boundaries between pockets of network agents and allow for widespread, if not global, consensus within the network. There were two main decisions individuals in Iraq could make: welcome the U.S. or resist the U.S. This is greatly simplified, of course, and one would hope that the majority rule would sway towards freedom and welcoming the U.S. This is certainly what the Vice President and Sec. of Defense tried to sell the American public on. However, what we must learn from Iraq is that the absolute necessity for a society to reach the peaceful decision and consensus is security.


Lesson 2: In social networks, law and order and security reduce environmental noise. If you do not maintain low noise levels, the local boundaries between those in the network who agree with you and those who disagree with you break down.

Iraq is a large country in terms of area, and it obviously required many more forces than we had to maintain order and security over the entire country. We did not provide this security. Very quickly, tremendous amounts of 'noise' were introduced into the social network of Iraq. Little or no food, clean water, power, money, jobs, medical facilities, trash collection, and large amounts of lawlessness appeared and remained for great periods of time that weren't there earlier. What almost certainly started off as a majority of individual agents in the network that would welcome the U.S. was now put into an environment where the boundaries between the majority and minority views, which hold steady with no noise, suddenly are perturbed from noise. The boundaries break down, and suddenly more in the majority change their view. As the research suggests (see figure 2 in the paper), a near consensus has been reached that no longer welcomes the U.S. in Iraq. It has simply been too 'noisy' for too long. In terms of time, another factor that plays into all this is the widespread exposure and use of communications technology. Individuals in the network are more connected within the network with more crosslinks, as opposed to just knowing nearest neighbors in the network. A combination of noise and 'rewiring probabilities' (in network language) accelerate the decision making process both locally and globally. Again, this is something we must learn from.

In a paper by Calvert Jones at UC Berkeley (check http://www.sims.berkeley.edu/~cjones/Full%20Text%20Papers/Calvert%20Jones_Al%20Qaeda's%20Innovative%20Improvisers.pdf#search='al%20qaeda%20and%20network%20theory'), which addresses al Qaeda's networks and the ability to learn, adapt and innovate, a similar idea is presented:

"Although its globally oriented ideology, spectacular successes, and US focus in a time of rising anti-Americanism have no doubt also encouraged worldwide collaboration, the loss of the state-based sanctuary is likely to have accelerated the process. According to Bruce Hoffman, the greatest change after the wars in Afghanistan and Iraq has been the breakdown in the rigid boundaries that used to separate militant Islamist groups.8 These groups are growing much more permeable, as individuals from varying backgrounds, of different nationalities, and with diverse skill sets are working together in temporary teams to pursue similar goals. Although some, primarily motivated by local grievances, may not share Al-Qaeda’s commitment to a globally oriented jihad, many have adopted a more pragmatic approach to terrorism, having recognized the advantages of increased cooperation and networking when domestic efforts fail. Al-Qaeda’s cooperation in varying degrees with Jemaah Islamiyah, the Abu Sayyaf Group, Hezbollah, and the Egyptian Islamic Jihad is well documented.9 The more decentralized, networked structure – as an adaptation to the loss of the Afghan sanctuary – is likely to encourage further interorganizational cooperation. "

The key is that the noise in the main network as well as the loose ties to other networks has broken down boundaries and allowed widespread consensus to be reached, leading to an insurgency that apparently has surprised most military personnel and war planners. It is time that traditional war games, planning and training need to move on and research into areas like network theory must become much more prominent. Perhaps the results coming out of network and organizational theory research would have changed some minds and resulted in a more prepared occupancy of Iraq.

I hope this spurs good discussion and debate, and I encourage comments from those who have a much better grasp on military planning than I have to see what you think about how what has happened on the ground in Iraq fits into some of the research that has been producing results and predictions from network theory.

2 comments:

mark said...

Very nice. One of your best posts.

" What almost certainly started off as a majority of individual agents in the network that would welcome the U.S. was now put into an environment where the boundaries between the majority and minority views, which hold steady with no noise, suddenly are perturbed from noise. The boundaries break down, and suddenly more in the majority change their view."

Phase transition state ? Or no ?
Is a Phase transition here a point where the system could go in any direction or merely linear from
A -----> B ?

Accumulation of "noise" minor perturbations equal a single large system perturbatiuon ? U.S. invasion was a " Big Bang" that created potentialities - occupation was the death of 1000 cuts that reduced them

Mark Vondracek said...

Thanks, Mark. It isn't clear to me if this is the social equivalent or analog to a physical phase transition. There has been a 'change of heart' among the Iraqi civilians, but I don't know if this change in attitude is a change of state.

The invasion certainly was teh 'shock and awe' big bang. But the public consensus appeared to be with us. The looting was the beginning of the breakdown, though, in my opinion. That was a clear indication we did not have a security plan or the troop level to fill in the vacuum. It both created and invited 'noise' that began to work its way into the network and had a negative impact on the decision of the people to back the U.S. Minds that were made up to support us began to change...a case where a rule set evolved where a majority state was undone and the initial minority became the majority. That would make for an interesting study to see if the nature of such rule sets could be determined. But it is obviously a complicated problem.