EPIDEMIC INSURGENCY (Part 1)

Iraq's insurgency is both growing and innovating quickly.  A good way to understand this speed is to dive into how epidemics spread and cascade in social networks.  In this first brief on this topic, I will look at how the innovation spreads through the global guerrilla network in Iraq -- the epidemic spread of information to individuals/groups that are highly susceptible (those that have already opted to join the insurgency). In the second brief, I will examine how the insurgency infects the general population -- is it an epidemic or not?  As always, I am open to ideas on how to improve this analysis.

Epidemic Innovation

Iraq's insurgency has demonstrated that their decentralized process of innovation (open source warfare) can yield effective methods of system disruption.  These innovations appear to be spreading quickly.  A good way to understand why, is to treat this tactical innovation as an informational epidemic.  The spread of an epidemic to susceptible individuals is determined by its reproduction rate (the ability to infect others).  The equation for this is:

Rate of reproduction = (duration of contact) x (likelihood of contact) x (duration of infectiousness)

Tactical innovation travels quickly in Iraq's guerrilla network because the rate of reproduction is extremely high.  The primary reason is that the "likelihood of contact" is large relative to the other factors due to the topology of its decentralized network structure.  Three topological features of Iraq's insurgency feed this high "likelihood of contact" are:

• Small cohesive cells.  These small organizational units exhibit high degrees of structural cohesion (Menger's Theorom).  Every member knows all the other members (this is different than the 9/11 operational network).  Each cell is composed of family members, friends, and neighbors.    
• Small world properties (Milgram).  Cross connections between members of the insurgency radically reduce the mean path length (the mean distance between any two members) of the network.  Baath party membership, military experience, etc. provide these short-cuts.  Further, a modern road network makes it easy to make physical connections.
• Scale free properties (Barabasi).  The guerrilla network in Iraq is likely scale free.  Early entrants, such as the Fedayeen and al Qaeda are highly connected hubs.  As the network grows, they become more central and powerful. 

How this works

The spread of tactical innovation in Iraq's guerrilla network is a combination of the overt (direct communication via personal connections in the bazaar) and passive (stigmergic environmental signals).  The process by which this works follows this pattern:

• Innovation.  Small groups (clusters) innovate within the open source model of guerrilla warfare we see in Iraq.  They are independent and therefore able to make decisions based on their own decision making processes.  Innovation is often incentivized by the reward of funding and other profit opportunities (as with any entrepreneurial activity).
• Adoption.  An innovation is widely dispersed when a network hub adopts it.  Hubs are influenced through a combination of stigmergic and overt communication.  If the innovation works, the hubs typically adopt the innovation.
• Propagation.  Hubs directly influence other groups through economic incentives in the bazaar (they have influence over sources of funding in their roles as violence capitalists).  They also have a high number of overt connections to entrepreneurial guerrilla start-ups. 

Breaking the Connections

One objective of counter-insurgency should be to lower the rate of reproduction of the epidemic.  Effectiveness in this area would slow the aggregate decision making of the group (Boyd) and make it easier to beat.  Here are some ideas (note:  some of the ideas here are only valid given that it appears that the US has given up the moral war against the Sunni insurgency, and has opted for a military only approach that targets other psychological factors that impact decision loops):

• Reduce the duration of the contact.  Indirectly, this was an objective of the Fallujah operation.  The collapse of the Fallujah TAZ hinders ongoing face-to-face high duration contact.  Unfortunately, the requirements of this operation are so excessive in terms of effort and personnel, we are limited to a single TAZ at a time.  A better method must be found given our limited resources.  Additionally, the Fallujah TAZ was less important to the insurgency than the military assumes.
• Eliminate connections by isolating cells (hubs in particular).  Iraq's state of emergency is aimed at this.  However, phone networks and Internet connections are still available in the region.  One method would include reducing the insurgent areas to older means of communication. Another would be a strict limit the on physical movement between cities and towns.
• Limit the duration of the infectiousness.  This requires rapid adaptation on how we defend targets in Iraq.  Early detection of an innovation against a class of targets, should be countered.  This will require a real "strategic corporal" (not merely one that avoids moral mistakes through good judgment) and new ways of sharing innovation horizontally.

A general note on many of these tactics:  they contribute to a moral loss.  They also indirectly undermine the government's moral legitimacy by disconnecting the economy from globalization (which is a goal of global guerrillas).

Comments

[Paul Snively]

John, you may find it interesting to read http://www.cs.utexas.edu/~lorenzo/corsi/cs395t/04S/notes/p1-demers.pdf, "Epidemic Algorithms for Replicated Database Maintenance," which dates to 1987, and influenced Xerox's "Clearinghouse" name server design as well as Apple's "Rumor Monger" app. It's a good formal study of how the mathematics of epidemiology can drive the disemination of information--in other words, right up your alley.

[Paul Snively]

John, you may find it interesting to read http://www.cs.utexas.edu/~lorenzo/corsi/cs395t/04S/notes/p1-demers.pdf, "Epidemic Algorithms for Replicated Database Maintenance," which dates to 1987, and influenced Xerox's "Clearinghouse" name server design as well as Apple's "Rumor Monger" app. It's a good formal study of how the mathematics of epidemiology can drive the disemination of information--in other words, right up your alley.

[nathan]

Yet another insightful post.
The connection to computer viruses and network interaction is illuminating to the extreme. I am a nurse by profession at the moment and the parallels of analogy between new age terrorism and the biology of the HIV virus inside the human body along the lines of things like receptor site up-takes are fascinating. The way to defeat AIDS-cells from interfacing with these receptor sites (Hubs) is use of decentralized, free roaming anti-bodies (loyalist paramilitaries?).
Using the above analogy it would seem that the military, in order to effectively conduct counter-insurgency operations would have to become more confederation than command. Semi-autonomous field units of varying size able to maximize response times with good coordination of communications and extreme flexibility of chain of command. If necessary the disparate units could come together following a pre-set SOP to swarm a battlefield. The Marine Corps have been more open to moving in this direction though it seems the Army lags behind in will to do the same.

[Greg]

for the second time, a high ranking US general has claimed the back of the insurgency is broken. is he accurate in the level of importance he applies to Fallujah as a hub? I'd tend to doubt it, mainly because the insurgency existed before it had that hub available. would eliminating more hubs suffice or is the internet and global media enough of a connection that terrorists can learn from each other's actions without direct contact between distant cells? just how much internet access does the Iraqi insurgency have? is it a significant medium for propagation of the insurgency?

[Daedalus]

Here's the main difference. This epidemic is lethal and there is not an unlimited supply of hosts.

But only is someone hired you to consult, this wouldn't be a problem, I'm sure.

[dingo]

There is always a danger in the use of analogies as analytical tools rather than as devices to explain the analyses themselves.

First problem in analyzing the insurgency is a lack of concrete information on it and an inability to test hypotheses about the way the insurgency works. The result is more speculation rather than fact, which may explain why so many competing theories about the insurgents and why they are fighting seem able to co-exist.

Second, treating the growth of insurgency as a single process (like an infection by a virus) rather than the result of the interaction of a variety of social forces to attain some result (think a political campaign), could be a flawed way to analyze the situation.

Third, the argument generally revolves around how America can win the hearts and minds of the local population. What if the history of American activity in the region and the way it exercises its power(i.e., through military domination by conventional forces) means that such a goal is impossible? What if ethnic and religious differences in the country prevent any kind of accomodation between competing factions in what has become an Iraqi civil war?

[John Robb]

Thanks Paul, I will give it a read.

Dingo, a couple things. This is a valid analytical tool and not an analogy. There is a large body of academic work on how information propagates within a network. The network's topology influences that flow. A finding from this research is that in certain configurations, information flows like an epidemic with a low threshold.

Analytical tools are only as good as their explanatory and predictive power. The current tools being used by our current crop of "military analysts" are highly subjective and prone to failure. My chosen techniques have been better at accomodating available information and predicting future events.

Thirdly, this brief isn't about hearts and minds and the general spread of the insurgency (I am planning to write about that in the future). It is about information flow (techniques, methods, etc.) within the existing insurgent network. It's useful to understand this and may yield effective methods of counter-insurgency.

[dingo]

Hi John:

Sorry, I was overly broad in the language I used. When I spoke about analogies taking over the analytical process, I was thinking more about the AIDS comparison another poster made and some things I've seen referenced on other websites. Less a complaint about the analysis than a caution. I can't argue with the use of models of network flows to gauge the spread of information among insurgents.

Also, I look forward to your views on the 'hearts and minds' issue. Seems to me that the military has begun with the assumption that 1) winning H&M is necessary to counter an insurgency, and 2) that hearts and minds can be won in Iraq. Yet when you look at the situation the Israelis face in the West Bank, you wonder why they begin at that point in the analysis. I haven't seen any commentators or theorists address that point, and it seems like somebody should think about it.

[nathan]

Uhm, Ok I take back the HIV stuff. Except, I would argue that HIV is a "lethal" epidemic as you described. I made an analogy to John's analytical model because the langue used and the "pathology" described do have many similarities. Use of analogies have two goals, to speak in a way that makes the reader/listener understand the subject matter more clearly, and two, to allow a "perspective" shift so that a problem can be looked at in a new or inventive way. Decentralized command structure may no be the answer, is I guess what your saying. May not be. May be. But be all means remove the analogy from the discussion.

[dingo]

Nathan:

No need to take it off the table. As I told John, I was overly broad. What I had in mind was the kind of analysis where there is a runaway analogy: rather than using the analogy to explain the mechanics of the phenomena, we start treating the analogy and the phenomena as if they're the same process.

[nathan]

Dingo--Right, the "can't see the forrest for the trees," analogy. Look, I just made an analogy about our discussion of analogies. Please help--I have a problem and I can't stop--lol

[Rikkert]

Hi John,
A very insightful analysis. Let me share my thoughts.
Despite the apparent small world and scale free properties of guerrilla networks, I think they are likely not to have nodes of very high degree since that would undoubtedly bring with it significant security risks. In this respect they are unlike typical Albert-Barabasi networks, preferential attachment probably turns into the opposite at very high node degree. It is not unthinkable media networks play an important part, substituting for high degree nodes within the network.
To counter the guerrillas I agree the coalition should alter its pace of action (duration of feedback- or OODA loop). Right now the guerrillas are dictating the troop deployments of the coalition which is inherently not a good thing for the coalition. Fallujah provides the perfect example, a group of guerrillas took over the town, most of them left, lots of coalition forces tied down and the guerrillas are having a field day elsewhere, taking over new towns for this whole scenario to repeat itself.
Coalition forces should either speed up and catch up with the guerrillas (you seem to propose this, but maybe it is not really possible?) or slow down. By that I mean concentrate forces on the general security in Iraq, don’t let guerrilla “strongholds” tie down forces in symbolic battles. Perhaps a “rapid response force” could deal with guerrilla hotspots, but it is absolutely essential to deny guerrillas the opportunity to just take over a town at a whim, otherwise the guerrillas will just move on to another town when threatened. That is the kind of basic security the coalition forces need to provide to scale the guerrilla infection back down to subvert levels. Increased security will also mean higher transaction costs at the bazaar (or limit the infectiousness of the disease if you like).
Any attempt to reduce the entropy of the guerrilla network that has the same effect on the other social networks of Iraq is likely to achieve the goals of the guerrillas by disrupting diverse systems on a large scale. Perhaps it is possible to radically increase the entropy in the network of links between coalition forces and the bazaar, for example by leveraging existing Iraqi social networks more effectively, in effect creating a “counter bazaar”?

[James Acres]

More on a conceptual level, but I found William McNeil's "Plagues and Peoples" and "Pursuit of Power" very useful in thinking about the ebb and flow of civilizations epidemologically.

Specifically (in "Pursuit of Power" I believe) for the current conflict he talks about how feudalism was reaction to the HRE's inability to protect society from the Viking threat. The HRE's civilizational network was far more centralized. It was difficult to react to Viking attacks before the Vikings dispersed to other targets.

Feudalism worked because it created a scale-free network of force. Instead of centralized, 'imperial' control of violence, it was pushed down to the level of petty feudal lords.

[John Robb]

Rikkert, thanks. Bazaar/market connections may serve as a means of supplanting overt connections. For example, I don't need to call all potential suppliers to inform them that I want a specific product, if I indicate my demand for that item in an open way. They will contact me. This is command and control through market manipulation. I will write more on this.

James, that is interesting. If global guerrilla system disruption spreads to revisionists world-wide, it would put pressure on states to fragment into a more viable size. Singapore anyone?

[nathan]

Vikings. Premier terrorists of the dark ages.
The idea of feudalism as modern analogy is very interesting. Two questions, since you guys seem to have such a good perspective. Was not the “Strategic Hamlet” model created by Marines in Vietnam a form of the feudalism? This did not work very well as I recall. Except that the Special Forces highlands models did work to a higher degree of success so perhaps win/loss is intrinsic to the character of the indigenous population and not the “model” itself? Don’t know.

My own lines along this was a decentralization concept as well. Putting it in feudalism terms seems a better way of expressing this. If asymmetrical (non-centralized) warfare is the new “chariot” then perhaps decentralized counter-measures will be the answer.

However (question here), centralization as an economic issue/model is seen as superior. Using analogies again--but if economic forces see centralization as a superior method of productivity then perhaps the reasons why this is seen as so by manufacturers can be seen as methods of strategic-thinking to apply to decentralized networks of Global guerillas.

One thing I think I see, however, is this. In order to enjoy “real” success, global guerillas must, like local insurgents, eventually parlay unconventional force into conventional force--the Vietnamese army is no longer made up of VC, it looks like the NVA now. For example, Zarqawi was decentralized, uncoordinated, and a terror phantom. He parlayed that into a fiefdom in Fallujah (either the city itself or just an AO inside it). The fiefdom provided a target. The target was destroyed. Now he has to build a fiefdom again from scratch--valuable lieutenants dead and instead of looking like an invincible ghost he is scurrying to make alliences with people he may not entirely trust to find a new place to make his CinC. Unless they (guerillas) want to remain marginalized, eventually they will have to conventionalize to actually control a Taliban-style country. That is: terror ideals= terror attacks= success generates currency in bazaar= ability to command a TAZ= enough TAZs= a failed state, and thus control. Each time they move up the food chain of importance they open themselves up to target and attack. However fighting them in such a manner is very reactive which is intrinsically bad. Thoughts?

[James Acres]

John:

My thinking is going in a different direction. Perhaps more in line with Nathan's.

I don't understand how Singapore is any less dependant on the hub-and-spoke infrastructure of globalization that the United States is. So I don't understand how switching to that level of governance will necessarily protect us from the GGs.

The Viking analogy isn't nearly perfect. They attacked or threatened the populace directly in order to carry away wealth. Feudalism worked because it gave the ability to counter-punch but also because it made it harder for the Vikings to strike their target. Peasants and wealth could hide within castle walls.

While city-states like Singapore might give us more supple control over violence it wouldn't do anything to protect the hub-and-spoke networks from attack. In fact it might make them harder to defend as no specific political entity would be responsible for the inter-city-state portions.

I wonder if the answer is going to lie in constructing scale-free distribution networks for all the vital services of globalized civilization. Transport, communications, energy, violence. Everything.

I don't know what such networks will look like; for energy at least my guess is it requires some replacement for oil. Something which, as Nathan says, the economic logic won't require hub-and-spoke distribution.

My only theory for what governance would look like in such a 'connected neo-feudal' society is something like the international standards commitees governing the internet and other high-tech realms.

Nathan:

I've a similar working theory regarding the GGs in Iraq. While their tactics are GG as John points out, their goals are explicitly statist. They wish to (re)gain control of Iraq. If we can demonstrate an ability to frustrate that goal they may quit before we do.

Flowing from this, the April 'retreat' from Fallujah was actually a tactic. Since our enemy wishes to control territory we allowed them to think they won territory from us. We then let them spend months demonstrating their ability to control territory.

This gave us a strategic target to attack. Retaking Fallujah is not just about the tactical end of denying the enemy a TAZ. It is more about the strategic end of proving to the enemy and spectators that the enemy cannot hold territory against us. This may allow us to achieve a political or moral victory.

[John Robb]

I don't see the drive towards a political/conventional force that most analysts do (it's a false historical lesson that will prove to be a bad assumption). The multiplicity of motivations for global guerrilla activity, its organizational structure, and the strength of coercion vs. direct rule argues against it.

Re: the Singapore comment. Just a little free thinking on "the open society and its enemies." I'm far from advocating state fragmentation as a likely historical outcome of this conflict. It does seem possible to me but not, as of yet, a high probability scenario.

[nathan]

James--I also so the "giving into" cleric Al-Sadr along these lines. He was allowed to thing he'd stood up to American forces. Was lured out of a political inconvienant enclave that made attack unpalatable and back into the slums of Baghdad where operations could be conducted with more force. The result--Sadr seemed to realize what had happened as soon as Sadr City was surrounded and came to the tables. He still snaps but it's not the same as when he was fighting from the mosque.

John--curious about one point, you think the idea that the GG's will eventually have to/desire to become more conventional is a bad or weak model? This would make for interesting reading as I have always operated under the assumption that the guerilla force is not an entity in and of itself but a phase in a larger plan ending with state control.

[jthomas]

It's interesting that the only overt methods for combatting insurgencies involve extreme oppression (shutting down the phone lines, cutting off even donkey cart transportation). Not only that, but these methods also increase exposure of the citizenry to foreign soldiers.

All of these situations cause great hostility within the occupied population, giving more people cause to join the insurgency, and undermining the moral authority of the "coalition."

A possible sub-military alternative might be to use covert actions such as quiet assassinations and "spamming" extremist networks with false prophets in the employ of the "coalition."

In Iraq, we have created a failed state, one that is already a breeding ground for a larger insurgency, and will continue to be so after we leave.

Our methods-- pervasive surveillance, lockdown, house to house searches-- tactically improve the situation, but strategically worsen it to a great degree. When we leave Iraq, we'll likely leave it in the hands of a weak, propped-up police state. If Iraq should descend into anarchy or become an Islamic extremist state, it will cause a creeping metastasis of insurgency in neighboring states.

It seems that what we're doing is so effective at producing this outcome, and that the outcome is so blatantly obvious from our course of actions, that I have to wonder if the outcome of an extremely destabilized middle east and creeping unrest across much of Europe and elsewhere might not be the ultimate goal.