The cover story of this month’s Nature features the work of a team of researchers examining the mathematical properties of insurgency. One of the authors is Sean Gourley, a physicist by training and TED Fellow, and this work represents the culmination of research by Gourley and his co-authors—a body of work that I have been critical of in the past. The article is entitled, “Common ecology quantifies human insurgency,” (gated) and the article attempts to define the underlying dynamics of insurgency in terms of a particular probability distribution; specifically, the power-law distribution, and how this affects the strategy of insurgents.
First, I am very pleased that this research is receiving such a high level of recognition in the scientific community, e.g., Sean tweeted that this article “beat out ‘the new earth’ discovery and the ‘possible cancer cure’ for the cover of nature.” Scholarship on the micro-level dynamics of a conflict is undoubtedly the future of conflict science, and these authors have ambitiously pushed the envelope; collecting an impressive data set spanning both time and conflict geography. Bearing in mind the undeniable value of this work, it is important to note that several claims made by the authors do not seem consistent with the data, or are at least require a dubious suspension of disbelief.
In many ways I reject the primary thrust of the article, which is that because the frequency and magnitude of attacks in an insurgency follows a power-law distribution this somehow illuminates the underlying decision calculus of insurgents. Without belaboring a point that I have made in the past, the observation that conflicts follow a power-law is in no way novel, and I am disappointed that the authors failed to cite though I am encouraged that the authors did cite the seminal work on this subject (thank you for pointing out my errata, Sean). The data measures the lethality and frequency of attacks perpetrated in the Iraq, Afghanistan, Peru and Colombia insurgencies, but the connection between this and the strategy of an insurgent is missing.
The authors’ primary data sources are open media reports on attacks; therefore, their observation simply reveals that open-source reporting on successful insurgent attacks follows a power-law. There are two critical limitations in the data that prevent it from fully answering the questions posited by the authors. First, there is some non-negligible level of left-censoring, i.e., we can never attempt to quantify the attacks that are planned by insurgents and never carried out, or those that are attempted by fail (defective IEDs, incompetent actors, etc.). Although they do not inflict damage, these attacks a clearly byproducts of insurgent strategy, and therefore must be present in a model of this calculus. Second, while the authors claim to overcome selection bias by cross-validating attack observations, this remains a persistent problem. Consider the insurgencies in Iraq and Afghanistan; in the former most of the attacks occurred in heavily populated urban areas, garnering considerable media coverage. In contrast, Afghanistan is largely a rural country, where the level of media scrutiny is considerably lower, meaning that media outlets there are inherently selective in what they report, or most reports are generated by US DoD reporting. How do we handle the absence of attack observations for Afghan villages outside the purview of the mainstream media?
The role of the media is central to the decision model proposed by the authors, which is illustrated in the figure above. Again, however, this presents a logical disconnect. As the figure describes, the authors claim that insurgents are updating their beliefs and strategies based on the information and signals they receive from broadcast news, then deciding whether to execute an attack. For lack of a better term, this is clearly putting the cart before the horse. The media is reporting attacks, as the authors’ data clearly proves; therefore, the insurgents’ decision to attack is creating news, and as such insurgents are gaining no new information from media reports on attacks that they themselves have perpetrated. Rather, the insurgents retain a critical element of private information, and are updating based on the counter-insurgency policies of the state—information they are very likely not receiving from the media. The framework presented here is akin to claiming that in a game of football (American) the offense is updating their strategy in the huddle before ever having seen how the defense lines up. Without question updating, in football both sides are updating strategy constantly, but it is the offense that dictates this tempo, and in an insurgency the insurgents are on offense.
This interplay between an insurgency and the state is what must be the focus of future research on the micro-dynamics of conflict. From the perspective of this research, a more novel track would be to attempt to find an insurgency that does not follow a power-law; but rather a less skewed distributions, such as the log-normal or a properly fit Poisson. Future research may also benefit from examining the distribution of attacks in the immediate or long-term aftermath of a variation in counter-insurgency policy. After addressing some of the limitations described above, such research might begin to identify the factors that contributed to why some counter-insurgency policies shift the attack distribution away from the power-law. The key to any future research; however, is to connect this to the context of the conflict in a meaningful way.
Again, congratulations to Sean and his team, I hope their piece will initiate a productive discussion in both academic and policy arenas on the methods and techniques for studying the micro-dynamics of conflict.
Photo: Nature
Automatically Generated Related posts:
Drew,
I wholeheartedly agree with your critique. Gourley et. al use a novel data source, but I think their interesting statistical methodology and heroic data collection can not overcome a poor research design. What do we gain by knowing that casualties from insurgency events follow a power law? Moreover, the idea that we can just infer insurgency dynamics without looking at how governments’ and incumbent forces respond to insurgencies I think is myopic (see Figure 4 in your blog post) . Finally, the most interesting dynamics to scholars and policy officials is not what makes these insurgencies similar (even though I am skeptical, that as you point out they are actually measure the power of insurgency events and not some artifact of country-level reporting procedures), but rather what makes Afghanistan, Iraq, Peru, and Colombia unique? Why did Peru’s insurgency die out and why is Colombia’s still going on? These are the important questions. Without looking at the internal political dynamics, the organization of the insurgency, relevant external actors etc. any study of insurgencies is extremely limited in its usefulness.
[Reply]
Drew Conway Reply:
December 18th, 2009 at 8:33 am
Your point is extremely critical; that is, we should be at least as interested in what makes these insurgencies different.
[Reply]
I think you’re going a bit too easy on Gourley’s methods. The Security Crank just wrote a piece detailing how badly unreliable open media sources are for determining mass casualty attacks in the war in Afghanistan. Similarly, the Lancet controversy indicates Iraq poses similar problems in terms of accurately gauging war casualties from media estimates. If that basic of a data set is unreliable, how could a broad study of insurgency that relies on it be useful?
I just don’t see how Gourley’s work is anything other than finding a convenient group of data, then drawing a curve to fit it with no other work. In Afghanistan there are literally thousands of deadly IED incidents each year, but only a few dozen make it into the media—even ignoring your very good points about about other factors he ignores, that basic issue with the data poses grave concerns about the conclusions he can draw.
The data problem is a fundamental one: the analysis of it is relative simple (especially considering the rich history of such data studies, going all the way back to Clausewitz and Jomini which Gourley doesn’t seem aware of), but getting a good set of data is very difficult. It simply doesn’t exist in ephemeral news accounts. I can’t stress how critical this is, especially since Gourley is trying to give this model predictive power: there is no justification for it (nor is there, frankly, for Nature to give his paper the cover).
I applaud Gourley’s effort, both from a bias that prefers cross-disciplinary approaches and because falsified theories are still useful. But this isn’t an especially revelatory, or explanatory thing he’s going.
[Reply]
Drew Conway Reply:
December 18th, 2009 at 9:00 am
Josh,
Yours is a classic conundrum of non-social scientist attempting to do social science. While there is clearly value in examining the same problem from many perspective, I find that often (and this piece is a good example) that the so-called “hard sciences” will ignore the human element of a problem when looking at it. In this case, the authors have focused on collecting a large data set, but ignored the human data generating process.
[Reply]
Hi Drew,
I agree with your critique to their methods and approach. In a discussion with a friend of mine over twitter, I actually likened their top down approach to developing a model that says “The insurgents haven’t done something major for a while, they must be planning something nasty these days”.
This is almost a characteristic in top down approaches to social sciences – pretty much like Markowitz’s portfolio model. A more fundamental research of population dynamics, and a deeper understanding of their decision making process would yield not only more credible, but also more interesting results.
However, I would like to indicate that I haven’t read the paper, so I don’t claim that what I have just written here is more than just an impression. If anyone has read it or even talked to the authors, then maybe they can enlighten us more on their methods and approaches, as well as how they think their findings could be of benefit in the future.
Y.
[Reply]
Drew Conway Reply:
December 18th, 2009 at 9:04 am
Yaser,
Thanks for your comment, and I certainly recommend reading the paper if you can get a copy. Your quotation from Twitter is quite telling, a classic example of how people are very good at interpreting the probability of events over the short-term when we have no idea what the underlying likelihood of an event is.
[Reply]
Yaser Helmy Reply:
December 19th, 2009 at 7:10 am
Hi Drew. I read the article, but i don’t have access to the paper. if you can share it, that’d be great.
Y.
[Reply]
[...] paper (which is gated) or the editors note or news article(s) or Q’n'A session or TED Talk or commentary or books like Naked Ape or Human Zoo or just about any sort of reference research whatsoever -1) [...]
[...] interesting linkswap between security geeks Drew Conway and John Robb over a recent piece of research in Nature about power laws and insurgency (which you [...]
Ahem… http://mathematicsofwar.com/
I can’t do the impressive mathematical acrobatics but I can spot a typo in a URL now and again
[Reply]
Thanks for the post Drew. You can read more about our research on the new website http://www.mathematicsofwar.com
On this site you can download copies of the current Nature paper and supplementary info. You can also access working papers, background reading and detailed discussion of the types of models we are using for the research.
[Reply]
It is simply amazing that when a research project in fact shows us something that no other methodolgy/targeting mechanism/military decision making process has not shown us it gets’ trashed instead of giving it some thorough thought–typical knee-jerk reactions.
Nothing up to now with the exception of the “open source warfare” thinking of John Robb is being used to explain just how it was possible to go to a full blown phase two and elments of a phase three guerilla insurgency in under three years especially during the years 2005 through to the end of 2007—AND nothing is out there that explains the fast evolutionary tempo on insurgency weapons/IED development against every counter IED concept we have thrown at them up to now–to include the ability to hack Predator video feeds. The insurgency ops tempo and sheer number of IEDs weekly were really never reported fully by US media.
No one it seems wants to touch the question that Sun Tzu formulated “no one can stop innovation”.
If in fact the scientific community on their own validates to a very large degree the “open source warfare thinking” and one can now in fact test different strategies without losing troops on the ground one would think it would be greeted with a large shout—but it appears that since it did not come out of the IC/JIEDDO worlds it is to be ignored and put down.
One really does need to pay super attention to the insurgent videos being released almost daily—to many these videos as insurgency propaganda to some of us deep insight into their C2, thinking and weapons development as they really do not care if we are monitoring them as it is for internal consumption.
I have seen recently TB internet released pamphlets depicting in Arabic and Pashto every feature for a satcon cell phone and to include frequency skip—I have seen a video depicting a laptop being connected to a satcom cell phone to get to the Internet in the middle of the snow covered mountains powered by a gas driven generator—think what the TB could be doing if they were literate.
If this research allows one to fully understand an insurgency then run with it -test it to confirm or deny the thinking–do not simply critize it.
[Reply]
Richard,
I have given this research considerable thought, which is why I chose to write no less than four posts on the topic. Also, you are incorrect in asserting that this research, or the work of John Robb, are the first to attempt to quantify or explain the dramatic escalation observed of violence in conflict. The statistical analysis used in this research goes back 60 years, and the analysis of the micro-level dynamics of insurgency goes back to Clausewitz.
[Reply]
Drew-having spent a large amount of time talking with actual Iraqi insurgents from the Ansar al Sunnah types through to AQI/QJBR members it was interesting that their input as to how they functioned and innovated was constantly “surprising” the intelligence community.
While you might be right that dramatic escalation has been researched and researched I still do not see that any of the previous research actually explaining in any fashion of just how does the insurgency counter at an increasing level of speed any new COIN military strategy and or counter IED TTPs thrown at them.
You as well did not mention just how was it possible that we found ourselves in a full blown phase two guerilla insurgency by 2005 (inside two years) —have not seen anything from the research community even remotely talking about the speed of innovation in 2004 and 2005 as John was discussing in 2004 and 2005.
The Nature research may in fact explain something that both the IC as well as the research community is not answering that was in fact was easy to watch, but not explain—namely the ops tempo of the Sunni insurgency from 2005 through to early 2008. 2500 IEDS in a given week, cycle of attacks meaning a series of attacks in the North that would draw US forces North, then a pause—then attacks in the South drawing our attention to the South with a pause and then attacks in Diyala drawing us deeper into Diyala Province—it was whiplashing the troops to death and the IC was never ever able to get ahead of the cycle.
What was interesting is that between every series of insurgent attacks you could in fact see the internet media releases and then a pause would occur before the next series of attacks which is something the Nature article in fact starts to explain.
What I miss in all of the previous research is some frank and honest understanding of “insurgency”, battle tactics of insurgents and the speed of innovation supported via the internet.
I would even go as far as to suggest that the use of the Nature research might in fact answer just how did the Taliban re-emerge even stronger starting in 2007—I would go as far as to suggest that the strong emergence of Taliban videos being released at a faster pace than in any previous period could be in fact be paralleling their battlefield activities.
As long as the battle/IO videos being released by global insurgent groups are considered “enemy propaganda” we lose vital insight into their thinking and innovation.
[Reply]
Drew—written by John in 2004 and saw in Iraq in 2005/2006 actual confirmation of this thinking. Even now in late 2009 I still see battle videos coming out of Iraq and Afghanistan to confirm to a high degree John’s initial thinking on the topic. Actually the “opitmal size of a terrorist network” may prove to be highly accurate if one does a thorough study of insurgent battle videos that are recorded as part of the Dark Web project at the University of Arizona.
Friday, 24 September 2004
THE BAZAAR’S OPEN SOURCE PLATFORM
Earlier analysis (see the “The Optimal Size of a Terrorist Network” for more) indicates that the disruption of al Qaeda network mega-hub in Afghanistan has put strict limits on the size of the surviving virtual network elements. This size limitation may represent a barrier to attacks on the US, but is likely well within the capabilities of what is necessary for limited regional attacks. However, new innovations in group dynamics and the emergence of new unaffiliated guerrilla networks in Iraq may provide a method for regaining strategic capability.
The Bazaar
The decentralized, and seemingly chaotic guerrilla war in Iraq demonstrates a pattern that will likely serve as a model for next generation terrorists. This pattern shows a level of learning, activity, and success similar to what we see in the open source software community. I call this pattern the bazaar. The bazaar solves the problem: how do small, potentially antagonistic networks combine to conduct war? Lessons from Eric Raymond’s “The Cathedral and the Bazaar” provides a starting point for further analysis. Here are the factors that apply (from the perspective of the guerrillas):
Release early and often. Try new forms of attacks against different types of targets early and often. Don’t wait for a perfect plan.
Given a large enough pool of co-developers, any difficult problem will be seen as obvious by someone, and solved. Eventually some participant of the bazaar will find a way to disrupt a particularly difficult target. All you need to do is copy the process they used.
Your co-developers (beta-testers) are your most valuable resource. The other guerrilla networks in the bazaar are your most valuable allies. They will innovate on your plans, swarm on weaknesses you identify, and protect you by creating system noise.
Recognize good ideas from your co-developers. Simple attacks that have immediate and far-reaching impact should be adopted.
Perfection is achieved when there is nothing left to take away (simplicity). The easier the attack is, the more easily it will be adopted. Complexity prevents swarming that both amplifies and protects.
Tools are often used in unexpected ways. An attack method can often find reuse in unexpected ways.
Scaling the Bazaar
The bazaar dynamic — replete with stigmergic learning and entrepreneurial ventures — is vibrant enough to keep Iraq in a state of chaos. The statistics speak for themselves. However, can the bazaar be exported to regional nations or strategic targets? Can it serve as a post Afghanistan (post al Qaeda) model for global guerrilla warfare? Yes. Here’s why:
Leveraged attacks. As we see in Iraq, if appropriately planned, small attacks can have amazing impact. The reason behind this are the system dynamics that amplify results. ROIs (returns on investment) in excess of one million fold have been measured in Iraq. This means that smaller groups can have tremendous impact at the strategic level if they adopt the Iraqi method.
Swarms vs. single group activity. The bazaar offers the potential of many smaller attacks that can in aggregate have an impact equal to several large attacks. Many hands make light work. Combined with system leverage, this could reduce a nation to economic chaos in short order.
Rapid innovation. The bazaar’s demonstrated ability to provide rapid innovatation makes defense much extremely difficult. Rather than a single 9/11 style attack, we may see small attacks (less planning and training, fewer people, less support) against a plethora of targets. With a sufficient number of guerrilla networks unearthing vulnerabilities (particularly ones with system’s leverage), security forces will likey be outmatched.
Posted by John Robb on Friday, 24 September 2004 at 01:35 PM | Permalink
[Reply]
“However, new innovations in group dynamics and the emergence of new unaffiliated guerrilla networks in Iraq may provide a method for regaining strategic capability.”
OK, so do you think Al Qaeda read John’s blog and acted correctly to the information provided, or do you think John reported the facts correctly and his analysis was after the fact and taken from computer warfare between Microsoft and a bunch of winners and losers?
Actually, there is a fine line between selling books and serving your country and the information on Drew’s site seems to be helping to sort this out. after all, you are assuming the people who need the information would receive it, if it was just posted on Drew’s site to begin with.
Because of the probability of irrelevancy, I don’t think you want Drew to be less skeptical of the power-law, maybe even more so. There should be no easy answers.
I am sure we all do what we can; the fact is most of us, and I consider myself in that statement, are just noise. The effort should be in giving you no false positives. When you hear truth, jump on it. As it is trust is more important than truth.
[Reply]
Larry—it should be noted that the initial writings on the Iraqi insurgency activities by John Robb were in the period 2004-2005. The development of the COIN concepts espoused currently really did not evolve into hands-on Army training center scenarios until 2007 and the those scenarios were based really on experience taken from Brigade Commanders who were indicating what was working and what was not working.
The COIN manual does not at least the last time I looked at it explain the innovation that we were seeing in insurgency structure and battle tactics as early as the beginning of 2004. There were indications found in documents captured during 2005 of journals written in 2004 by insurgent commanders that they were testing and deploying IEDs as early as late 2003, lists of funding expedientures for IED trigger devices and explosives, money exchanging to support the insurgency, and names of virtually a Who’s Who of insurgent leaders who by the way still have not been captured such as al Duri.
The ability to build bounding IEDs that have in fact “downed” US copters is not your normal insurgency innovation easily explained by current methodologies–but though through the concept of “open source warfare” it does in fact explain why/how the innovation is occuring and in what direction it will go as a particular weapon systems evolves from an initial single insurgent group attack using it to it’s becoming a standard weapon of choice for all insurgent groups such as the evolution of the RKG-3 anti tank grenade.
Now that we are looking at the Afghanistan front as the Iraqi front is winding down there is in fact more and more open source literature indicating that the Attack the Network (AtN) concepts and methodologies used in Iraq are not working in defeating IEDs while the importance of the IED weapon is not as important to the Taliban as is their controlling of population living in villages and valleys. Now it looks like over six years of COIN has gone in the wrong direction and in fact “open source warfare” does go deeper in explaining what the “boots on the ground” are actually seeing.
I would like to see a far deeper understanding of just how has technology changed the face of guerilla warfare and how does one quantify the innovation being driven by that technology as it is driven over the Internet.
What is interesting in the Nature article is that it appears to be the first attempt in looking at the concept of “open source warfare” from a scientific viewpoint. After watching countless analytical methodologies come and go in the last forty years we are still searching of the “correct viewpoint”. So take the input of the Nature article and test it as thoroughly as necessary to see if it holds water.
The use of COTS and off the shelve devices are being used in ways that scientific research cannot predict unless one really understands how a human is using that device or product. But you can in fact trace the way the device or software is being spread by watching the media releases of the various insurgent groups.
[Reply]
Larry and Drew—read the just released from CNAS document on the failure of military intelligence in Afghanistan you will really understand where I am coming from.
Couple the “open source warfare theory” together with the Nature magazine research, coupled with input from cultural, human terrain, and governmenal aspects each does to a degree feed into the “open source warfare theory” and you will have a vibrant intellectual exchange that is simply missing from current intel analysis thinking.
There needs to be a deep thinking out of the box that does not tie itself to just one or two research methodologies, but is willing to challegene and examine new areas in new ways in a fast paced environment to get ahead and stay ahead of the insurgency developement curve. Only when you get a tad ahead can you stay instep with their developements.
[Reply]
I read the CNSA report and it sounds a lot like you are sending bloggers to Afghanistan, to bridge the gap between the kinetic and potential energy in the information being used there. While the military has, as the report indicates, many potential bloggers to choose from, it doesn’t sound like the system you plan to use will work. Perhaps, what I consider as the Chinese model would work better. Apparently, and I do not have any explicit knowledge of this happening, the Chinese are buying up memes and using them as sounding boards against this administration. You could do the same thing by hiring memes already in place and documenting, with help from the blogger, the structure they use to penetrate the internet. With this structure, you could then put in place one or more of your writers to accomplish the stated goal of connecting with the people of Afghanistan.
As for the Nature report, I found the math of figure 4 to be wrong. This is not unexpected, because the math once said that the sun revolved around the earth, and it was not wrong, but you could have never gotten there, because the structure was actually something entirely else. While the feedback loop and the Data Analysis look ok, to coalescence and fragment is a judgment call, and I don’t see that form of acceleration within the equation. It may be there, but then I don’t do calculus. Given time, I will re-read the Nature report and information on The Mathematics of War web site again, now that I know what I am looking for.
[Reply]
Larry—reference the CNAS article. This came to me from the schoolhouse of US Army Military Intelligence.
This article is being circulated; while MG Flynn is taking professional risk in telling the emperor he has no clothes (intel). It’s a war that has never been fully supported so policy makers are not interested in hearing from the good idea fairy. Risk adverse behavior will continue to be the norm for conventional/ISAF leaders.
What is causing some heartburn in the intel community is the emphasis on the intel needs of anti-insurgency vs that of intel needs for true counterinsurgency and there is a great difference between the two.
While it is far easier to conduct anti-insurgency as it the simple concept of kill or capture and the “winning” is easy to quantify–but to truly understand and fight a “fourth generation insurgency” is far more time consuming and extremely difficult to counter. And when you have a entire defense contracting world pulling on the anti-insurgent rope it is difficult to go in the opposite direction.
The article release by Gen. Flynn was in fact an eye opener—and nothing made the US mass media.
[Reply]
Larry Dunbar Reply:
January 11th, 2010 at 1:19 pm
The US mass media is made for the masses not by the masses. There is little incentive to produce nation-building articles, in the past forms of entertainment based recordings of history. One of the biggest owners of a mass media outlet in America is an Australian who is at the same time feeding from the trough of a divided America, while his own country is being over-run by hordes of Chinese using the same roadmap that the Japanese used during WWII. I should note, only this time the Chinese have the bulldozers and not the Americans who came in to help the Australians during WWII.
The old mass media has as much incentive to see nation-building going on in America as the Taliban have in Afghanistan. The only difference I can see is that the Taliban don’t eat their young, as do those of the media industry, they just feed them to the enemy.
Of course in today’s globally connected world there really isn’t any enemy as before. America should understand that we are going to become more totalitarian (or as the conservatives now call it Authoritarian) than before, with our resources coming to those who need resources and away from those who don’t have resources. In fact those without resources are going to have what they have taken away from them, except out of the benevolence of those with resources. Our corporations are demanding this and we the people don’t really have much choice, because of the complexity and corruption in our political system. As there are only a handful of the old mass media corporations that have most of the resources, it pays them to use those resources to keep viewers entertained instead of informed on where those resources are going to or coming from.
The closest form of old mass media nation-building reporting comes from C-Span and I watched several stories about the CNSA article. But, as you say, I bet as many people watch C-Span as read a post coming from ZIA.
I think the ideas in the CNSA report are the correct ones, but are too little too late. The best you can do now is to hit the Taliban hard, and knock the holy fuck out of them, while protecting the population, at as close to a 50-1 ratio, as much as possible. Once pressure is applied, if it is even possible, then you need to negotiate, with the Taliban, from a position of strength that this new surge might afford you. Changing the way you use all the information collected takes time that I am not sure you have, unless within the next 3 years you have magically inserted the correct DNA into the Afghan military. Also, these bloggers you’re deploying in Afghanistan need to be deployed in D.C as well as on the front lines, if you can find the line, and the front. Afghanistan is actually a center.
[Reply]
[...] Zero Intelligence Agents by Drew [...]