Small Wars Journal

Populations as Complex Adaptive Systems: A Case Study of Corruption in Afghanistan

Fri, 08/26/2011 - 9:56am

Download the Full Article

In August 2009, the North Atlantic Treaty Organization’s International Security Assistance Force mission shifted from an enemy-centric, anti-insurgency campaign to a broader, population-centric counterinsurgency.   This strategy change immediately presented challenges to field of intelligence and its existing analytic methods as it struggled to cope with the complexity of population analysis.  However, a wave of innovation under the name of complexity theory is spreading through mathematics, computer science, biology, economics, and sociology that do cope with the challenges of analyzing entities as complex as populations.  The Emergent States Assessment  (ESA) is an analytic tool that attempts to exploit this wave to support decision makers in Counterinsurgency and Stability Operations.  To demonstrate the potential power of viewing populations through the lens of complexity this article examines the phenomenon of corruption in Afghanistan through the ESA framework.

Download the Full Article

About the Author(s)

MAJ Tom Pike is a Strategic Intelligence officer in the US Army. He has studied Complex Adaptive System applications to analysis for the past 8 years, initiated and contributed to Inter-Agency Agent Based Modelling efforts, and served as a representative to the Defense Intelligence Socio-Cultural Capabilities Council.

MAJ Eddie Brown is a Military Intelligence officer in the US Army. He is a SAMS graduate and has served as an intelligence planner for multiple tactical and operational military units.


Eddie Brown

Thu, 11/10/2011 - 8:13am

In reply to by G Martin

I'd like to amplify a point Tom made: that the article was an opening statement in what we hope will be a continuing discussion of how to apply a developing science to the analytic art form (and yes, we are shamelessly stealing these ideas from the richest minds we can find). The topic of Afghanistan provided a vehicle for that discussion. Of course it was never our intention to encourage false assumptions or silver bullets. And as far as your request for hypothesis testing: agreed. Whether or not that is modest enough, we leave to your judgment. The article was the first data point in an iterative process of testing and refinement.


Tue, 11/15/2011 - 1:51pm

In reply to by G Martin

G Martin:

I have two points. The first point is that we are in complete agreement with your main gripe. So much so it took us awhile to really understand it, because we did not even think it was an element of the debate. To explain let me give you some quick context, this article is based on a few years of discussion between myself and Eddie. I then attended the National Defense Intel College (now National Intelligence University) and wrote an unclassified thesis on which this article is based. Never was it even considered to try to model the AfPak or any other theater as the thesis. The intent the entire time was to build an analytical methodology that could be exploited from the tactical to the strategic level to aid operations to influence the system. My internal metric of success, since I just came from teaching and commanding at the MI schoolhouse (These comments and my article are my personal opinion and do not represent the views of the US Army or Intelligence Center) was a method that could be easily understood and leveraged by Soldiers right out of their individual training. The assumption implied in this decision is no one person can really understand these complex adaptive systems in which they are working. From the smallest level – the individual (the brain is a complex adaptive system), to the AfPak region and everywhere in between are an overwhelming array of complex adaptive systems. Therefore everyone needs to be involved and it seemed the best way to influence these systems was to appreciate its rich diversity of the engaged system and then leverage the diversity within our own CAS by figuring out how to improve the tools of all those working to influence their specific problem set. So for the quote the "If" should have been bolded, underlined and be about five font sizes bigger, because you cannot model it. The article is trying to promote a well-researched perspective that will help analysts and decision makers understand the interdependencies of the system they are trying to influence.

This leads right to the second point, how do we disprove our hypothesis? We cannot or more specifically we do not know how beyond trying in real life (because it is too hard to model). But has the status quo that is being used proven itself? Saying we can’t prove it so it shouldn’t be tried is I believe a false sense of security. Every action in these current theaters and ones past has been efforts to influence a complex adaptive system. Those attempts were based on the best knowledge of the time combined with experience. They begin to fail when the system adapts (the fitness landscape changes) and the particular methods are no longer as effective. This attempt is based on solid research and personnel experiences. Based on our own experiences we feel with a complex adaptive system based understanding of the world that we can understand issues, form Small Wars to International Politics to history more effectively and more easily than before we adopted this perspective. Now we are trying to convince others and if we can and if they find it successful then maybe enough people will employ this perspective (not necessarily ESA) to test how effective it is in places like AfPak. However, I would challenge anyone who says what is being done now is the proven, most effective method.

In summary, the article was an attempt to present a method to help analysts and decision makers understand the complex adaptive systems they are influencing more effectively. Not solve the world’s problems. It seemed the most effective way to present this was lay out the method and employ it. Although I agree with my analysis and conclusions the point is to present an alternative method. If someone said, “we are going to employ the conclusions but not the method across the organization.” We would adamantly oppose such a plan.

Thank you again for your comments.


G Martin

Thu, 11/10/2011 - 9:47am

In reply to by tompike

<em>"...I also argue that if we could run an effective agent based model comparing competing approach to influence Afghanistan this approach would have a higher probability of success...."</em>

I guess this is my main gripe- in my experience when statements like these are made by military folks they are usually followed by dogmatic attempts to ignore data to the contrary and focus on proving the theory.

I understand that rigorous testing has concluded that interdependencies are key to complex adaptive systems- but I would seriously doubt that rigorous testing has proven anything about our ability to target those interdependencies vice doing something else that is more effective for our purposes (for instance, maybe because of the impossibility to accurately predict where a complex adaptive system will go once we "influence" an interdependency- less direct targeting of them or even ignoring them may be the best COA in some/most circumstances...).

And I still think even if targeting interdependencies works in some circumstances- there are still other methods that may work better in others- depending on the specific complex adaptive system (CAS) and position in space/time it occupies. I mean, after all- that's what a CAS does- is adapt, and if our targeting methods work, I would guess that CAS's as a whole might change to make interdepencies even more resilient, less targetable, or more mysterious...



Wed, 11/09/2011 - 9:56pm

In reply to by G Martin

G Martin:

Thank you very much for your comments. Your thoughts and critiques are great. Let me start by saying that I concur with your concerns about the “dangers of a false sense of security” and that are no “silver bullets” for these problems. The problems associated with complex adaptive systems and their related emergent phenomenons are never easy to solve nor is there a clear point of victory. Clearly our qualifiers of “ideally put them on a path to democracy’ or “will allow for more effective action” were not sufficient to emphasize these points. However, there are also a few points that myself and Eddie would argue.

First, the most basic and central point to our argument is that the military, intel community etc must shift from a node focused approach to an interdependency focused approach. You pointed out that “This is a very positivist philosophy- similar to what we already have: identify key interdependencies (read:center of gravity or key nodes or whatever the latest systems thinking term is en vogue.)” In that statement I would say you glossed right over the main issue. The military focuses on key nodes (nodes are easy to target) but we don’t focus on interdependencies and we continue to try and use these system theory ideas as you pointed out - whatever the latest systems thinking term that is en vogue- while missing a fundamental concept. Yet using the terms and applying them is two different things and the central point of complexity is that the interdependencies make the system. By extension ESA is designed to provide an easy to use method to shift focus from nodes to interdependencies. Focusing on nodes will not work because the parts are less important than how they are connected. (Why we continue to apply system ideas is another discussion) Only if people accept that there is a fundamental difference between targeting nodes and targeting interdependencies will they realize what it means to attempt to influence a complex system. (This isn’t saying yo don’t need to blow stuff up but the understanding of why your blowing it up changes). This shift from nodes to interdependencies is a major change in how one typically views the world and in this shift rests every part of our argument.

With this base and in regards to some specific arguments:

You pointed out that it is arguable in 2009 ISAF went to a pop centric strategy – I agree and that is exactly the point, how do you really shift to a pop-centric strategy. Our answer/hypothesis is after years of research and talking with lots of experts is focus on the interdependencies- these like all complex systems is the essence of the population system one is engaging. This shift in and of itself will not be a silver bullet but I believe it will improve how fast we understand our environment and mitigate many mistakes by having a better understanding (not complete) of why the population is acting they way they are and what options we have to try and influence their behavior. Through this research it also became apparent that intel can add as many human/population related factors to JIPOE and IPB as they want but it will have little effect as they always end with “this is what the enemy will to do us.” Without question understanding the enemy should be emphasized. I have gone on hundreds of patrols and I had an intense desire to survive but I also had an intense desire to make sure my fight meant something and to do that I needed to understand the population (as best I can and in as short a time as possible) and influence it. So both defeating the enemy and influencing the population are non-negotiable. Yet, as intel drive operations, a focus on the enemy to shape our intelligence will never provide a shift to a pop-centric strategy.

Concerning your point about logic and casual linkages first let me say this article was reviewed by complexity experts both in its thesis form and as an article prior to publication and not one had a issue with the hypothesized casual linkages. Complex systems are deterministic (and logical if viewed from the right perspective) although that interaction of variables may produce counterintuitive effects. Understanding which interdependencies have the strongest correlative effect on specific emergent phenomenon is the purpose of much complexity research (ideally this research will also find certain immutable laws of complex systems) Yet, as you pointed out it is exceptionally hard to determine if the hypothesis concerning certain interdependencies and their emergent phenomenon is correct. Particularly as the standard reductionist approach to test the hypothesis don’t work. However, any other strategy, concept, attempt etc is subject to the exact same criticisms. Any effort in Afghanistan is trying to influence a complex adaptive system--- this is a statement of fact. So any other attempt is just a different hypothesis. Which leads to the point about the Afghan govt and its governing institutions. I will predict that the Afghan govt with its current array of interdependencies will at best lead to an autocracy that provides some modicum of stability…at least for awhile (see Arab spring). Does this meet US/ISAF obj? I think that depends on which obj you are talking about.

We are absolutely saying that we believe with this perspective there will be better understanding which will lead to more effective action. I concede that that action will produce its own slew of problems and will absolutely require more modification to our understanding and idea and will not produce a nice clean definitive victory. However, I would argue “we are not advocating using a different tool to get the same problems” we are advocating a different tool to get less severe problems. I also argue that if we could run an effective agent based model comparing competing approach to influence Afghanistan this approach would have a higher probability of success.

Yes, we should be more humble when dealing with complexity and ESA will not solve all the world’s problems. Hopefully it helps decision makers have an understanding of the interdependencies of the population/system they are trying to engage. As we are in a foreign country “to bend others to our will” of the available ways to approach the problem I hypothesize that the better choice is to shift to an interdependent focused strategy. I know of no way to test this or any other strategy, but the concept within it (complex adaptive systems are a result of their interdependencies) has been rigorously tested and we shamelessly stolen it.

Thank you again for your comments.


G Martin

Tue, 11/08/2011 - 5:08pm

Interesting article. My initial impression was negative, since I cringe at anything analytic that has the potential to give our ORSAs and Intel folks a false sense of security when acting in complex environments. I would caveat using this as a tool (among others) to assist in attempting to falsify assumptions as opposed to developing plans with which many will become advocates for and lose all objectivity. As complexity theory grows and tools are experimented with, refined, thrown-out and new ones adapted- we should also seek to see where they may assist us, but become skeptical if anyone offers them as a "silver bullet".

A few specific points:

- I think it is arguable that in 2009 ISAF went to a pop-centric strategy. I have talked to some who have started to study the operational record of McKiernan vice McChrystal/Petraeus and there are indications that we talked pop-centric more but in practice we actually were <em>MORE</em> enemy-centric post-McKiernan.

- There seem to be an awful lot of attempts to show causal linkages between factors. Some complexity theorists would have issue with attempting to draw that many conclusions about a complex system vice submitting them as possibilities. One example:

<em>"...Afghanistan has no effective local tax collection, which means the provincial state has no dependency on the local population. Without a dependency on the local population, the national and provincial governments have little incentive to provide public services to the population..."</em>

and later: <em>"...Government officials are motivated to expand patronage networks to increase power."</em>

This shows a logical linkage between several factors- starting with Afghanistan not having an effective local tax collection and eventually concluding that Government officials are motivated to expand patronage networks- thus corruption is tied logically to a sufficient cause of lacking tax collection capability (unless I read that wrong). Although this may be true, as I understand complexity theory it is very problematic to state this logically AND prove this linkage. We currently have the same problems in Afghanistan: the logic behind our actions is tied to faulty sufficient causal linkage and conventional wisdom/groupthink. Again, instead of stating all of these logic linkages as facts- as opposed to assumptions that must be tested- we fall into the same logic trap we have already fallen into.

<em>"...As the dominant feature of the fitness landscape is the patronage network, and this allows the patron-client agents to flourish it logically follows that the flow of goods, services, wealth and power in Afghanistan follows the patronage networks..."</em>

again, as I understand complexity theory- complex environments rarely show things connecting through any kind of logic. Instead, complex environments are characterized by connections that are not logical, that are not easily identified, and change often.

The final conclusions:

1) <em>"...Intelligence analysis should identify current interdependencies and their associated emergent phenomenon, and then the USG should conduct operations to either enhance or alter them. The emergent phenomena produced by these decisions will then ideally produce population behavior consistent with ISAF objectives. It is critical to identify the flow dynamics across the system, analyze how changes to the dynamic affect entities considered external to the objective system, and monitor internal and external attempts to alter the objective dynamics. Instead of building advanced governing institutions, the purpose is to establish simple and basic interdependencies that grow the desired institutions, which favor specific agent traits..."</em>

This is a very positivist philosophy- similar to what we already have: identify key interdependencies (read: centers of gravity or key nodes, or whatever the latest systems thinking term is en vogue) and then act. As I understand complexity theory, assuming one can even identify these components is difficult in a complex environment, not to mention actually establishing basic interdependencies. And to think that whatever we do establish will be likely to grow desired institutions and favor specific traits, well...

On the practical side of things: I submit that this solution ignores that there is already an Afghan government with governing institutions- advanced or not- and attempting to coax behavior out of people that will further ISAF objectives assumes that the political objective of transition in 2014 is furthered by ISAF objectives. Many would argue that ISAF objectives are not in synch with the political mandates of transition in 2014 or Afghan objectives- both of which will be problematic if we are pursuing the wrong objectives (so, we may have found success in inserting the right interdependencies and great things are "emerging", but when we transition all is for naught. This goes back to the primacy of strategy).

<em>"...Instead of trying to build developed Afghan institutions regardless of the cultural and situational dynamics ISAF elements across the country should adjust local interdependencies based on local conditions. This change will alter the Afghan populace’s decision-making considerations causing them to favor choices compatible with U.S. objective and allowing the Afghans to develop their own governing institutions consistent with their unique and varying cultural dynamics..."</em>

There are issues with applying local solutions everywhere in Afghanistan- not the least of which are that some European and most Afghan central government entities do not support them. But- again we seem to have a positivist prediction: if we do "x", we will get massively complex change (i.e.: "altering Afghan populace's decision-making considerations to be compatible with U.S. objectives"). Instead of seeing this as a viable alternative to what we currently have in Afghanistan- it seems to me to advocate using a different tool to get to the same problems!

<em>"...Decision-makers who adopt a population-centric, complexity –based approach should be more effective in setting the conditions to create a stable and productive Afghanistan, which is intolerant of terrorist safe havens..."</em>

I would caution strongly all who offer solutions to complex problems to be more humble in their statements about success. Every "silver bullet" idea I've seen for the last 10 years proclaims that it will "create a stable and productive Afghanistan intolerant of terrorist safe havens". I would suggest caveating statements like these to be more modest and transparent about the impossibility of simple, step-by-step, if-then statements predicting success and more statements that propose a hypothetical and ways with which to attempt to falsify that hypothetical. Statements like these do tend to get "emergence"- but that emergence is in the form of stakeholders who place value in proving the hypothesis right.

<em>"...analytic efforts and positive actions that can grow a new emergent state of Afghanistan, shaped by mutually supporting actions, across a complex country...</em>

The idea that we can somehow control emergence- or even influence it- shape it- within a complex environment again to me smacks of hubris. It sounds like we are promising too much and that we've already drank the kool-aid. We must remain as objective as we can- not focused on proving our theories were right, but concentrating on coming up with the right theories.

Overall, I'd recommend coming across more modestly, being up-front about possible places we may be wrong, and structure our causal linkage statements to be more like hypotheses- and recommending ways in which we can test these hypotheses.

Thanks for the paper-

Eddie Brown

Mon, 11/07/2011 - 9:23am

In reply to by tompike


Thanks for the feedback and glad some of the article helped. And I couldn't agree more on our oversteps you pointed out. Tom and I continue to read and think about how this new science applies to our profession, and of course our method of analysis follows. As we told a group at OUSD(I) to explain ourselves, "we're hobbyists."

I wish you luck in your efforts to improve thought processes in the field. Many people use the word complex. Fewer seem to be changing to adopt best practices. And as I have found out, some will go as far as to offer one-way conversations.

Thanks again for not letting us off easy.



Fri, 11/04/2011 - 8:16pm

In reply to by Hubba Bubba


As one of the authors, concerning your comments on control I agree (and I am sure Eddie will too). You are absolutely right we should not have used control with complex adaptive systems. We have limited ability, at best, to influence adaptive path of the complex systems.

Concerning patronage networks again, I whole heartedly agree. Transforming is much better. I don't think you ever get rid of that 'patron-client' dynamic (ie people in power dolling out favors) completely. As a case in point, and at the risk of sounding heretical, you can see that dynamic at play in our government (in my estimation pork is not a fundamentally different dynamic). This, from my understanding, is quintessential complexity, there a no clear casual linkages in the system dynamics instead there is a subtlety of interdependencies that produce behavior. Unfortunately, I think these "soft reasons" are a hard pill to swallow for human beings in general and is often seen as nay saying. To finish the example, accepting that America has some similar dynamics as Afghan patronage networks is not saying that the American system and the Afghan systems are the same (my comments interpreted at their most ridiculous extreme). Instead, it is saying that certain dynamics emerge in all human interactions. Accepting this allows us to have better understanding and analysis in order to take more effective action.

Hubba, thank you very much for your comments. We will absolutely refine our choice of words. Complex adaptive systems can be influenced but they can never be controlled. Our entire scope of diplomatic, NGO, military power etc may influence the adaptive path of a people but it will never dictate it. (I suppose this is intuitively obvious to anyone with kids)


Hubba Bubba

Mon, 10/31/2011 - 6:31am

Forgive my tardiness on this one; I have enjoyed this article for a month now and got to apply some of it's principles in some recent endeavors that relate to design and military planning.

It is interesting that even within the Intelligence community, there is in-fighting and folks at logger-heads over terminology, methodology, and what is "on-target with our assessment" and what is not. In fact, I used the 'patron-client-non client' model in one such engagement with another intelligence organization that wears Army uniforms, and they wrinkled their noses at any deviation from their beloved "criminal-patronage network" concept. Essentially, if it differs from the group think, it is not considered or discussed. This is tragic, because this article represents some of the very useful discussion that occurs on SWJ. Critical thinking, creative thinking, and the free discussion of ideas. It serves no purpose to simply recycle doctrine and existing ideas (or old ones) that function as an echo chamber for the organization that seeks to continue it's merry existance, even at the expense of national objectives. This article is great because it challenges thinking- but the intelligence community may not be entirely receptive to this right now. Some are; many I encounter are not- in an almost openly hostile way. The world is what we want it to be, instead of what it really is...

I would venture that the patron-client phenomenon discussed here may follow a complex adaptive system process where feedback loops and swarming processes influence the evolution of the system...

I do disagree with page 8 where the authors contend that they might "break the patronage network over the long-term"- this has echoes of 'Effects Based Operations" thinking that has long infested the intelligence community's logic on interpreting the world. We cannot master a complex system, and reduce it down enought to "break" it. Breaking implies control- complex systems resist control; and presistent phenomenon like the patronage network are like narco-syndicates and criminal enterprises- they adapt and can wipe out an entire cartel or network, but you do not destroy the idea. It persists, and manifests in a new and likely more effective form. I would propose the authors consider "transforming the patronage network" or "influencing" it into a future state where the conditions are supportive towards our interests. Essentially, the mafia still functions, but may retreat into a penal system gang network and gambling enterprise instead of doing shake-downs along Main Street and bribing local cops.

P9 continues the dissillusion that one might "control" the complex system; the authors state that "emergent phenomena produced by these decisions will then ideally produce population behavior consistant with ISAF objectives." That is, in a nut-shell, delusional. Complex systems are, as the title of this article states, 'adaptive.' This means you cannot train them like a dog. They are more like a zebra versus a horse. Nobody rides a zebra because they do not conform to that sort of controlled relationship. The intelligence community needs to stop trying to tame zebras and recognize that while a horse can be broken, sometimes you need to handle the zebra by moving it's source of food, water, and predator relationship to influence it's behavior.

regardless of my last few comments on complexity, this was overall a great article and worthy of being discussed in planning groups now. I did, and found it quite 'value-added.'

Just blowing bubbles-