The Odd Sheikh Out: A Complex Problem
The imminent publication of former Secretary of Defense Robert Gates’ harshly critical memoir is likely to trigger another around of recriminations about the prosecution of the wars in Iraq and Afghanistan and the preservation of the peace. While specific explanations for failure range from flawed policies to civil-military squabbling, we can decompose them into political scientist Scott Page’s conception of uncertainty, difficulty, and complexity as types of social problems. While undoubtedly uncertainty and difficulty are a large part of the problem, it is complexity that really has plagued American war-fighting the most. Any future policy or strategy should be interrogated by what assumptions it makes about all three characteristics of the problem, and which characteristic the American design assumes will be the dominant barrier to victory.
Uncertainty concerns an actor’s belief about the fixed payoff values of a set of choices. Assuming he knew the payoff for every choice he considers, he could simultaneously evaluate them and pick the most optimal decision. Difficulty is a measure of how large the fixed solution space of a given problem is – and the cost of different ways of searching the solution space for the best choice. Cost can be measured in both resources and time – one might have the best solution, but it could cost too much to discover and the right answer may take an eternity to find.
Complexity arises from subtle alterations to the characteristics of uncertainty and difficulty. One can consider uncertainty and difficulty purely by the question of whether or not the decision-maker is optimizing or rule-based. An optimizing decision-maker chooses the best solution given the constraints. A rule-based decision-maker, in contrast, follows a recipe for making decisions. As the size of the solution space to be searched grows, search rules that scan the entire space of solutions become prohibitively expensive.
Recently, it has become popular to use shortcuts that make very few assumptions about the nature of the problem and thus apply to a variety of situations. These search strategies find a “good enough” solution and are particularly useful when the analyst does not know what an optimal solution to the problem looks like or how to go about finding it. With at least some knowledge about the problem that can allow them to tell good and bad solutions apart, the analyst can use these methods to iteratively improve solutions that work and discard ones that do not.
Complexity, however, arises from an often-heterogenous group of rule-based decisionmakers, This introduces several enormous problems. First, everyone’s payoff depends on choices taken by everyone else. Second, no one can be assumed to choose the most optimal payoff – only that which is dictated by the conditions of their decision rule. Members of the group often have different decision rules. History matters, so decisions made earlier can evolve the structure of the solution landscape. What was once a peak value at one point is now a valley. Finally, complexity generates unpredictable behavior over time. Multiple points of equilibria for the social system exist, and systems can cycle.
The very abstract (and often mathematical) nature of such ideas makes them seem forbidding to practitioners and more qualitatively-oriented researchers. Hence they are either dismissed outright or discussed at a superficial level of generality. A short parable (consider it War on the Rocks’ version of the Grimms’ fairy tales) can illustrate the real-world implications of these distinctions.
Suppose an American military unit is patrolling a village governed by Sheikh al-Evans of the Cask Strength tribe, which dominates the misty and peaty heartland of the war-torn state Laphroaigistan. American troops currently occupy Laphroaigistan, and are struggling to deal with a tenacious insurgency. Stabilization must be carried out village by village, and the key to stabilizing this village is al-Evans. A gregarious man known for well-attended happy hours and eloquent discussions of classical realism, Sheikh al-Evans is the village’s political and military authority. However, the unit has heard rumors that he is a prominent supporter of the insurgency.
At first blush the choices available to the unit seem limited and obvious. Either the sheikh is on their side or he supports the enemy. Stabilizing the village either involves supporting the sheikh’s men or moving against him. Treating the problem as one of uncertainty, intelligence assets are deployed to suss out which team Sheikh al-Evans is on. Unfortunately, reducing uncertainty to the degree that intelligence allows reveals that the village stabilization problem is actually mostly characterized by difficulty.
The sheikh is on his own side. Intelligence reveals that the sheikh seems to back whoever is strongest, and since the unit has failed to suppress the insurgents in the village he will back Americans when they are visible and present and insurgents when they are not. Due to restrictive rules of engagement, the unit cannot move against the sheikh unless he is confirmed to be a “member” of the insurgent Laphroagi Liberation Army (LLA) – a meaningless stipulation given he is clearly a fence-sitter who alternates his support between the two sides.
So now the potential solutions to be evaluated range far beyond simply “support or depose the sheikh.” They have to find some way to convince the sheikh to cooperate with them – which requires tackling expansive questions of village development and counterinsurgency tactics. There is now a large solution space that the unit has difficulty searching through with its existing resources and its time in village before being relieved by a green unit. Frustratingly enough, the unit has no idea of what an optimal solution would look like, or how to go about finding it.
The general problem of the influential al-Evans’s influential Cask Strength tribe triggers a high-level meeting by the International Laphroaigistan Assistance Force (ILAF)’s high command. The high command agrees that the problem cries out for a comprehensive politico-military effort to develop the war-torn state’s institutions, relieve its material burdens, and counter the insurgency in every village where a sheikh sits on the fence. They assume that all sheikhs share the same preferences, so what solves the al-Evans problem in one village can scale to every other village. Ignoring country experts’ warnings about the diversity of the Cask Strength tribe, they begin a detailed process for reducing the difficulty of the strategic problem. In order to carry out the comprehensive effort, they must make decisions about resource allocation, political-military-economic planning, and organizational structures and relationships.
At first ILAF carry out top-down planning, but their formal and rigid planning process – in which every possibility is calculated, wargamed, and fed through bureaucracy – is slow and inefficient relative to its resource costs in both staff time and contractor billable hours. Meanwhile American troops are being blown up by fiendish enemy IEBs (improvised explosive beverages) and everyone from Congressmen to critical bloggers questions the war effort. And like the troubled unit, ILAF lacks both a firm idea of what an acceptable solution would look like and a non-exhaustive way to search for it. They have at least some idea of how to distinguish a good idea from bad outcomes gleaned from historical study of insurgency, but not much else.
A new general is appointed who believes that the best solutions will come from the troops in the field. He stops ILAF from commissioning studies, punching scenarios into high-powered computers, or running through high-level planning. Instead, he disseminates rules of thumb for coping with insurgency along with a detailed explanation of the Sheikh al-Evans problem across ILAF. He believes that a “good enough” solution can arise from the bottom up. Promising ideas will be rewarded and bad ideas will be discarded, with the goal being a situation in which good solutions become more and more alike and bad solutions drop out of the picture entirely – thus pointing the way towards a “good enough” national-level solution. Ideas are being tried without a central logic, but with the goal of climbing the hill of quality towards a peak in the solution landscape.
Though promising at first, the ILAF effort stalls. Through a process of trial, error, and strategic learning the ILAF standardizes its procedures. But even though ILAF’s increased capacity for adaptation should have reduced the difficulty of search, it nonetheless keeps getting stuck on local optima. The problem becomes apparent when an Alcoholic Terrain Team (ATT) report reveals that the problem – while involving difficulty and uncertainty – is principally one of complexity.
The ILAF had treated the Cask Strength tribe as a homogenous mass that all – like the aforementioned village master at the beginning – wanted greater security and government authority. It was this basic similarity that was supposed to help the ILAF reduce its search cost – given the same nation-wide problem, it should have only been a matter of time before a solution that worked in one place could work everywhere. But the Cask Strength tribe is only united in name. In reality different sheikhs in the tribe have strikingly differing criteria for deciding whether or not to cooperate with ILAF.
Though ostensibly sharing a common ancestry, different subsets of the tribe occupy different ladders of its hierarchy and serve different functions. Al-Evans himself merely seeks to protect his village and seems most attuned to who is most powerful. Other Cask Strength sheikhs may have different objectives, such as drinking their scotch with a dash of water or with whiskey rocks. Some of the more heterodox sheikhs prefer the Smoking Gun cocktail (2 oz Laphroaig, 3/4 tsp Fernet Branca, 1/2 tsp brown sugar cordial, 2 dashes Fee Brother’s whisky barrel bitters, mint leaf for garnish). There is even a lesser sub-tribe that mixes their single malt with soda water, inspiring enmity from the rest of the Cask Strength tribe.
Making matters worse, each Cask Strength sheikh’s decision is influenced by every other Cask Strength sheikh’s decision. Using the history of past recent aggregate tribal decisions, each sheikh tries to anticipate what the aggregate decision will be to avoid being the odd sheikh out. The ILAF assumption was that if they could convince enough sheikhs that they were in control of the villages, the aggregate tribal choice would be predictable. After achieving control of enough villages to gain a bare majority of sheikh cooperation, recalcitrant sheikhs would change their votes to avoid being left out and supporting sheikhs would keep cooperating. But a solution that achieved village control (and thus pacifying one sheikh) might prevent another sheikh from enjoying their Smoking Gun Cocktail or earning a living off the illicit narcotics trade in single malt with soda water. The more things of value to the Cask Strength that a comprehensive political-military effort changed or altered, the greater the potential for displeasing a Cask Strength member.
Because of this complication, the ILAF realized something deeply unsettling about the entire problem. The sheikhs were making choices each according to a decision rule, rather than simply evaluating decisions based on what was optimal. Because each sheikh’s rules were different and each sheikh used the last aggregate tribal decision to predict what they should do in the next one, the tribal system’s aggregate result at any given point of time was essentially unpredictable. Making matters worse, the ILAF’s idea of dramatic, multi-layer changes in the fabric of Laphroagi society was making an already unpredictable tribal system more fractious and dangerous. And even a solution that had stood the test of time on multiple Laphroagi battlefields was still nothing more than a shot in the dark.
Chastened and now regarding the problem as principally one of complexity, the ILAF called area experts on Laphroagi tribal politics back in. A professor of Laphroagi Studies, accepting a Smoking Gun cocktail as payment, reported that over time he had noticed some qualitative tendencies on which the council had eventually converged based on different historical circumstances. He counseled that the most successful Laphroagi rulers considered the complexity of getting all of the Cask Tribe to cooperate, and often made very minimal demands on the Laphroagi political system. By this point, it was too late. The President had already committed to a “Laphroagization” program to withdraw U.S. troops – leaving the civil war to rage afterwards.
The point of this intentionally absurd scenario is to show the importance of understanding the dominant characteristics of a strategic problem. Certainly elements of uncertainty plagued the ILAF. But at a certain point investing resources in intelligence-gathering would produce diminishing returns. The issue was not really just that the values of decision payoffs were uncertain. Rather, it was that there were too many potential solutions to a problem of such size and an efficient way of searching the solution landscape had to be devised. This also eventually hit diminishing returns because the heterogeneity of the Cask Strength tribe meant that any one solution could not be guaranteed to hold true over time. The solution landscape was evolving. What was a good idea at time t seemed to be suboptimal at t+1, because its underlying assumptions about the peaks and valleys of the landscape were invalid when the situation changed.
Had the ILAF recognized at the beginning that they should have modeled the “Odd Sheikh Out” problem as complex, it would have consulted the area experts in the beginning. It might have given up on the idea of a complicated and rigid solution that forced the Cask Strength tribe to a deterministic “end state.” Instead, it might have looked at potential solutions that might have created a steady state of agreeable outcomes – created by simple adjustments to the war torn country rather than a comprehensive political-military effort with far more unpredictable effects.
The ILAF, Sheikh al-Evans, and the Cask Strength tribe are all fictional. But the problems the example represents are real. Strategies and policies ought to be evaluated by the assumptions they make about uncertainty, difficulty, and complexity. And most importantly, the proposed idea also ought to be evaluated as to what kind of strategic problem it assumes deserves the most attention and resources. The answer will often vary – but too often complexity is dismissed or ignored altogether, with material and human costs that even the most well-prepared whiskey rocks can’t wash away.
Adam Elkus is a PhD student in Computational Social Science at George Mason University and a columnist at War on the Rocks. He has published articles on defense, international security, and technology at CTOVision, The Atlantic, the West Point Combating Terrorism Center’s Sentinel, and Foreign Policy.
Photo credit: isafmedia (adapted by War on the Rocks)