Every week, the Islamic State (IS) makes further headlines with its ruthless behavior. Beheadings, mass executions, burnings and extreme acts of brutality are the methods of a terrorist campaign intended to cow opponents and rally potential fighters. At the same time, the group is fighting a guerilla war against Iraqi forces while engaging in more conventional infantry battles against Kurdish Peshmerga and Free Syrian Army cadres. The many tactics of IS raises the question: which type of war are we fighting against it?

President Barack Obama recently announced his strategy for stopping IS militants, presenting a draft authorization of military force. The debates about his request have used the language of traditional war – which take into account clear geographic boundaries and fully committed allies. This is misguided. There is no single front in this war and the level of our allies’ commitment to the effort varies from place to place. The fight against IS is, in fact, several different wars against the same enemy, with porous fronts that only occasionally resemble contiguous state boundaries.

Therefore, finding the right strategy will require working with our allies in the Middle East to define their objectives and limitations—including an honest assessment of which territories matter enough for them to commit troops on the ground and eventually govern—and build our policy accordingly.

(Recommended: Five American Weapons of War ISIS Should Fear)

But doing this is hard because some of our basic notions about war mislead us when examining the conflict with IS. Our world map, with its thick black lines winding between countries, was created by wars fought between nations, wars that were usually won by the side with superior weapons and larger armies, who then went on to exert full control over the territory they conquered. So our cultural conception of war – its dynamics and the ways to prepare for and win it – is based in large part on those clashes, with the two World Wars casting a long shadow.

Such a conception of war is historically accurate but dated and does us a disservice when we approach most current conflicts, including the one against IS. Fronts in this conflict are porous and shifting and run along a complex web of political fault lines that only occasionally resemble contiguous state boundaries.

(Recommended: Is China About to Declare War Against ISIS?)

Applying analogies from nation-versus-nation wars can lead us to overlook critical constraints on what the United States can accomplish against IS, constraints that stem not from our vast physical capacity to wage conventional war, but from the interaction of two factors: the essential role played by local allies and the sometimes limited objectives of those allies.

For the last five years, our multi-university research collective, the Empirical Studies of Conflict Project (ESOC), has worked to understand the complexities of subnational conflict, both on the ground and academically. We have assembled micro-level conflict data from nine different countries on political violence and are now working on several other countries. We map territory and study conflict at a very local level, and combine a rich appreciation of local politics with economic theory and large datasets to generate and test clear predictions. We are interested in how policy choices influence trends and outcomes within wars. In short, we bring data driven evidence to hotly contested arguments that have typically been adjudicated on the basis of anecdote.

(Recommended: Martyrs Wanted: ISIS' Devastating Defector Problem)

One outcome of our research is that we can say a good deal about what outsiders can do to help in suppressing threats emanating from IS in Iraq and Syria. ESOC models suggest multiple important steps the United States and its allies can take to create better strategies.

Subnational conflict: Symmetry, Asymmetry and Information

Subnational conflict is not new; there have always been wars that do not operate according to a nation-versus-nation dynamic. For example, in Napoleon’s struggle to control the Iberian Peninsula, he didn’t face one central opponent. Instead, he fought many “little wars” or guerilla wars, as the Spanish called them. Nearly a century later, after Spain ceded the Philippines to the United States, the United States waged a three-year war within this newly acquired territory against multiple insurgent groups. Although this conflict officially ended in victory in 1902, there was sporadic violence for decades afterwards. On the Eastern Front during World War II, Hitler’s armies fought various insurgencies including the Yugoslav Partisans, the Polish Underground State, and the Ukrainian Insurgent Army, which also fought the Soviets.

(Recommended: ISIS vs. Al Qaeda: Jihadism’s Global Civil War)

Recent research by Stathis Kalyvas of Yale University and Laia Balcells of Duke emphasizes the distinctions between symmetric and asymmetric subnational conflicts. Symmetric civil wars involve protagonists of roughly equal capacity and are fought mostly over territorial control. In the latter stages of the Vietnam War, for instance, combatants fought along well-defined fronts as in international wars, with victory usually secured by a combination of superior weaponry, numbers and strategy.

But symmetric conflicts account for a minority -- 46 percent according to Kalyvas and Balcells -- of subnational conflicts. Every war the United States has fought in since the Korean War, except for the first Gulf War, has been an asymmetric, subnational conflict. This will likely remain the case for the foreseeable future due to American technological and geopolitical superiority. In asymmetric wars, the struggle is not over territory. The stronger party can seize territory whenever it wishes to do so. Rather, the struggle is over holding and administering territory.

One enormous distinction is that while civilians are not that relevant for victory in a symmetric conflict, they are vital in asymmetric conflicts.

Consider Iraq from 2004 through 2010, when Coalition forces and their Iraqi allies had massively superior conventional, military capacity. Whenever insurgents could enlist the support of the population, they could move forces, acquire weapons and conduct attacks with roadside bombs and other improvised devices. This prevented the Iraqi government from consolidating control. However, whenever civilians accurately cooperated with governments in identifying insurgents, it was relatively easy for the Coalition and Iraqi government to act on this information. In such cases, victory requires a flow of accurate information provided by civilians.

The primacy of information is one lesson we have learned from our research at ESOC. A second important lesson is that in asymmetric wars, small shifts in the flow of information can achieve significant results. These shifts are triggered by small day-to-day, operational changes and not massive political events.

Game theory gives us a way to model complex interactions, assign different motivations to separate actors, and then test predictions using real-world data – methods that have proved useful in disciplines from economics to computer science to evolutionary biology. In a 2011 paper, we constructed a model of asymmetric conflict with three players: rebels, community and government. The government seeks to reduce rebel-initiated violence through force and by providing services that win civilian loyalty. Rebels seek to impose costs on government by attacking it. The community seeks to maximize its own well-being either by sharing information about rebels with authorities – thereby helping deliver territorial control to government – or by refraining from doing so.

This model predicts that insurgents will produce violence up to the level where the community suffers so much that it then decides to align with the government. We have seen this played out in reality.

During a raid near the Iraq-Syria border in 2007, Coalition forces uncovered a trove of internal documents revealing the local strategy of al-Qaeda in Iraq (AQI). The documents clearly portray an organization trying to determine the limits of violence that the community will tolerate when planning its attacks. After an AQI operative assassinated local tribal leader Sheikh Nasr al-Fahadawi, leaders issued a memo with instructions to kill more carefully lest their assassination campaign provoke resistance. The memo advised AQI members to “leave no evidence after the killing and bury [the victims] in far-away places and not like what we did with Nasr al-Fahadawi in front of the people.” Additionally, a 2004 AQI announcement attempted to curb looting from locals with a vow to “expel those who allow themselves to exploit the blood or the money of the Muslims.” This and other announcements showed AQI’s willingness to give up a measure of secrecy and security in order to win over a skeptical public.

Our model generates a number of hypotheses about how to improve information flow. The focus of our 2011 paper was aid spending – whether, how, and in what amount aid can win the “hearts and minds” of civilians, as has been the aim of many campaigns since the Cold War. Previous studies have found that, broadly speaking, aid may stoke insurgent violence. However, extremely detailed administrative data recorded by the U.S. government and allies and reconstructions of spending in Iraq show that smaller projects delivering targeted public goods to communities reduced insurgent violence. This has been especially true since the “surge” began in 2007, because it increased the ground presence of development experts. While these projects were too modest to have large effects on the level of employment, they could have well increased information flow. A key implication of our model was validated: low-profile aid, targeted so as to lessen the odds it will be captured by rebels, can push the community over the tipping point into sharing information with the government.

In later papers, we expanded our analysis to include the extension of mobile networks in an area. This allowed us to examine the question of how mobile communications impact violence: do they allow insurgents to organize and increase violence, or do they allow civilian informants to communicate tips to government, thus decreasing violence? It turns out that in the asymmetric conflict in Iraq, expanding cell phone coverage did reduce violence. Importantly, our findings suggest that the mere threat of information sharing by noncombatants can reduce violence.

Other research, using a variety of different methodologies, supports our model and adds nuance to our findings. In a 2012 study, we used surveys in Pakistan to show that, contrary to popular conception, the poor are more likely to disapprove of insurgents because they are more likely to suffer from insurgent violence. Furthermore, in a 2014 paper, we used large-scale experiments to show that the mere perception of poverty and violence can change civilian support for insurgents.

In summary, smaller, smarter aid packages can reduce violence. So can the threat of information-sharing. Shifts in civilians’ perceptions of violence, and poverty can alter their support for militant groups, with potential implications for information sharing. All of this points to the fact that, in asymmetric conflicts, small-scale and often low-cost interventions can yield big results – information control being the key.

Multiple wars against a single enemy

So, is the current fight against Islamic State symmetric or asymmetric? The answer is “both.” The involvement of the United States in aiding the Iraqi government and Kurdish forces is spurring the evolution of the conflict in an asymmetric direction. If the war continues along this trajectory, then it will be crucial for the Iraqi government and its allies to make political accommodations with communities to increase the flow of information. Additionally, aid packages should be low-profile, informed by development experts and conditional on communities cooperating with the government. Iraqi forces should follow the model of community interaction exemplified by the “surge” of 2007, and the Iraqi government needs to find a way of convincing skeptical Sunnis that they have a future in the country, as the United States convinced Sunni tribal leaders in 2006 and 2007.

In Syria, by contrast, the atrocities of the Assad regime have precluded the kinds of international intervention and support to the government that would in turn allow it to fight a predominantly asymmetric conflict. Thus, while the Assad regime has fought its multiple enemies to an asymmetric stalemate -- none of them can overcome Assad’s army -- a symmetric conflict is unfolding between the Islamic State and the Free Syrian Army (FSA). This conflict features a familiar volleying for power along front lines. We know how to provide support in such a conflict: firepower – which can be provided from the air – is critical; information from non-combatants, less so, though it is still obviously useful.

The picture in Syria becomes more complex, however, when we consider the FSA’s struggle against a dominant government army, especially in cities. Our model provides us with recommendations for the kind of war the disadvantaged side should wage in an asymmetric conflict. It should apply as much violence as the public will tolerate. This is a thorny proposition, given the catastrophic results of the deterioration of the Syrian state. It is important to recall the end result of United States' support for Afghan rebels in its asymmetric war against the Soviet Union.

Identifying the areas where the fight is symmetric or asymmetric is important not only for strategy, but also for resource allocation. Applying data from past conflicts to our model reveals the margins where small expenditures in aid and information-gathering can achieve big results. Our results suggest that taking a blanket approach will be counterproductive. Instead, we should think of fighting different types of wars in different areas areas of conflict. The more asymmetric a war gets, the more important the flow of information becomes.

Commitment and shades on the map

In subnational wars, it is critical to understand not only the nature of the conflict but also the level of the local ally’s commitment. The hard truth is that our allies are sometimes unwilling to pay the price of exercising complete control over their territory. The black lines on our world map give a false impression of  territorial solidity. Not only are the majority of the world’s nations involved in one or another kind of border dispute – approximately 70 percent of the countries with significant military forces – many contain areas that their governments do not control. This is true, for example, of the Federally Administered Tribal Areas (FATA) which the Pakistani government has long given over to local tribes. It is also true of large chunks of Yemen with its patchwork government, and even pockets of Central and South American nations effectively ceded to drug cartels. Governments do not exercise control over these areas from lack of capacity or lack of will – the financial or political returns of exerting control over marginal territory is often low, even when these territories cause real problems for the rest of the world.

Therefore, we must ask the question of whether in a subnational conflict, the government is bringing all available resources to bear in order to win, or whether it simply does not feel an area is worth fighting for. This adds a further level of nuance to our predictions regarding asymmetric warfare. Perhaps the local ally is willing to conduct incursions, but not to provide sufficient governance to ensure a flow of tactically critical information from civilians. In Iraq, we have observed that the government does not try that hard to assert control over certain areas; it ceded much of the Sunni areas of Iraq very quickly to the Islamic State despite a superiority in numbers and military technology. The United States assisted Iraq in waging a successful asymmetric war to gain control over the western province of Anbar in 2007, only to later find that the Iraqi government lacked the will to govern that area. Anbar is now an Islamic State stronghold.

Therefore, in order to wage an effective war against the Islamic State, the United States should sit down with allies active in Iraq and Syria and identify territory that matters enough for them to commit troops to and eventually govern. Such a conflict will not feature actors controlling blocks of territory but chains of outposts, and will be characterized by various levels of symmetry and asymmetry. However, there will be areas, especially in Syria, that neither the Assad government nor the FSA will want to contest, and there is little we can do about that.

That brings us to the difficult issue of ungoverned spaces, which are uncontested by any state or subnational ally. Neither a symmetric or asymmetric war can be waged effectively in an uncontested area. These areas are especially dangerous because they could incubate the greatest terrorist threats, dangerous to Paris as well as Baghdad.

The United States has two constructive options to deal with such spaces. First, we can give allies reasons to police them, through incentives, through development programs that make them attractive, or through occasionally helping nudge noncompliant leaders such as Prime Minister Maliki towards the exit. Doing so is quite challenging, as Pakistan’s case demonstrates. Failing this strategy, a second option is to directly suppress the terrorist threat. The United States has strategies for this kind of warfare – developed in Pakistan and Afghanistan. They involve the unilateral employment of drones, special forces, and signal intelligence. A similar strategy may be unavoidable in some areas of Syria.

The commitment of allies is also important. The United States will remain dependent on how well its European allies, for example, control the radicalization of their Muslim populations.

The bottom line is this: we have long recognized that effective counterterrorism requires the cooperation of countries who are willing to governing all their citizens and territory. We must also recognize the limitations of working through countries and allies with more opaque and limited objectives, such as Iraq, the FSA and the Peshmerga.

Failing to recognize the type of war one is in can lead to massive waste and unrealistic expectations, as can failing to accurately diagnose the reasons for ungoverned space. The war against the Islamic State is not an old school war involving states contesting territory. Instead, the United States is fighting several very different wars against the same enemy: symmetric, asymmetric and a mix of the two as part of its efforts to combat terrorist threats. The first steps toward victory may involve creating a detailed, dynamic subnational map of which war is going on where, setting feasible expectations, and organizing our allies appropriately to deal with different types of conflicts in different ways using appropriate evidence-based strategies.

Eli Berman is a research director at the University of California Institute on Global Conflict and Cooperation and professor of economics at the University of California, San Diego.

Joseph H. Felter is a senior research scholar at the Center for International Security and Cooperation at Stanford University.

Jacob H. Shapiro is associate professor of politics and international affairs at the Woodrow Wilson School of Public and International Affairs at Princeton University.

Image:Flickr/Defense Department​