Max Abrahms is an expert on asymmetric warfare. He specializes in studying the dynamics and motivations of militant organizations and individuals that engage in terrorism. In “Leadership Matters: The Effects of Targeted Killings on Militant Groups Tactics,” a forthcoming paper in Terrorism and Political Violence, Abrahms and co-author Jochen Mierau argue that targeted killings of militant leaders can often result in the promotion of lower level militants to positions of responsibility, introducing new behaviors and preferences into the organization. Consequently, militant groups may gravitate more toward attacking softer civilian targets than hardened military targets. To test this theory, Abrahms and Mierau examine the New America Foundation’s statistics on U.S. drone strikes in Pakistan and a dataset of Israeli targeted killings published by Asaf Zussman and Noam Zussman.
Although the American targeted killing program in Pakistan and Yemen has long received a great deal of attention from academics and journalists, there are many aspects—such as the targeting process—of this covert campaign that remain vague and inconclusive. Likewise, the secretive nature of militant organizations makes it difficult to understand how an alteration in the chain of command can have tactical and strategic consequences. The goal of the “Leadership Matters” paper, explains Abrahms, is to understand the changes that these strikes can introduce in an organization.
Dr. Max Abrahms is an assistant professor of political science at Northeastern University and a term member at the Council on Foreign Relations.
Interview by Dan Gettinger
Center for the Study of the Drone Why did you decide to study the decapitation strategy?
Max Abrahms I come at the examination of drones and decapitation strikes from a very different perspective from most researchers. Most researchers totally understandably go into the study of these things because they’ve become the cornerstone of our counterterrorism strategy. Researchers and practitioners alike want to know whether decapitation—whether it’s from drones or from other means—is effective. There’s no consensus how to measure efficacy; it’s been measured in all sorts of ways in terms of whether the attacks expedite the demise of the group or reduce its ability to generate violence in the short term. All of these questions are important and I eagerly read that research, too.
For a long time, I’ve been interested in the question of “why do militant groups attack civilian targets in particular.” There’s no consensus over the definition of terrorism but the one that is generally used by social scientists—especially in North America—is a three-part criteria involving a non-state actor attacking some sort of civilian target for some kind of a presumed political goal. The conventional wisdom among political scientists is that terrorism is strategically rational behavior—that terrorists groups are rational political actors.
But whenever I looked at targeted countries having their civilians blown up, what I saw was governments getting very angry, moving to the political right, becoming very mobilized politically and militarily, and trying to crush the perpetrators. Very seldom did I see governments become more dovish. A lot of my earlier empirical work focused on testing whether or not it is indeed true that indiscriminate violence against civilian targets is strategically effective for terrorist groups. My main empirical finding is that the strategic value of non-state attacks depends to a very large extent on their target selection. That not all kinds of attacks are equally useful politically from the vantage of militant groups, that there’s variation in the quality depending on the target and that in particular, indiscriminate violence against civilian targets are more politically risky for the group than more selective violence against military targets.
If the strategic model—that civilian targeting is strategic behavior for militant groups—is wrong, then this raises what I call “the puzzle of terrorism.” And the puzzle is “why do militant groups attack civilian targets if doing so carries high political risks.” The theory that I have proposed is that to a large extent, whether a militant group wants to attack civilian targets depends on whether that group is led by a weak leader because there is an inverse relationship between one’s position within the organizational hierarchy and one’s incentives for harming civilians. That is to say, lower level members are more interested in attacking civilian targets because these members at the low end of the organization have stronger incentives to harm civilians. I think that lower level members have strong incentives to harm civilians for many many reasons. One is that it helps them to rise up through the organization by demonstrating their bona fides. Like Jihadi John, for example: he would be unknown to Baghdadi had he not become so notorious for chopping off his captives’ heads. My understanding is that, although he may want to leave Syria, because of his misdeeds he eventually rose up and became the leader of the foreign jihadis.
The theory that I have proposed is that to a large extent, whether a militant group wants to attack civilian targets depends on whether that group is led by a weak leader because there is an inverse relationship between one’s position within the organizational hierarchy and one’s incentives for harming civilians.
I think that leaders are also less interested in attacking civilians than lower level members because they have more organizational resources to draw upon; they are in a position to mount more sophisticated operations against hardened targets, whereas lower level members are relatively starved of resources. If they want to commit an attack, they gravitate to softer targets. There are all sorts of reasons. I’ll list one more. Lower level members are more likely to be driven by emotional impulses rather than by strategic thinking like the leadership. For example, al-Nusra recently went into a Druze area and one of the Nusra soldiers was shot by a Druze guy and the Nusra soldiers were really pissed off. They sought revenge on the Druze population and then the Nusra leadership very quickly reprimanded the foot soldiers and said, essentially, “Use your head. This kind of violence is very counter-productive for us.”
If indeed it’s true—this is an interesting theory—that civilian targeting is driven by the tactical empowerment of lower level members, then perhaps we can find evidence for this theory by looking at decapitation strikes. After all, the point of decapitation strikes is to weaken the leadership to essentially hand over the reigns within the organization to lower level members who are less strategic, less smart, less savvy, etc. What I predicted is that decapitation strikes should induce a change in militant group target selection. As the quality of people within the organization is eroded through decapitation strikes, so too should the quality of the organization’s violence. In particular, in the immediate aftermath of operationally successful decapitation strikes, militant groups should be more likely to engage in indiscriminate violence against civilian targets. That’s really what got me thinking about doing the tests in the terrorism and political violence article.
Drone Is the decapitation strategy different from the campaign of targeted killings carried out by the U.S. in Pakistan and Yemen?
Abrahms Essentially drones are an instrument of decapitation strikes. Decapitation strikes—sometimes called targeted killing—have been around for probably a millennia, a long time. But they really got a push for a number of reasons. One is advances in drone technology. Another is the greater recognition that non-state actors pose a strategic threat to states. And another is the diffusion of these threats all over the world; it simply became impractical to deploy troops in all of these different theaters. Since 2008-ish, there has been a rapid rise in decapitation strikes, especially through drones. My understanding is that the selectivity of this targeting has changed over time. Whereas initially it was designed to strike mainly the leaders, over time, we’ve gotten less fussy about legitimate targets and now drones engage in signature strikes against even low-level militants. One of the many methodological challenges of the paper was to basically truncate this sample of targeted killing attempts so that we were really looking at a sample of just decapitation strikes.
As the quality of people within the organization is eroded through decapitation strikes, so too should the quality of the organization’s violence.
Drone We published a piece a few months ago in which we also tried to sort through the drone strikes to find some that had killed leaders. We found that it could be difficult to decide when a strike had killed someone who actually wielded power within the organization.
Abrahms I think that is definitely a weakness of the paper. There was definitely a certain amount of guesswork that went into that. However, if my theory is correct, we should really find evidence of a change in target selection even when the instrument of choice is not a drone. We also used the Zussman and Zussman dataset which focuses only on decapitation strikes against Palestinian terrorist leaders. So I think that’s helpful in avoiding the problem that a lot of these drone strikes are not necessarily directed against leaders.
Drone There appears to be both strategic and tactical consequences of drone strikes.
Abrahms A lot of the research on drones has focused on whether they are strategically effective against militant groups. What we focus on is how does leadership decapitation change the tactical choices of militant groups. What we essentially posit is that lower level members of the organization are more inclined to engage in civilian targeting than the leadership. We do find evidence for this. In the two-week window after an operationally successful decapitation strike, militant group violence tends to become more indiscriminate by favoring civilian targets over military ones.
Initially I was not going to say anything at all in the paper whether this tactical change was strategically effective because I wanted to underscore how our study was different. There have been many other good studies about the strategic effectiveness of drones, work like Patrick Johnston’s for instance or Jenna Jordan—even though they don’t agree with each other they are both illuminating. And then some reviewers said, ‘well jeez, if there is a tactical change in the militant group violence, then surely this affects the militant groups on a strategic level as well.’ And I suppose that is true. That if not all violence is equal, and that if civilian targeting tends to carry risks for the group, and if decapitation strikes promotes civilian targeting, then decapitation strikes at least in the long term might indirectly be strategically useful by essentially isolating the group from the population.
Many groups, it should be pointed out, have imploded historically because their target selection was too indiscriminate. A very salient example is al-Qaeda in Iraq, which really just went ballistic, shooting anything that moved and that ultimately led to the group’s demise. I think a lot of people today are overstating the strategy of Islamic State. Islamic State is also using violence indiscriminately and of course bragging about it over social media. And there’s a sense I think that basically the international community is helpless to contend with this kind of strategy because the more brutal the violence used by Islamic State, the more members the group will ultimately attract. And I actually disagree with that; I think it’s an ahistorical understanding of militant groups because generally speaking, civilian targeting is costly, at least over the long term. I do think that decap strikes, to the extent that they erode the quality of militant group violence might be counterproductive for the group in terms of gaining members. Another example might be the Taliban. The Taliban does a lot of civilian targeting and right now the group is increasingly having a manpower problem. Although the focus of the paper is on the tactical effects of decapitation strikes, I do think that there is a strategic impact as well.
Drone We often read or listen to these debates over whether militant groups are becoming more centralized and hierarchical or more dispersed and horizontal.
Abrahms You’re absolutely right, that’s a very important point. I basically have two complementary papers. The paper that I shared with you is forthcoming and we basically test my theory only by looking at decap strikes both in the Afghan/Pakistan and Israeli/Palestinian theaters. However, there are other observable implications to the argument. For instance, as you suggest, the structure of militant groups—if my theory is correct—should also affect their tactical choices. If the leaders tend to favor selective violence against military targets, then decentralized grooves in which lower level members are endowed with additional tactical autonomy, they should be more likely to direct their violence against civilian targets than more centrally structured militant groups in which the leadership can really dictate the tactical choices of its fighters. We find evidence for that in that all else being equal, decentralized groups are more likely to engage in this suboptimal violence against civilians presumably because lower level members with stronger incentives to harm civilians are calling the shots. These two tests of course are not completely separate because leadership decapitation has a similar effect by making groups more diffuse, by empowering lower level members, etc.
We find evidence for that in that all else being equal, decentralized groups are more likely to engage in this suboptimal violence against civilians…
Another proxy we used to test this theory is the location of the attack. The idea was that the further the fighter goes away from the leadership, the less clout the leader has in terms of the fighter’s target selection, and so the further that he goes away from the leader, proxied in terms of whether the attack was international and things like that, the more likely it should be against civilian targets.
Also, interestingly, in the International Organization paper, which is co-authored with Bill Potter, we find that not only do operationally successful decapitation strikes erode the quality of group violence by making it more indiscriminate, but also even when drones are raining down [strikes] on the leaders of militant groups, this has essentially the same tactical effect by making the violence more indiscriminate. You mentioned in the start that you are familiar with bin Laden’s documents seized from his compound. One of the things that he says I think is that although the strikes did not directly impact him, they did convince him to assume a lower profile within the organization, which basically made communications between the leadership and lower level members less reliable, endowing these lower level fighters with more tactical decisionmaking of their own. And so we get at that by testing whether drone strikes density is correlated with this rise in indiscriminate violence and we find that when governments rain down drones, it doesn’t necessarily have to kill the leader to affect their role within the organization and by extension the group’s tactical decisionmaking.
Drone As you say, we have trouble defining terrorism and defining our counterterrorism strategy. I know your paper deals strictly with the tactical side of things, but how do you think your findings contribute to this broader debate?
Abrahms I think that it just broadens the view of how this drone research might be used. I think that in political science in particular, we often get bogged down in these stylized debates that can go on forever. Realism or liberalism, decapitation strikes—do they work or do they not? The truth is clearly somewhere in between and this paper is saying that it is interesting in itself how drones affect the decisionmaking of these militant groups. In terms of my thinking on the efficacy though of drones, and that really is what most people care about, I’ve gone through a bit of an evolution. My emotional response is that, look, if we’ve got a clean shot at a terrorist, especially a terrorist leader, we’d be crazy not to take it. And, in theory, decapitation strikes make perfect sense. In all organizations, I think it’s fair to say, the best and the brightest are more likely to rise to the top than be at the bottom. As a general principle, I think that membership quality improves as you move up through the ranks. It makes perfect theoretical sense that we would take out the leadership and deplete the organization of its most capable members and that that can only have a positive impact in terms of tripling the organization.
Yet, empirically, I haven’t seen strong evidence for this theory. One of the reasons why, of course, which makes it difficult to assess the efficacy of decap strikes is that, and I’m not the first to say this, there’s essentially a selection issues where governments are much more likely to focus their attacks on the more dangerous groups, the harder groups to take out. It’s no coincidence that we use drones so much against al-Qaeda and all its affiliates and ISIS. The fact that these groups are still in operation isn’t necessarily an indictment against leadership decapitation per say and more of a reflection of the fact that we’re contending with a very determined and enduring enemy. Nonetheless, I look at where we use drones and these hotspots don’t seem to be improving. I do think we need to update our assessments of the value of this technology.
Drone Wouldn’t you say that part of the problem with studying drone strikes is that the available data can be unreliable?
Abrahms I don’t disagree with that. I think that we were probably overly generous in categorizing the attacks as against the leaders but we were at least cognizant of that issue. Again, our other dataset [Zussman and Zussman] is better at that. What we tried to do to combat that potential selection issue (that governments are more likely to strike at groups that are seen as more fearsome) is try to exploit variation in the operational outcome of a decapitation strike. So our sample was of both the operational successes and failures and then we compared the tactical effects when it was a success in order to really try to establish causation. The problem there is that we’re much more likely to know about decapitation strikes when they actually hit the target than when they miss. We do not purport to have the universe of all decapitation strike attempts. I suspect that when a drone misses in a field or something, we’re not necessarily going to hear about it. My sense is that if there’s reporting that the leader has been taken out, that’s generally a reliable statement. Not always—I don’t know how many times Baghdadi has been taken out. We don’t always know when the leader has been taken out but generally I think those reports are accurate. Where the drone reporting is probably most inaccurate is in trying to capture these failed attempts.
[O]ur sample was of both the operational successes and failures and then we compared the tactical effects when it was a success in order to really try to establish causation.
Drone Did you look at the roles of some of these high-value targets who were killed? Is that something that would interest you?
Abrahms It is something that would interest me. One of the assumptions of the paper is that when a high-value target has been taken out, the organization is more likely to turn to lower level members to assume responsibility for that position that higher level leaders. If it were true that higher level leaders were more likely to be the successor then I would predict the exact opposite—that the militant group would have more selective violence against military targets. I didn’t study this empirically, but based on what I read my understanding is that when someone is taken out, there is essentially a void within the organization that creates an opportunity for a greater portfolio by lower level members. There is some evidence for this. In Afghanistan—I read a book about the human terrain program—and it was saying that in response to all of these decap strikes, the mean, the average age in these groups went down dramatically and that correlated with the rise in indiscriminate violence. So this isn’t an important assumption that I make but I would love to find real evidence of who these replacements tend to be.
Drone We found that in a few cases, there was a clear successor to a position that was emptied due to a drone strike. A clear line of succession.
Abrahms I mean, these groups would be crazy not to have that. We tend to think of terrorist risk in terms of the risk to the victims—in Washington, D.C. or in Boston. But being a terrorist is itself extraordinarily risky. One of the things I really like about Jenna Jordan’s paper “When Heads Roll,” is that she basically says that there is variation in the resilience in militant groups depending on their characteristics—larger or older groups tend to be more resilient. Both of those factors would make an organization probably more inclined to have these successors identified. Because there is enough manpower, enough time, etc. And this might be one of the reasons why decapitation strikes are so ineffective is because we intend to be targeting with drones, not small newer groups but more established groups that have these successor lines in place.