May 29, 2013

I Might Need You To Kill: Signatures, Patterns, and Alternatives

From what we know from seemingly deliberate leaks on the eve of Obama’s major counterterrorism speech at the National Defense University last week, the most widely criticized ongoing aspect of America’s counterterrorism program, the “signature strikes,” against those engaging in suspected militant activity without confirmation of their identities, are set to decline - though they will apparently continue in Pakistan (where, we should note, the tempo of strikes is already falling drastically). Where many of the early objections to the targeted killing campaigns focused on the erosion of norms of assassination, many critics of signature strikes objected to killing targets when, as Senator Rand Paul put it in his filibuster, “We don't have to know what your name is, who you are, [or] who you're with.”

Whether strong critics or general proponents of targeted killing, drone strikes, the war on terror, or however one prefers to frame it, very few are willing to come out in support of the signature strike. They are a fitting symbol for everything wrong at the current iteration of the war - the remote and reckless killing of individuals under the auspices of what is supposedly a targeted and surgical campaign, rendered anonymous not simply in the media reports the public consumes but even in the bureaucratic and operational processes that bring about their deaths. Yet we should know by now that rehabilitating or discarding a symbol is no substitute for addressing its underlying causes. Signature strikes, as with most other representation of the policy problems that seem to extend beyond comprehensible scale, are easy to digest as moral wrongs, yet they arise from a fundamental frictions between the laws of war, the behavior or insurgent and terrorist groups, and the constraints on American operations in its global battlefield. There are many good cases for reducing or ending signature strikes, but doing the most good requires assessing why we conduct signature strikes and appraising the likely alternatives for replacing them or abandoning the objective they sought outright.

From a conventional military perspective, anonymous killing, in and of itself, is not the issue with signature strikes. Behavior, not your personal identity, provide the bedrock for established distinctions between combatants and non-combatants. Tactics and technology dictate that combatants often, at best, know they’re shooting at someone acting like somebody it’s permissible to shoot at.

The problem is what behavior qualifies you as a target. In a classic military conflict, it is simple enough to know that the enemy is bearing arms and distinctive markings. Distinguishing yourself from a civilian is an important tradition for militaries seeking legitimate authority to engage in violence. That many militant groups, such as Colombia’s FARC or the Philippine MILF consciously adopt semblances of uniforms to counteract government claims that they are merely terrorists rather than legitimate combatants in civil wars, is an important reinforcement of this norm. A wide number of insurgent or irregular groups adopt some overt type of combatant markings, whether they be helmets, camouflage fatigues, or even masks, in addition to simply openly carrying arms and engaging in hostilities.

Of course, even groups such as FARC which place a conscious priority on legitimating themselves as combatants still employ operatives who do not openly present themselves as combatants, such as those which assemble and plant bombs in populated areas where wearing a uniform would quickly call the attention of Colombian security forces. More broadly, insurgent infrastructures rarely seek to identify themselves to the enemy or population. Even in countries with traditional militaries, logistics, intelligence, propaganda, and political command and control are all areas which blur the line between what combatants may and may not target, where civilians play a great role in facilitating the war effort.

It is telling when we use the Phoenix Program and the associated anti-infrastructure efforts as shorthand historical analogies for modern military and paramilitary counterterrorism operations. As William Rosenau and Austin Long point out, what we frequently call the Phoenix Program consisted of a variety of interrelated efforts, ranging from intelligence sharing to CIA-South Vietnamese paramilitary units to carry out kill/capture raids, to sophisticated electronic and human surveillance of border regions. It was not an effort to decapitate the high leadership of communist movements in Indochina. It was an effort to sever the lines of support to Viet Cong insurgents at the most local level possible.

We do not know much about the criteria for signature strikes. The most recent leak at time of writing suggests that those which will continue in Pakistan will target “groups of unidentified armed men presumed to be extremists.” Nevertheless, in heavily armed areas and militia-ridden areas of the world, such as the tribal areas of Pakistan, this alone does not explain how stringent or broad signature strikes are. The real issue with distinction, though, is that basically no current U.S. target wants to readily offer one. How do you ethically and practically target the behavior of entities that already deliberately obfuscate and submerge as much of their activity as possible within civilian lifestyles and populaces? I understand when people far wiser on the laws of war than I, such as Kenneth Anderson, sound concerned over perversely increasing the incentives for AQAP leaders to use civilians as shields. Yet as Gregory Johnsen warns, AQAP goes out of its way to exploit civilian deaths from U.S. causes, and AQAP efforts to ingratiate itself among the local populace (though very often unsuccessful) frequently benefit from lethal U.S. errors as is. In Pakistan, where not just al Qaeda leadership and affiliates but Taliban insurgents (who can easily target Americans across the border), the criteria for these types of strikes are even wider. Nevertheless, even groups with relatively little local popularity require and establish complex infrastructures to sustain their operations.

Without multiple types of intelligence collected over an extended period of time, it is difficult to ascertain whether individuals are even directly involved in infrastructure operations for a hostile group. Even then, it’s not entirely clear international humanitarian law provides wide enough scope to target civilians engaging in the full range of infrastructure activities. As one scholar skeptical of the targeted killing program puts it:

In his view, two types [of 14 strikes examined] are legally justified: strikes against unknown individuals who are transporting weapons and attacks on known Al Queda compounds....

Finally, Heller claimed that one signature — “facilitating terrorist activity” — was inherently neither legally adequate or inadequate, as it depended on how the United States interpreted the conditions in question.

In his view, direct participation in hostilities, gathering military intelligence in enemy territory, acting as a guide for an organized armed group, or providing ammunition during hostilities fulfilled the requirements of legal adequacy for this signature. However, “war-sustaining activities” such as recruitment, propaganda, fighters, financing, or providing fighters food, lodging, or logistical support, did not.

In essence, the problem is not actually anonymous killing, it’s what it’s permissible for the U.S. government to kill you for, period. Of course, in most cases of internal conflict, domestic laws open up a much wider scope of authority for local or allied forces. Nevertheless, it is important to recognize a fundamental problem with why signature strikes exist - an attempt to navigate an area where terrorist and insurgent perfidy is rampant and political-military barriers to effective intelligence collection and capture programs are frequently limited.

Additional Protocol I of the Geneva Convention arguably does not apply to Non-International Armed Conflict, but there is a case that Additional Protocol II’s civilian protections amount to roughly equivalent prohibitions on claiming noncombatant status. Nevertheless, the notion that governments need to know the names or even the affiliations of non-state actors they target is not an accurate reading of either the letter or custom of international humanitarian law. While establishing somebody not immediately engaging in hostile behavior as a legitimate target requires corroborating intelligence and surveillance, it does not necessarily entail a targeted killing seeking out a specific personality.

The difficulty of targeting an enemy which lives among the population compounds the dangers that insurgents and terrorists already pose to civilians by blurring the distinction between combatant and noncombatant. Manhunting and assassination, especially by remotely-piloted aircraft, seem, for whatever reason, distasteful and cowardly to some. Yet it is hardly an act of technological cowardice for the infiltrators, informants, and other spies whose job it is to impart a degree of precision upon these strikes. They enjoy virtually no substantial protections under the laws of war, and even then, insurgent groups and terrorists are manifestly unlikely to extend them to spies or traitors anyway. Faced with an enemy disguised as civilians, counterinsurgents frequently seek to leverage civilian information, defectors, or infiltrate civilian communities and insurgent networks to generate intelligence for strikes.

Philip Mudd, in a rare direct defense of signature strikes, highlights increasing terrorist and insurgent fears of infiltrators and provoking aggressive counterintelligence as benefits of maintaining the tactic:

There is an intangible factor that reinforces the effectiveness of signature strikes: the fear factor, coupled with the suspicions and paranoia that result from organizations searching desperately among their ranks to find out who is providing the Americans information so detailed that we can wreak such havoc over such a long period of time. Time and again, intelligence has clearly told us that the adversary dreads these operations -- lethal strikes that come anytime, anywhere, and that eliminate entire swaths of organizations. And these same organizations then turn around and further degrade their operational capability by engaging in savage hunts for leaks.

The life of an intelligence collector in enemy territory, where strikes do nothing to guarantee your personal safety, is not an enviable one. Just ask the informants or suspected informants militants in the Pakistani border region began executing at alarming rates, or the suspected Israeli informants Hamas dragged through the streets. Anyone who’s familiar with Kalyvas and other logic of violence approaches can intuit the problem here. In areas where insurgents and terrorist groups have preponderance, but the number of potential infiltrators is high, militant and civilian incentives align to produce high rates of civilian killings to target defectors. By taking on civilian guises, and necessitating the use of civilian informants or counterinsurgent infiltration, even the most theoretically surgical approaches to manhunting and air policing create dangerous second order dynamics for civilian populations, which the counterinsurgent has little ability to mitigate.

Of course, it is important to note these violent dynamics are hardly unique to signature strikes or aerial assassinations. Though improper targeting and munitions selection can rapidly magnify the danger of collateral damage in those operations, it is important to remember the enormous potential costs of seeking to kill or capture militants with any instruments that are unable to ensure security for civilians.

Night raids, such as the infamous botched Gardez raid, can easily falter on poor intelligence and the mistaken use of force against civilians. While the U.S. has learned much since Black Hawk Down, compare the amount of force the U.S. has had to bring down to ensure adequate force protection in operations on or across the border with Pakistan. In 2008, the Angor Ada raid, involving dozens of US ground troops and multiple aircraft, killed at least several civilians, and meant that the next raid provoked a major standoff that threatened to cause large-scale firefights and civilian displacement. Or take the example of 2011, NATO and Afghan forces reported fire from Pakistani positions. The result was that NATO unleashed more airpower to protect its troops in contact than MQ-1s or MQ-9s ever could. Two Apache attack helicopters, two F-15Es, and an AC-130 gunship pummeled targets on the Pakistani border, killing as many as two dozen Pakistani soldiers.

Even then, a regime of raiding into territories where we are not willing to actually create a sustained military presence does nothing to mitigate the dangerous dynamics for civilian cooperation and intelligence collection. Raids that leave territorial control an open question for host governments and militants do not give civilians much incentive to provide the intelligence necessary for more precise targeting, leaving them to try their luck at the dangerous game of avoiding militant counterintelligence efforts. The solution to such a quandary, especially when providing security to positively incentivize informers, is to remove potential informers from enemy retaliation through detention or concentration, and the use of high-tempo raiding operations to generate as much actionable intelligence as possible through the raiding process itself.

The face of a robust capture program is not the FBI effort which retrieved the 1993 CIA shooter, which in the relatively sanguine climate of 1997, the Pakistani government was unwilling to publicly admit its role in handing over a citizen to the US. American law enforcement wisely worked with the ISI to lure the suspect into Punjab. In today’s climate, against targets part of active militant networks, an operation that relies on relatively unsavvy suspects and highly compliant host government security and intelligence seems less than forthcoming.

The face of a capture program in Pakistan’s border regions with Afghanistan, Yemen, Somalia, and similar environments, is not going to be law enforcements, but the types of programs that, past and present, we praise with intimidation or decry with disgust as “industrial-scale killing machines” or “executive assassination rings.” Wartime friction ensures that any well-intentioned capture program in denied or contested areas will live on as an assassination program. Just ask those involved in the Phoenix Program, who had even their own President thinking they were running a massive assassination machine.

Of course, programs like the CIA-Vietnamese Provincial Reconnaissance Units, for all the reputation they gained as an unstoppable assassination machine, pale in comparison to the sanguinary behavior of other paramilitary efforts to dismantle insurgent infrastructure and disrupt irregular opponents. While the Anbar Awakening receives massive praise, enlisting irregular forces with relatively little opportunity to control their behavior, and far less “skin in the game” with regard to the political situation on the ground, frequently results in incredibly excessive killing and the incorporation of civilian populations into horrific, racket-like forms of extortive governance. Even relatively antiseptic terms such as extending the reach of governance and strengthening the state, in the context of civil wars or internal conflict against irregular opponents, frequently involves the tacit or explicit cooperation between host government and paramilitary forces to purge not just insurgent infrastructure, but political sympathizers and threats to elite interests. Extending the reach of the state under such conditions is frequently a nasty thing, and while it is in vogue to speak of the death or decline of counterinsurgency, the clean language of empowering local partners and expanding state capacity is still counterinsurgency, just of a much different sort than the kind wealthy liberal 3rd-party interventions might try to steer their clients towards.

In light of all this, it’s not surprising why advocates of reforming drone strikes now criticize signature strikes with desires for more surgical targeting. While there remains debate over when and what ought to trigger a targeted killing, discomfort with the requirements of anti-infrastructure operations and a wide public and elite consensus over the value of leadership targeting now see an institutionalized, accountable, reliable form of assassination as the desirable alternative to the comparatively more brutal and hamfisted method of signature strikes.

There are, of course, other, significant changes afoot. The implication that “threats to a partner like Afghanistan or Yemen alone would not be enough to justify being targeted” will force a renegotiation in the terms of cooperation through which the Yemeni and Pakistani governments privately bless or at least tolerate American strikes. If the U.S. is no longer targeting groups of militants who do not pose a threat to Americans, or Nek Muhammad-types who are primarily a threat to partners, the targeted killing program is, in some ways, listing towards more unilateralism, just as the emphasis on precision (versus signature) quietly further entrenches assassination and targeted killing. Lastly, if and when the U.S. finally ceases signature strikes, it marks basically the end of an air-policing like attempt to conduct anti-infrastructure attacks with standoff weapons. If ending such operations is an admirable and necessary goal - and there is plenty of evidence to suggest it is - we ought think about what will replace those efforts, or what strategy will allow us to bypass such operations. The targeted killing program grew out of a strong but fuzzy commitment to taking the war to al Qaeda, frustrations with capture, detention, and trial, and bureaucratic maneuvering and inertia unleashed through dissatisfaction with the flaws plaguing the counterinsurgency campaign. So long as the basic dynamics of policy and strategy that animate our counterterror campaign today remain - agent-principal relationships with some of our least-savory allies, a murky and frequently flawed view of the adversary we seek to vanquish, and the skewed incentives of our domestic political system, the problems which afflict policies past are soon to manifest themselves anew to afflict their successors.