[Editor’s Note: Stay tuned later today for a post by Ryan Goodman and Derek Jinks responding to the guest post by Chris Jenks and Geoffrey Corn.]
In The NSA’s Secret Role in the U.S. Assassination Program, Jeremy Scahill and Glenn Greenwald overpromise and under-deliver.
But their article may serve as a useful targeting Rorschach test.
If you read the following in their article what conclusions would you draw?:
- Cellphone communication content and patterns are major components of the threat identification process conducted in support of US targeting operations;
- National Security Agency (NSA) signals intelligence (SIGINT) capabilities provide remarkably detailed cellphone based information to US commanders engaged in the target decision-making process;
- Cellphone exploitation allows the NSA to develop pattern analysis of suspected terrorists, including voice identification of an individual terrorist operative, who his associates are, who his commander and subordinates are, and precise location of the cellphone;
- US commanders have relied on this intelligence to identify individuals who meet the enemy belligerent definition;
- This determination has resulted in night raids and drone strikes to kill or capture the individuals in possession of the phones that intelligence indicates are being used by belligerent operatives;
- Before a drone strike or raid is ordered there must be at least two sources of intelligence;
- It can take 18 months or longer to gather sufficient intelligence for a drone strike or raid to be approved;
- The vast majority of individuals targeted based on NSA SIGINT capabilities have been terrorist belligerent operatives and their associates engaged in ongoing efforts to attack US and Coalition personnel;
- These drone strikes and raids have been instrumental in disrupting enemy efforts to kill US and Coalition personnel in Afghanistan by use of improvised explosive devices;
- Despite best efforts to limit the consequences of such attacks to actual enemy belligerent personnel, mistakes have been made, frequently as the result of the enemy giving a cellphone SIM card to an innocent civilian;
What would you think? Would you condemn reliance on this intelligence? Would you condemn all such attacks if you learned that intelligence assessments were not 100% accurate? Would you condemn a US commander if a SIGINT based targeting decision killed an innocent civilian because, unknown to the commander, the enemy had transferred the cell phone to that individual? Unlikely. Yet these are the conclusions Scahill and Greenwald ask of you.
Why anyone would be surprised that the US military is relying on SIGINT and network operations provided by the NSA, the US governmental agency whose stated mission is to use SIGINT and network operations to support the U.S military, is utterly perplexing. Even more perplexing is why anyone would object to the US targeting individuals determined, through extensive intelligence analysis, to qualify as terrorist belligerent operatives who have killed or are trying to kill members of the U.S. military?
Apparently Jeremy Scahill and Glenn Greenwald are both surprised and troubled.
Scahill and Greenwald condemn both the reliance on SIGINT to support targeting operations and the resulting attacks on suspected terrorist operatives. But their article offers little more than blinding statements of the longstanding and obvious and an unbalanced critique of targeting operations which, on the whole, are legally justified and appropriate. In support of this condemnation, they rely almost exclusively on the opinions of two low ranking former drone sensor operators, and have built upon this feeble foundation not only a recent article, but their current speaking tour with hyperbolic titles like “death by metadata.” As Ryan Goodman has already described, despite its claims, this is not a story about metadata. Ryan also labeled their article a “must read;” we disagree. But in mentioning the talismanic words Snowden, NSA, and metadata, the story is receiving more attention than it should. Given that, we feel it is important to highlight some of what Scahill and Greenwald failed to mention – about their argument and their sources.
In addition to confusing metadata with SIGINT, Scahill and Greenwald ignore the legal and moral responsibility the terrorists bear for civilian deaths from SIGINT based drone strikes. And they do themselves, and the article’s credibility, a disservice by purporting to write on contemporary targeting operations while relying on a former junior member of the U.S. military discharged almost three years ago, with no consideration of targeting doctrine, legal requirements related to the targeting process, or the reality that no targeting process can ever guarantee absolute accuracy.
The article reflects a view that targeting based on signals and electronic intelligence is both new and inherently inaccurate. This is demonstrably false. Since the inception of electronic communications, SIGINT has been a key component of threat assessment and target decision-making. Indeed, the NSA evolved to provide US commanders at every level of operations timely, accurate, and vital insight into the organization, activities, and intentions of actual and potential wartime adversaries.
Even a cursory understanding of military operations indicates that in the overwhelming majority of encounters in combat involving the use of deadly force, the two sides never have “eyes on” or see their opponent with the naked eye. Instead, a wide range of sensors, radars, and communications intercepts provides the intelligence relied on by US forces to determine who and where the enemy is. For example, when insurgents fire mortars or rockets at U.S. outposts, electronic sensors and counter battery radars determine the point of origin. Relying on that information, a military commander seeking to achieve a rapid response might order an artillery counter-attack, call in an air strike, or rely on an armed drone to attack the individuals assembled at that location. Another example is the anti-radar missile used by U.S. attack aircraft, which locks on to the emissions of enemy air defense radar systems and guides the lethal missile to that target. The suggestion that reliance on this type of intelligence is either new or invalid is therefore simply false.
This fallacy also extends to reliance on communications pattern analysis – intelligence historically provided by the NSA – to justify a lethal attack. For decades US targeting operations have relied on such analysis to identify enemy command and control (C2) networks, pinpoint the location and vulnerabilities of those networks, assess when those involved in the C2 function are most vulnerable, and attack based on this information. In all of these examples, the U.S. employs lethal force based on signals and electronic intelligence, often in a decision-making process measured in seconds.
It is of course true that in our current struggle, U.S. forces rarely attack presumptive military targets like enemy radars. This does not, however, undermine the importance or validity of signals intelligence. Instead, it alters the nature of the source and significance of enemy communications, which for this enemy is the cell phone.
That’s different. And the U.S. treats it as such. Indeed, Scahill and Greenwald acknowledge that unlike the lethal force decisions made in seconds described above, the intelligence gathering, assessment, and targeting process for a drone strike based on cell phone signals is measured in months. Furthermore, as they note, the concern for targeting accuracy has resulted in a self-imposed policy requirement that there must be two sources of intelligence to support an attack decision.
This should not, however, support an assertion that SIGINT-based attack decisions must be 100% accurate. Such an expectation defies both the realities of combat (especially when fighting an unconventional enemy), and the legal standard of reasonableness applicable to all such decisions. Scahill and Greenwald, however, rely on several examples of targeting error to support their condemnation. But consider the error examples from the article and ask yourself whether they are a problem, and if so, who is responsible.
The first type of error the article describes involves top Taliban leaders “purposely and randomly distribut[ing] SIM cards among their units in order to elude their trackers.” As a result, U.S. forces believe they are targeting Taliban leader A, but A has swapped SIM cards with Taliban leader B, so the drone strike ends up killing B. Other than A still being alive, that an unintended Taliban leader was killed is squarely in the “not a problem” category.
The other type of error the article describes is where the suspected terrorist gives his cell phone to “friends, children, spouses, and family members.” This could, and purportedly has, led to drone strikes that unfortunately kill the individual who took possession of the cell phone (and those in close proximity) instead of the intended target. This is, of course, terribly unfortunate. But is it justifiable to label it “a problem” when the “problem” implies culpability on the part of the US commander? The answer is clear: mistakes happen in war; they are problems only when they result from unreasonable judgments. For this “problem,” we should therefore start by questioning why the authors would have readers impose legal or moral culpability on the U.S. if the U.S. has undertaken reasonable precautions leading up to the strike, the unintended civilian deaths are attributed to the suspected terrorist.
There is nothing remarkable about the notion that a targeting mistake, even when it results in the death of innocent civilians, does not always justify condemnation. So much is inherent in the law of armed conflict, which implicitly acknowledges the simple reality that “reasonable” does not always mean “right.” While the targeting principle of distinction requires the U.S. to distinguish between civilians and belligerents, during both conventional and unconventional conflicts, the ultimate touchstone for compliance with this principle is the good faith effort to gather timely and accurate information and take feasible precautions to mitigate the risk of targeting error. Scahill and Greenwald, however, would have us believe that the law imposes a strict liability standard on commanders: any error, even when reasonable under the circumstances, justifies condemnation.
Perhaps even more misleading is their failure to point out that the law of armed conflict also imposes obligations on the potential target: a prohibition against deliberately using civilians in an effort to shield themselves from lethal attack. Why is this omission so damaging to the credibility of their condemnation? Because it supports a fundamentally different conclusion: it is the terrorist operative who uses and then gives his cell phone to a civilian in an effort to avoid being the object of US attack, who is both legally and morally culpable for the death of that civilian, not the US commander. Indeed, this tactic is even more pernicious than a uniformed enemy giving his tunic to a civilian, knowing it might result in the civilian being attacked. At least in that situation the civilian realizes he is assuming a risk by wearing the tunic. In contrast, when the terrorist operative gives the cell phone to an innocent civilian, he is not only exploiting the limitations of US intelligence to gain the advantage of avoiding attack, he is potentially exploiting the ignorance of the civilian by making that individual an unknowing decoy. That is what Scahill and Greenwald should condemn.
In addition to distorting both the targeting process and the law that guides that process, Scahill and Greenwald also fail to provide credible support for their conclusions. This is made clear through their misplaced reliance on an “insider directly involved in the drone program,” Brandon Bryant. Scahill and Greenwald omit the fact that Bryant was a junior enlisted member of the U.S. Air Force in no position to provide informed comments about the NSA’s role in the targeting process. Indeed Bryant acknowledges as much, stating that “I don’t know who we worked with….We were never privy to that information. If the NSA did work with us, like, I have no clue.”
Bryant is by no means the only clueless one here. For Scahill and Greenwald to rely on a Senior Airman in his early twenties for strategic analysis and critique of the U.S. drone program, targeting procedures, the integration of intelligence into the process and the NSA’s role is patently absurd. Bryant’s perspective is neither particularly substantive nor contemporary. According to the U.S. Air Force, Bryant was involved in seven missions in which a drone fired a missile, all of which occurred prior to February 2009, over 4 years ago.
Toward the end, the article drifts into the surreal. Scahill and Greenwald claim that Bryant has taken personal responsibility for his actions in the killing of Anwar al-Awlaki and that the killing of Awlaki and his son haunt Bryant. Bryant is free to be haunted by his actions while in uniform. But Awlaki was killed on 30 September 2011, four months after Bryant was discharged from the US Air Force in April 2011. It is therefore impossible that he played any role whatsoever in the attack that killed Awlaki, let alone the subsequent drone strike two weeks later which killed Awlaki’s son
Returning to the Rorschach test, if we told you the authors of an article on contemporary targeting operations relied on a former low level service member who was involved in a handful of drone strikes over 4 years ago; that the authors failed to acknowledge that SIGINT and pattern analysis have for decades been relied on as a valid source of information to support targeting decisions; that the authors failed to describe accurately either targeting doctrine or the law that regulates lethal targeting; that the authors drew the inference that some mistakes in the targeting process indicate that all decisions are flawed; that the authors never addressed the tactical and operational benefits that have resulted in these operations, what would you think?