Then Yemen retaliated. Though its president and a few members of his security cabinet had known of the strike in advance, the working layers of Yemen’s intelligence apparatus, which had functional ties in some instances to militants, retaliated. They explained to the press precisely how the operation had been coordinated and how it had worked. Edmund Hull, the U.S. ambassador to Yemen, had acted as an intelligence scout, sneaking away from Sana’a to meet with tribesmen who had clues to al-Harethi’s whereabouts.3 Hull got a tip, and the NSA found the cell phone pings from a number belonging to one of al-Harethi’s associates.4 Deniability was no longer an option for Yemen, which in subsequent years would struggle under the weight of an al-Qaeda invasion and renaissance and would simultaneously try to find ways to demonstrate its independence from the United States. Yemen made it clear that they would allow no U.S. troops on the ground, except for special operations forces trainers whom they could keep tabs on. The United States did not abide by the agreement. Teams of CIA and JSOC intelligence operatives rotated in and out of the country, often being airdropped into the desert at night or using the cover of a SIGINT mission, of which there were many. The Pentagon quietly established a military liaison element—an MLE—that would ostensibly be used as a cover for direct action missions.
But Hull forbade them from engaging suspected terrorists directly; instead, they gathered human intelligence and helped aviation assets, both manned and unmanned, to find their targets.∗ Though the number of U.S. personnel on the ground at any one time was small, the pace of operations was intense. Even Marine helicopters were used to drop ordnance.5
When the United States negotiated secret agreements with Pakistan to allow drone strikes in restricted areas, it showed sensitivity to Pakistan’s internal politics. Pakistan would be seen as providing intelligence for the strikes; the United States would play a supplemental role. Indeed, when the first Hellfire struck a Pakistani target in June 2004, the Pakistani army claimed responsibility.6 But then the missiles began to go astray, and invariably innocents were killed. The indignation of the average Pakistani was roused. Political factions in Pakistan fanned the flames of protest. In order to rebut the notion that the strikes were causing too many civilian casualties, the U.S. government began to speak about the program to some journalists, on background. Other journalists developed sources within Pakistan’s government, which was quite willing to confirm the details of every subsequent strike, including their intended targets. By the time President Obama joked about the program at the White House Correspondents’ Dinner in Washington in 2009, the “cover” for the program had evaporated. The CIA would swear by the program (operated in conjunction with JSOC), pointing to the number of high-ranking al-Qaeda officials who had been killed.
Obama’s first director of national intelligence, Dennis Blair, wanted the CIA to use its capability more strategically. His reading of intelligence suggested that the collateral harm of the operation—the anger that the strikes caused among Pakistanis, even though the targeting was precise—was damaging to U.S. security interests. The CIA, in a deft bureaucratic move, simply stopped providing Blair’s office with advance notice of strikes.7 The dispute went all the way to the Office of the Vice President, which sided with the CIA, although Blair “won” the ability to have a director of national intelligence representative at CIA covert action briefings at the White House.
Here again, we see the difference between a “good secret”—the technology platforms themselves, such as unmanned aerial vehicles—and the ambiguity resulting from such secrets, in this case essentially invisible, clinical robotic warfare, where missiles appear from nowhere and annihilate villagers with few fingerprints.
Nobody feels bad for dead terrorists. But when a U.S. F-16 fighter pilot makes a mistake and civilian deaths result, there’s a moral element that doesn’t exist when a literally heartless drone does the same thing. Like all creatures of biology, humans evolved to understand the intimacy of killing. Drones change the equation, and its unclear how our mammalian brains cope. But the reality is that no matter who is behind the strike—the United States or the local government—those on the ground predisposed to hating the United States will always blame us. And this technology allows mission creep in a way never before seen. Spying on Pakistan is nothing; to a certain extent, the United States can now spy on or bomb any country in the world at any time. While Spain probably doesn’t have to worry, the Third World does. And terrorists in such failing or failed states depend on the United States to employ boneheaded uses of military force. Covert applications of military force are the most tempting actions of all. Good secrets can very easily go bad.
For this reason, it is hard to stay technologically neutral about drones. They are a ubiquitous presence in global airspace. More than a dozen major police agencies in the United States are testing them. The U.S. military owns about 6,500 of them; the intelligence community probably controls about 500. A small percentage is equipped with missiles; most are used for surveillance, signals intelligence collection, and clandestine tracking.
When McChrystal assumed command of JSOC, it didn’t have a single drone to its name. To schedule orbits, JSOC reconnaissance planners had to ask the Joint Chiefs of Staff, the Air Force, or even the CIA. There were in fact very few intelligence, surveillance, and reconnaissance resources. General Michael Flynn, JSOC’s intelligence chief, had to beg for time on specialized collection platforms such as Medium Altitude Reconnaissance System airplanes, with which he could track insurgents on the ground, and RC-12 Guardrails, innocuous jets that contain highly sensitive signals intelligence collection equipment. With the National Security Council’s assent, Flynn expanded a unit called the Technical Development Activity, which secretly developed manned reconnaissance and surveillance aircraft. He chained it to JSOC for use in Iraq, Pakistan, and, later, Yemen. In Iraq, Flynn and his counterparts established a Joint Reconnaissance Task Force (JTRF) to manage all theater-based requests for special operations force drone orbits. The JRTF was classified; it has now become a JSOC headquarters element and is commanded by one of the Navy’s top special operations officers. In conventional battlespaces, drones are assigned to Combined Air Operations Centers that schedule their orbits, and to several U.S.- and Europe-based reconnaissance fleets, which operate them remotely. The Air Force Distributed Common Ground System links all tasking orders together. More than sixty military installations in the United States house the machines, mostly for testing purposes.8 A dozen unclassified UAV variants are operational, along with a half dozen, which remain classified.
As JSOC ramped up its task force operations in Iraq, the CIA was initially reluctant to provide institutional resources. The National Security Agency, however, under the directorship of General Keith Alexander, was quick to see the benefits of giving Flynn the best personnel and manpower. Several joint CIA/NSA Special Collection Service teams rotated through Balad, Iraq. Alexander personally participated in a secure teleconference with General McChrystal at least once a week. He sent dozens of engineers directly to Flynn’s headquarters in Balad and to other forward-deployed sites, where they implemented a TiVo-like system for signals intelligence that allowed analysts to rapidly process the take from the NSA’s near-total tapping of the telecom networks of several Iraqi cities. SOCOM Technical Surveillance Elements set up cameras and RFID (radio frequency identification) chip tracking sensors. A quiet Pentagon procurement office, the Rapid Response Technology Office, and a classified department called the Special Capabilities Office provided more than three hundred technological assets to assist intelligence and special military operations in the CENTCOM (U.S. Central Command) theater.9 About 60 percent of them went operational.10 The Army’s Technical Operations Support Activity figured out how to merge sensor data collected on the ground with experimental drones in the air, providing what commanders call “persistent” surveillance. Commanders could now track the bad guys and see their activities 24/7 and could analyze patterns with incredible effici
ency. (One promising project involves sensors attached to balloons—the Persistent Threat Detection System.) Tactical satellites were fielded, and by 2009 units could task them to view multiple targets and track as many as ten thousand objects per pass. (The unclassified Pentagon office that works on these projects is called the Operationally Responsive Space Office and is run directly out of the Office of the Secretary of Defense.)
One operational technology that was first attached to drones and later to Artemis geostationary satellites helped JSOC (and later Task Force ODIN in Iraq) figure out where IEDs (improvised explosive devices) might be placed by analyzing how recently the soil had been disturbed. A joint National Geospatial-Intelligence Agency (NGA)–U.S. Strategic Command project called NGA SKOPE allowed JSOC units to merge data collected from virtually any intelligence source and predict, based on patterns of movement, where insurgents were likely to be and what they were likely to do. (To understand how this works, imagine sensors surreptitiously placed on cars belonging to suspected IED planters. Based on the cars’ locations and orientations during an IED attack, the SKOPE cell could predict future attacks based on similar movements.)
Three technologies developed by the NSA during this short burst of time proved pivotal. One, which the government has asked us not to describe in detail, involves the ability to pinpoint cell phone signals to within inches of their origin. (In their book Top Secret America, Dana Priest and Bill Arkin refer to an “electronic divining rod” that allowed operators to hone in on cell phone–using bad guys as though the operators were using metal detectors at a beach.)11 Another involves the use of RFID chips in what can only be described as an ingenious way. (Again, details are withheld because the technique is highly classified and still in use.) One technique that SOCOM has shared with researchers, originally code-named BLUE GRASS, involves attaching tiny RFID emitters to vehicles and tracking them through a variety of different platforms.12 In 2005, Project SONOMA helped analyze where cells of insurgents planning IED attacks were clustered. And JSOC was using dyes and perfluorocarbons to track insurgents before the rest of the military was aware that this capability existed.
It also helped when NSA scientists figured out a way to “unwipe” supposedly cleared cell phones and extract every number ever called by that phone. When a cell phone is captured at a site, the NSA techs download its data using the new technique and feed it to other analysts who are monitoring the data pulled from cell towers across Iraq. If two numbers match, a team is sent to the area to investigate. The NSA, with help from British intelligence, has created a massive database of computer hard drive and thumb drive identification numbers, allowing analysts to trace connections among militants through the technological litter left at sites.
JSOC fusion teams and their augments also benefit from the completion of a comprehensive biometrics database that allows for quick identification of insurgents, as well as a quiet revolution in DOCEX (document exploitation) techniques. Using technology relying on sophisticated algorithms that assign values to data based on the probability that a faint “I” might indeed have been an “I,” DOCEX specialists can reconstruct documents that have been burned. Meanwhile, the Defense Intelligence Agency (DIA) consolidated its media exploitation center and figured out a way to speed up its analysis.13 As late as 2003, lumbering military transport planes had to fly into Andrews Air Force Base in Maryland to drop off unsorted pocket litter by the crate, leaving the DIA’s teams with reams of paper and little context.
To the credit of the Department of Defense and SOCOM, most of these technologies were classified only until they were fielded and then were quickly downgraded to allow the people fighting the wars to gain access to them and push their limits. Had the intelligence agencies been stingy—if they’d been unwilling to relax security controls or had set up shielding special access programs—the fusion cells that eventually beat back the insurgency in Iraq and have been used by U.S. forces ever since would simply not have existed. NGA’s SKOPE cell, for example, was highly classified for about two months. Now it is ubiquitous, and its main architect is permitted to acknowledge its existence in the press.14 “A lot of organizations like this—the Rapid Equipping Force, the Robotics Systems Joint Program Office, the Joint IED Defeat Organization, the Biometrics task force, and probably down at JSOC—were essentially start-ups deploying technologies that were unique to the threats of counterinsurgency,” said Brian Smith, an Air Force captain who worked on energy projects for the Rapid Equipping Force during the early years of the Iraq War.
In October 2011, Flynn, the former JSOC J-2, was promoted to lieutenant general and appointed a deputy director of national intelligence. He has been outspoken about the need for reform within the military intelligence community. Many of his fellow flag and general officers in the intelligence community consider him to be too outspoken. His nomination took more than eight months to gestate, as forces within the Pentagon—inside the Army in particular—pushed back, whispering into the ears of senators that Flynn’s tactics did not work when he followed McChrystal to Afghanistan in 2009. (Flynn had the last laugh. Today he is the director of the DIA and commander of all intelligence forces in the Department of Defense.) Some Democratic senators on the Armed Services Committee believe that Flynn’s championing of bulk data analysis provided a brutally efficient way to kill too many innocent Afghans and may not have been as effective as the military suggests. By this, they mean that instead of targeting people, infantry and special operations forces targeted telephone numbers—they would target gatherings of people who had been surreptitiously tagged with chemicals or RFID chips, even if they didn’t precisely know who these targets were.
Officially, the DIA’s Joint Interagency Task Force–Counter Terrorism vetted the targets, with input from the NSC. In reality, the Joint Prioritized Effects List—the target board—had a lot of phone numbers that a computer had associated with the broad periphery of the insurgency, rather than names of specific terrorists.15 Flynn’s response to this is simple: for one thing, raids weren’t ordered because some Afghan villager happened to call a Taliban commander; there needed to be a better reason to send Americans into harm’s way than that. One of the hardest tasks that Flynn’s intelligence team faced was figuring out whether contacts between innocent Afghans and those associated with the Taliban were innocent or nefarious. Doing that required a significant amount of collection and analytical time. It required a granular level of knowledge about each village. Plenty of people were collecting all sorts of information—Provisional Reconstruction Teams, Human Terrain Teams, Civil Affairs officers, and intelligence gatherers—but it wasn’t being fused or analyzed or appropriately disseminated.
Flynn wanted to create a middle level of what he termed “information brokers,” who could analyze everything and determine patterns that would allow all parts of the Afghanistan effort—especially the mission to rebuild civil society—to succeed. As he described it, “This vast and underappreciated body of information, almost all of which is unclassified, admittedly offers few clues about where to find insurgents, but it does provide elements of even greater importance—a map for leveraging support and for marginalizing the insurgency itself.” He attempted to divine, in villages and provinces, who was good and who was bad and attempted to flesh out (as much as possible) which members of the Taliban were secretly cooperating with the State Department or the CIA and which members were susceptible to U.S. influence. But Washington saw American body counts and ordered General McChrystal, as the commanding general of the International Security Assistance Force, to force Flynn to reprioritize his resources. He had to stop the flow of money and trainers from Iran who were arming insurgents. He had to counter growing Pakistani influence in the region and deal with the nettle of cross-border political complexities. And he had promised to provide conventional forces in Afghanistan with the same high-grade, high-velocity intelligence that special operations forces received.16
On a fresh patch of land in the northwestern
corner of Fort Bragg, specially cleared construction workers are completing a massive 110,000-square-foot building that will serve as the crown jewel in JSOC’s empire. The building will headquarter the JSOC Intelligence Brigade (JIB), which analyzes raw and finished intelligence for the Command’s special missions units. The JIB has quietly existed for more than three years, escaping the notice of congressional intelligence overseers. Under a program started by Generals McChrystal and Flynn, JSOC borrowed hundreds of intelligence analysts from the sixteen U.S. intelligence agencies, many of them rotating through quick frontline deployments. These augmentees greatly helped JSOC conduct its operations, but the Command was not able to develop a cadre of analysts who were JSOC’s own, with institutional muscle memory that would make the fusion of intelligence and operations more efficient in the future. The JIB, in essence, sets in stone JSOC’s new way of doing business.
In 2010, Admiral William McRaven attended the quiet ribbon cutting of his newest jewel: the Intelligence Crisis Action Center (ICAC) in Rosslyn, Virginia, funded through a classified line item in the Pentagon’s budget. Until just recently, it operated on two floors of a nondescript office building that also housed a language learning center and a dry cleaner. At the time when McRaven christened the center, its existence was a secret to many U.S. intelligence officials, who learned about it by way of an Associated Press newsbreak in early 2011. According to a senior military official, it has about fifty employees and reports directly to the JSOC Directorate of Intelligence and Security.17 Its primary function is to serve as a command post for JSOC operations around the world. It is informally known as the Targeting Center, and because of operational security concerns it has changed its name twice.
These entities are sensitive subjects at the highest levels of the U.S. counterterrorism community, because each represents the extraordinary achievements of the JSOC units and also reveals by its own existence the inadequacy of the other intelligence fusion centers set up by the government to do mostly the same thing. JSOC’s successes have brought with it blowback and envy and more than a bit of criticism from military officials who think that conventional forces and regular special operations forces units were just as important as the smaller, secretive standing task forces in degrading al-Qaeda’s infrastructure.
Deep State Page 18