On the last day of 2003, Macedonian border guards arrested Khaled el-Masri at the Serbian border. He had a suspicious name, and the Macedonians didn’t like the look of his passport. After beating el-Masri, stripping him and searching him with painful and highly invasive thoroughness, the Macedonians turned him over to the CIA, which flew him to the ‘Salt Pit’, a black site in Afghanistan. He told his American captors the same thing he’d told the Macedonians: he was an innocent man, a German citizen on vacation. He made a living selling cars. He had no connections to al-Qaida or any other jihadis. After el-Masri had spent 149 days in captivity, the CIA realised he was telling the truth. The man they had arrested was not Khaled el-Masri the jihadist, he was Khaled el-Masri the used car salesman. The CIA dropped el-Masri in Albania, from where he had to make his own way home to Germany. He was three stone lighter. His apartment was empty. In his absence, his wife and four children had moved back to Lebanon. He was given €14,500 and no apology. One of those who argued in favour of the kidnapping and imprisonment of el-Masri was Alfreda Frances Bikowsky, one of the leaders of ALEC Station, the special CIA unit tasked with hunting Osama bin Laden. ‘She just looked in her crystal ball and it said that he was bad,’ one of her co-workers told Jane Mayer of the New Yorker.
The error was exposed in 2005 by the New York Times. The CIA completed its own review of the case in 2007. It turned out that Bikowsky didn’t have probable cause, or even reasonable suspicion – she had a hunch. ‘Available intelligence information did not provide a sufficient basis to render and detain Khaled el-Masri,’ the agency’s inspector general wrote. He recommended setting up an ‘accountability board’, an internal tribunal, to investigate the handling of the case, a step that could have led to disciplinary measures. The final decision on this fell to Michael Hayden, the CIA director at the time. He chose not to. ‘It was a pretty easy call,’ he writes in Playing to the Edge, his new memoir. He doesn’t describe el-Masri as ‘innocent’, noting his ‘clouded past’, but does admit he was a ‘false positive’. The Macedonians, on the basis of his name, had positively identified el-Masri as a terrorist. This identification turned out to be wrong. Had Hayden disciplined Bikowsky, it would have sent a dangerous message to CIA agents that ‘the most important thing is to avoid false positives … even if it means a few true positives slip through.’ He made the same argument a few years earlier to the Senate Select Committee on Intelligence: ‘With regard to counterterrorism operations in general, and the el-Masri matter in particular … the scale tips decisively in favour of accepting mistakes that overconnect the dots against those that underconnect them.’ El-Masri is not the only innocent person to have suffered when the dots have been overconnected. There’s also Brandon Mayfield, Jamal Tarhuni, Mustafa Elgobi, Abdullah al-Kidd, Adama Bah, Yasir Aladdin Afif and Rahinah Ibrahim, among others.
How many false positives for how much safety? In the canon of Western jurisprudence, it usually takes only a small number of false positives to call the legitimacy of the whole system into question. It is ‘better that ten guilty persons escape than that one innocent suffer’ according to William Blackstone’s famous ratio. For the US Supreme Court Justice Stephen Breyer, the 4 per cent of convicts exonerated after execution are significant enough to justify sparing the other 96 per cent. It was Benjamin Franklin who said it was folly to trade ‘essential liberty’ for ‘a little temporary safety’. According to Hayden, we must also consider the number of innocent people who may be killed in a future terrorist attack. ‘What might be admirable for a court system is unconscionable for an intelligence agency,’ he writes.
Hayden was appointed to lead the National Security Agency in 1999 by Bill Clinton, and stayed on through 9/11 up until 2005, when he became the principal deputy to John Negroponte, the first director of national intelligence. In 2006, George W. Bush appointed him director of the CIA, where he stayed until 2009, when Barack Obama replaced him with Leon Panetta. During Hayden’s ten years at the top, the intelligence establishment assumed many quasi-judicial powers. Domestic surveillance, detention, evidentiary findings and adjudication became part of the remit, as did the killing of US citizens – at least one, Anwar al-Awlaki, with explicit intent. For Hayden, this is all just ‘playing to the edge’. The boundaries of his imaginary football field are the constitutionally-derived limits of the intelligence community’s authority. The threat of terrorism requires that they be pushed to the limit, or even further.
El-Masri was one of at least five innocent men mistakenly detained by the CIA after 9/11. (The total number of black site detentions is still unclear: ‘Early bookkeeping had been sloppy,’ Hayden writes.) Then there are the more than 700 million airline passengers screened by the Transportation Security Administration last year, and the hundreds of billions of domestic US phone calls that have had their metadata, including origin and destination numbers, logged by the NSA and continue to be held for government review by the phone companies. There are also the lists that the US government keeps of people whom it deems to be worth watching. The best known is the No Fly List, which contains tens of thousands of names. The more obscure Terrorist Identities Datamart Environment (TIDE) has more than a million names, a figure that dwarfs even the most pessimistic estimates of the number of terrorists among us. More than 700,000 names appear in the FBI-maintained Terrorist Screening Database, which is shared with state and local police as well as with 22 foreign governments. Finally, there are the people who, in the words of the US Department of Homeland Security (DHS), are ‘not on the terrorist watchlist, but are misidentified as people who are’. According to the DHS, 99 per cent of people who were stopped at an airport and then contacted the agency to clear their names were never flagged to begin with. Like the German used car salesman, their names were just similar enough to arouse suspicion.
A list known as the ‘disposition matrix’ or ‘kill list’ spells out who is to die, where, and by what means. An article from 2012 by Greg Miller of the Washington Post called it an ‘operational menu’ and emphasised how carefully it is updated and reviewed. Hayden seems untroubled by the ‘special emotional strain’ experienced by the NSA linguists whose job is to eavesdrop on those at the top of the list. The analysts ‘sometimes know their target better than they know their own families’. Some have opted out of confirming a subject’s identity before they’re killed. The analysts’ distress may be owed, at least in part, to knowing so much about their targets that they begin to have doubts about their guilt. When it comes to counterterrorism, human intelligence is far more acute than any computer program, but it’s also expensive and hard for the intelligence services to come by – especially in a foreign language – and human analysis can’t be scaled up to cover hundreds of thousands of potential targets. ‘We kill people based on metadata,’ Hayden has said – meaning which phone called which phone, when and where from. One wonders how often the words said during these calls are taken into account, and how often they are used to validate a kill decision that has already been made.
Using its lists, the US government defines a set of people that pose a degree of risk in the eyes of the intelligence services, but are still largely innocent in the eyes of the courts. The problem of how to manage these people poses a dilemma for Western governments. The US strategy has four prongs: electronic surveillance populates the lists; algorithms and analysts sort and rank them; secret tribunals pass judgment on some of the biggest names; and the drones cross them off the list. Further down the list, a much larger number of suspects undergo milder forms of punishment: secondary screening at airports, interrogation, physical surveillance, cancelled passports, approaches by informants and pressure to inform. The last technique can sometimes resemble bribery, according to a recent Buzzfeed investigation, with the FBI offering to exchange a green card for information.
The smaller the group of people you are looking for in comparison to the size of the population, and the lower your tolerance for missing even one of them, the greater the number of false positives your search will produce. This is why unusual but harmless debit card activity so often leads to a phone call from your bank’s fraud department, and why most men don’t get screened for prostate cancer until they are fifty. Even if there were a test that could flag terrorists with 99 per cent accuracy, the vast majority of those flagged would be false positives, because the test is being applied across a population where the number of true positives is minuscule.
Hayden never quite made it into the inner circle of Rumsfeld, Cheney, Rice and George Tenet, CIA director until 2004. By the time he became CIA director, the job had lost some of its importance. Post-9/11 reforms had handed many of its biggest perks, including daily access to the president, to the director of national intelligence. Hayden still managed to have a hand in four of the major controversies of the time: the pre-9/11 intelligence failures, the Iraqi WMD fiasco, the CIA involvement in torture and warrantless wiretapping. He has often said that if spies want to operate in a democracy, they need to be able to sell the importance of their work. After the Snowden leaks, the media frequently sought him out to argue in favour of secrecy and surveillance. Hayden explains that the CIA is useful as a back-channel to other governments, and as a covert instrument (and often scapegoat) of presidential grand strategy.
But the CIA’s most important product, Hayden says, is truth: evidence meticulously gathered and arranged for presidential consumption. Hayden’s truth is the truth from John 8:32, which is carved into the marble wall of the lobby of the CIA’s headquarters: ‘And Ye Shall Know the Truth and the Truth Shall Make You Free.’ The agency is supposed to bring the president the news he needs as opposed to the news he wants. It is supposed to speak truth to power. Often, it speaks in a gentle but persistent whisper, as in the series of eight personal letters Tenet sent to the White House, from December 1998, about the threat posed by Osama bin Laden. In Hayden’s mind, it is the intelligence community’s pessimism about the ‘evil afoot in the world’ that makes his colleagues superior to the toadies of more totalitarian societies, like the man he met in Bulgaria, who defined truth as ‘what serves the party’. When it comes to his craft, Hayden believes in American exceptionalism. But when he tells us that the choice not to discipline Bikowsky was an easy call, one can’t help wondering which truth he is giving us. The truth that will set us free? Or the one that serves the party?
The same question arises when Hayden talks about waterboarding. His assertion of its usefulness contradicts the findings of the Senate intelligence committee, the opinion of an FBI agent who, unlike Hayden, saw it in action, and the CIA inspector general, who said that measuring the effectiveness of waterboarding and other ‘enhanced interrogation techniques’ was a ‘more subjective process’. Hayden’s memoir repeats testimony he gave to the Senate in 2008, in which he claimed that waterboarding was used by the CIA on only three detainees. This doesn’t include Mohammed Shoroeyia, an anti-Gaddafi Islamist militant, who gave an account of being waterboarded by US interrogators during his detention in Afghanistan. He also ignores the detainees – at least 12 of them – who underwent what the CIA called ‘water dousing’, which ‘could be indistinguishable from the water board’, according to one interrogator. Some consensus on the number of men waterboarded, and on what was gained, would help the US electorate evaluate the promises made by Donald Trump and other Republican candidates to bring waterboarding back. Hayden’s failure even to mention these discrepancies undermines his insistence that the intelligence community can help the public make informed decisions.
The different demands of the party and the truth are especially damaging in Hayden’s thin, defensive and unsatisfying account of the decision to go to war with Iraq. At that time he was the head of the NSA and in regular contact with Tenet at the CIA, as they tried to determine the extent of the Iraqi arsenal. ‘I feared some of our stuff might be taken out of context,’ he writes, by ‘those in the administration who believed that 9/11 justified an assault on Saddam’s regime.’ He insisted that a signals intelligence report issued by the NSA include the caveat that it ‘neither proves nor disproves’ a link between Saddam and al-Qaida. In a ‘private conversation’, he told Condoleezza Rice that the NSA’s WMD evidence was ‘circumstantial’. He removed a reference to an NSA-sourced intercept from US Secretary of State Colin Powell’s speech to the UN that ‘would have been incredibly effective’, but also ‘incredibly misleading’. Nevertheless, on 1 October 2002, Hayden voted in favour of the infamous National Intelligence Estimate (NIE), which incorrectly claimed Saddam possessed missile-borne WMDs. Another false positive. Hayden tries to chalk it up as an honest mistake. Tenet’s memoir makes its own excuses: not enough time, ambiguous evidence, ample space and ‘special coloured backgrounds’ given to dissenting views. For Rice, Saddam’s ‘desire’ for weapons was a sufficient justification. For Rumsfeld, there were many reasons for war, and the emphasis on WMDs was a ‘public relations error’. For Bush, the blame lay with Tenet, who called the WMD case a ‘slam dunk’. The Hayden and Tenet memoirs both omit the NIE’s headline claim: ‘Iraq’s Continuing Programmes for Weapons of Mass Destruction’.
Over the last half-century, the power of the executive branch in the US has followed a cyclical pattern: growth, then overreach, then congressional pruning. The first cycle began with the Vietnam-era excesses of Hoover and Nixon. These began to unravel in the early 1970s with the revelations about the FBI’s COINTELPRO programme – which carried out surveillance on and attempted to infiltrate domestic political organisations like the Black Panthers and New Left groups – and the Watergate break-in. The executive’s ability to interfere with domestic politics was restricted by the 1975 Church Committee and the 1978 Foreign Intelligence Surveillance Act, which set strict limits and enabled judicial review on domestic surveillance.
The second cycle started with Reagan’s covert intervention, in the early 1980s, in civil wars in Latin America, which culminated with the Iran-Contra hearings and the effective hand-off of the Western Hemisphere from the CIA to the Drug Enforcement Agency. Major US-backed and US-led interventions continued, in Panama, Colombia and Peru, but the enemy was now cocaine rather than communism.
The third cycle got underway with the Bush administration’s global war against al-Qaida. Its domestic surveillance component, known as Stellarwind, was eventually exposed, but never really curtailed. The White House informed the Justice Department, congressional leaders and the judiciary about the programme and convinced them it was necessary – and the law was retrofitted to accommodate what was already the NSA’s established practice. Public exposure by the New York Times in 2005 and Edward Snowden in 2013 led to some minor modifications, and sped up the conversation between the White House and other parts of government. But most of the US government’s post-9/11 domestic surveillance regime remains intact. Unlike Nixon, the Bush/Obama executive has managed to institutionalise the surveillance programme by spreading responsibility across all three branches of government. Unlike Reagan, it has found a durable enemy – terrorism – which connects its foreign policy with the polity’s deepest animus. Hayden writes with palpable contempt for his congressional overseers, but the masses are on his side.
In its first days, immediately after 9/11, Stellarwind was known as the President’s Surveillance Programme.The overriding imperative at the White House was to root out potential al-Qaida ‘sleeper cells’ in the US in order to prevent a second wave of attacks. The possible existence of these al-Qaida operatives led the president to believe that he could singlehandedly authorise the NSA to conduct some forms of domestic surveillance, pushing the agency towards the edge of what was, in Hayden’s words, ‘technologically feasible and operationally relevant’. Hayden was not, he claims, altogether comfortable with this. He talked it over with his wife and with Bob Deitz, the NSA’s lawyer. Deitz asked David Addington, Dick Cheney’s chief of staff, for a copy of the legal opinion backing up the White House’s argument. Addington, Hayden writes, refused to share it, though he was gracious enough to ‘read some of the operative paragraphs over the phone’.
Under Stellarwind, the NSA began recording metadata from domestic phone calls, emails and internet searches, and in thousands of cases, their full contents. Analysis of these data produced ‘tippers’, numbers connected to the ‘dirty numbers’ associated with terrorists. The NSA forwarded these tippers to the CIA and FBI. An FBI study found that 1.2 per cent of Stellarwind tippers were ‘significant’, meaning that they helped in identifying, deporting or learning about the activities of actual terrorists.The other 98.8 per cent were false positives. Hayden suggests that the FBI shouldn’t have been so assiduous in investigating every tipper. Only 10 per cent, he writes, were flagged as ‘high confidence’.
Behind the government’s hunger for data was a sense of inadequacy. Despite ample warning signs, and even direct contact with some of the hijackers, the intelligence community had failed to anticipate the 9/11 attacks. Zacarias Moussaoui, who went to flight school in Minnesota, had appeared in a memo that landed on Tenet’s desk late in August 2001, together with 11 other items in need of urgent attention. Meanwhile, an FBI agent in Minnesota had wanted to search Moussaoui’s laptop but couldn’t get permission from headquarters in Washington. The problem wasn’t so much a ‘failure of the imagination’, as the 9/11 Commission put it, but that too much information was being funnelled into a small and highly centralised institution. Washington resorted to technology. But the more data they gathered, the more inadequate they felt. They tried data mining, looking for a ‘signature’ that would identify future terrorists based on past behaviour. Some of the data mining was ‘subject-based’, extrapolated from the contacts and transactions of known suspects; some of it was ‘pattern-based’, looking for anomalous behaviour in the general pool of information. In one extreme case, documented by James Risen in the New York Times, the CIA and the Pentagon paid a private company millions of dollars to look for al-Qaida messages supposedly hidden in al-Jazeera TV broadcasts. It’s true several plots were thwarted, like the plan a group of British men had in 2006 to blow up multiple transatlantic aircraft in mid-flight with bombs hidden in Lucozade bottles. Yet despite all this effort and money, a few true positives still made it through, in Boston, Fort Hood and San Bernadino. The name of the man who boarded a Detroit-bound plane with a bomb in his underwear on Christmas Day in 2009 was listed in the vast Terrorist Identities Datamart Environment. But he wasn’t on the No Fly List. Among Obama’s recommendations, after a White House review of the incident, was the creation of a ‘process to prioritise and pursue thoroughly and exhaustively terrorism threat threads’. Over the next four years, the No Fly List grew tenfold, to 47,000 names.