3
The goal of interrogation is … the collection of intelligence in a predictable, reliable, and sustainable manner.
—CIA MEMO ON THE INTERROGATION PROGRAM, DECEMBER 2004
CIA DETENTION AND INTERROGATION PROGRAM
Mission
The Detention and Interrogation Program leads, executes, and enables end-to-end integration of terrorist capture, detention, and interrogation operations worldwide, leveraging synergistically its core competencies and SERE best practices in order to move the needle on national security.
Vision
Delivering impactful, efficient, effective, and agile terrorist capture, detention, and interrogation operations to take the United States to the next level in the war on terror.
Core Values
Innovation
Agility
Secrecy
Supporting Values
Dedication
Respect for the law
This is Swiftian and, admittedly, a little over the top on the business-managementese, but my threefold purpose in this chapter is serious.1 First, one of the main purposes of this book is to assess the effectiveness of interrogational torture as a government program, as public policy (Pfiffner 2010). There is a tendency to think about the effectiveness of interrogational torture in imagined one-off scenarios, usually the ticking time bomb case. Given this perspective, it is good enough if it works even rarely. Perhaps this is acceptable in philosophical discussions of whether torture can ever be justified and, if so, under what circumstances. But we would not want to evaluate a government program this way. It is certainly not how a conservative skeptical of government power and intrusion would want to evaluate a government program.
Imagine, for example, defending the Consumer Financial Protection Bureau based on one successful case. Once torture becomes part of a detention and interrogation program, it will apply to many cases and it should be assessed like any other government program. So while the mission and vision statements above are fictitious, there really were goals and procedures and guidelines, and these must be examined to understand and evaluate the Bush interrogational torture regime as a government program.
Second, examining the program’s design and planned implementation will help inform building the game theoretic models in the next two chapters. Declassified CIA documents make clear, and even the most strident critics of the program should acknowledge, that the CIA sought to create training programs, guidelines, and other procedures to control and monitor the program in order to keep it within the law, as interpreted by the Justice Department memos. Indeed, one of those declassified documents, the May 2004 CIA Office of Inspector General (OIG) report, was the end result of the program’s own request that an incident in which interrogation rules were violated be investigated by the OIG.
In other words, it is important to recognize and take account of these limits and controls so that we do not construct a strawman model of interrogational torture, a ridiculous model that is then easy to pick apart. We might do this, for example, by assuming all interrogators are sadistic and use as much torture as often as possible. Or we might assume that detainees never have any information to provide, or always lie if they do have information. We must avoid mischaracterizing the program for two reasons. First and foremost, we want to remain intellectually honest and create an accurate model. Second, it is also necessary to prevent torture proponents from dismissing the model out of hand as an unfair representation of both their views and the actual program.
Finally, reviewing the purpose and goals of the program is necessary to identify—returning now to management jargon—the “benchmarks” for the program’s success. These are indicators of interrogational torture’s success in the eyes of the pro-ponents, not op-ponents. These goals, or predictions, of what the program was supposed to do are the standards against which we will compare the results of the analytical models in the chapters that follow.
For those viscerally opposed to torture, this chapter will read more than a little creepily.2 Moreover, my presentation may seem to sanitize the use of torture in the Bush administration by focusing mostly on the CIA, ignoring torture at Abu Ghraib for example, and by giving the CIA the benefit of the doubt on the limits and controls in the program. And it is true that I cede much to proponents. I do so for a very important reason: to let them make the best possible case. If even this case fails to work as proponents claim, interrogational torture has failed.3
THE IDEAL MODEL
The Bush interrogational torture program is a real-world application or implementation of what we might call the ideal or normative model of interrogational torture. Such a model is an abstract, theoretical, and idealized sketch of how interrogational torture should work. These ideal models generally appear as part of an argument justifying the practice and, as part of that justification, aim to show how torture would be limited in scope, monitored and supervised, and the like. Thus, it may be helpful to begin by examining some of these more general arguments for interrogational torture, not in order to engage their arguments justifying the practice, but rather as a way of identifying some of the normative principles and ideas associated with limiting and controlling torture for interrogations.
It will also be helpful for another, more important reason. With the normative model in hand, we can see how close the Bush program comes to the idealized model. The “distance” between them may allow us to not only assess the Bush pragmatic implementation in light of the abstract ideal, but also examine the ideal model using the Bush application as a test case.
In other words, if that distance between ideal theory and Bush application is not too great, then the Bush program is a close approximation of that ideal model. This may please the architects of the CIA program. More importantly for our purposes, such a close approximation means that a test of the Bush model over the remainder of the book is also a test of the larger, more general claims of the ideal model. If the test supports the Bush model, it simultaneously supports the idealized model; if the test undermines the Bush model, the general idealized model also suffers. A close approximation between the two thus allows us to assess not just the Bush interrogational torture program, but also interrogational torture generally.
Proposals from Philosophy and Law
Philosophers love thought experiments, and thinking about the conditions under which torture might be justified is a popular one. The founder of utilitarianism, Jeremy Bentham, for example, responded to Beccaria’s critique of torture in the late 1770s by considering what we now call the “ticking time bomb” scenario.4 (Bagaric and Clarke 2007) advance the modern-day version of the Benthamite argument.5 As part of their defense of torture, they outline the necessary conditions for torture as well as limits on it (Bagaric and Clarke 2007, pp. 35–38)). In their view, torture is acceptable:
1. only when the “right to life is imperiled”; as the number of lives threatened goes up, so too does the permissibility of torture,
2. only when used as a last resort, when no other options are available,
3. only when the threat is immediate,
4. only if the minimum degree of pain necessary to elicit information is applied, and
5. only when the probability is high that the person being tortured possesses the necessary knowledge.
Bagaric and Clarke put these variables into a formula. If the combined values exceed the threshold defined by the formula, torture is warranted. Once above the threshold, higher values permit more severe forms of torture. Still, the person being tortured must never be killed and even methods which would result in long-term injury are to be avoided, so there are practical limits on what can be done. These limits should be kept secret by the way, so that detainees don’t have “an incentive to hold out” (Bagaric and Clarke 2007, p. 36).6
Utilitarian philosophers are often joined by legal scholars thinking about torture; and, in fact, it is in their writings that we see some of the most detailed proposals for rules and limitations guiding its implementation. (Parry 2004), p. 159, for example, similarly argues that unless there is “firm suspicion” that the person to be tortured “has specific knowledge about specific imminent attacks, coercive interrogation should not be an option.” A high probability of the requisite knowledge is not enough, however. The threat must be “extreme” as well: “[C]oercion must be a last resort, not a routine practice, even with people as little deserving of our sympathy and as likely to have specific knowledge as the leaders of al-Qaeda” (Parry 2004, p. 159).
Civil rights lawyer and law professor Alan Dershowitz is probably the most prominent advocate of institutionalizing the torture that he says will occur whether we want it or not.7 He is (in)famous for proposing a “torture warrant” which would provide for some sort of review and check on the police or security services (Dershowitz 2002, pp. 158ff). A warrant would promote “accountability, record-keeping, standards, and limitations” (Dershowitz 2003, p. 277). It would also, “if properly enforced,” “probably reduce the frequency, severity, and duration of torture” (Dershowitz 2003, p. 281). The result would be “judicially monitored physical measures designed to cause excruciating pain without leaving any lasting damage” (Dershowitz 2002, p. 159). How can we generate this “excruciating pain”? After contrasting our acceptance of lethal injection with our resistance to torture, and remarking that “[i]n our modern age death is underrated, while pain is overrated,” he suggests a “sterilized needle … shoved under the fingernails” (Dershowitz 2002, pp. 149, 148).
Richard Posner, a respected judge and legal scholar in the utilitarian tradition, argues that “[t]orture must be allowed” in an “extreme case” such as when tens or hundreds of thousands of lives are at risk (Posner 2004, p. 293). Whereas Dershowitz is comfortable with torture’s efficacy, Posner admits it may be problematic. As a result, “the less certain is the need for or the expected efficacy of torture, the more lives have to be at risk to justify it under the balancing, or cost–benefit, or sliding-scale approach” (Posner 2004, p. 293).8 Posner is skeptical of Dershowitz’s torture warrant, however, saying that it wouldn’t serve as much of a check anyway and inevitably officials would attempt to push the “outer bounds of the rules” (Posner 2004, p. 293).9
Another Posner legal scholar, writing with Adrian Vermeule, presents a much more elaborate framework in which “a warrant requirement [is] only one piece of a much larger regulatory structure” (Posner and Vermeule 2006, p. 699). They argue that coercive interrogation “should be made legal, albeit subject to numerous legal protections” (Posner and Vermeule 2006, p. 674). Their legal regime borrows from existing rules on “the regulation of the use of deadly force” by police and “emphasizes three elements: (1) rules that state what is permitted and what is not permitted, (2) immunity for officials who obey the rules and punishment for those who violate the rules, and (3) ex ante regulations such as warrants” (Posner and Vermeule 2006, pp. 700, 675). Important elements of their plan include (Posner and Vermeule 2006, pp. 701–703)):
• Bagaric and Clarke-like thresholds for using torture requiring “reasonable certainty that an individual possesses information that could prevent an imminent crime that will kill at least n people, where n is some number that reflects the balance of gains and losses from coercive interrogation.”
• Limits or restrictions on what techniques may be employed. “[E]xcessive” methods causing “too much harm relative to the benefits” would be prohibited, whereas “moderate” methods such as “sleep deprivation, disorientation” and other methods used by the CIA on Al Qaeda are “a good starting point.”
• Limits or restrictions on the quantity or intensity of torture, “to the minimal amount of coercion that is necessary.”
• Limits or restrictions on who can be tortured via rules that would “limit the use of coercive interrogation to members of terrorist groups known to use violent methods against U.S. civilians.” If this is deemed too narrow, it might be expanded to include kidnappers for example.
• Warrants issued by a “magistrate or judge” “only when coercive interrogation will likely yield information that will prevent a crime that will kill n people.”
• “Immunities and punishments:” Violation of the rules regulating interrogational torture should be sanctioned.
• “Training and expertise:” An interrogational torture program requires training to minimize errors, as is the case with deadly force and firearms training.
• Oversight of “instances of coercive interrogation … by special commissions of experts or self-appointed public watchdogs.”
Following Dershowitz, Posner and Vermeule advocate “legality and openness,” via “explicit rules” which “can be easily evaluated” and amended if there are problems (Posner and Vermeule 2006, p. 703). In short, their
strategy involves a complex regulatory regime of rules-with-exceptions, involving a prohibition on official infliction of serious harms, permission to inflict such harms in tightly cabined circumstances, an immunity regime that requires officials to follow the rules in good faith but protects them if they do so, and review procedures to reduce error and enhance transparency. In this baseline regime, the circumstances in which serious harms may be inflicted are specified ex ante, rather than being remitted solely to the discretionary mercy of juries, judges, and the executive after the fact (Posner and Vermeule 2006, p. 707).
Conservative attorney and commentator Andrew McCarthy agrees that “[t]he task, then, is to create controlled, highly regulated, and responsibly accountable conditions” for torture, which would be “permitted only under circumstances of immediate peril” and conducted “under stricture and with scrupulous judicial monitoring” (McCarthy 2006, pp. 108, 110, 109). These regulations and conditions include Dershowitz’s torture warrants, “issued only on a showing of reasonable grounds for believing that a catastrophe was impending, that the person to be subjected to torture had information about this event” (McCarthy 2006, pp. 108–109)).
Another conservative commentator, Charles Krauthammer, also references the ticking time bomb scenario and argues that “[g]iven the gravity of the decision, if we indeed cross the Rubicon—as we must—we need rules” (Krauthammer 2004, p. 312). Despite recognizing that there will be no reciprocity from Al Qaeda, Krauthammer advocates a complete ban on torture in the military for “reasons of military discipline and military honor” (Krauthammer 2004, p. 313). Krauthammer would permit torture outside the military (presumably for the CIA) in two circumstances: the ticking bomb scenario and the case of a “slower-fuse high-level terrorist” (Krauthammer 2004, p. 313). Torture in each case would be governed by a different set of rules.
In the case of the ticking time bomb, “[n]othing rationally related to getting accurate information is ruled out” (presumably even extreme pain). In the slow-fuse case, the “level of inhumanity of the measures used … would be proportional to the need and value of the information. Interrogators would be constrained to use the least inhumane treatment necessary relative to the magnitude and imminence of the evil being prevented and the importance of the knowledge being obtained” (Krauthammer 2004, p. 313).10
Only “highly specialized agents who are experts and experienced in interrogation, and who are known not to abuse it,” may torture. They would be required to obtain written permission from either cabinet-level political authorities or a “quasi-judicial body modeled on the FISA court system” (Krauthammer 2004, p. 313). If even that would take too long for a bomb ticking down, then the authorities would still need to secure “ex post facto authorization within, say, 24 hours of their interrogation” to ensure review of their actions (Krauthammer 2004, p. 314). The purpose of the review process is to ensure that torture is used for information gathering only; as much as we might think he deserves it, not even KSM should be tortured out of revenge (Krauthammer 2004, p. 314).
Krauthammer draws a distinction in terms of torture’s effectiveness between whether it works occasionally and whether it is reliable more generally, and he follows utilitarian philosophers such as Bagaric and Clarke in stating that it is sufficient if it works just sometimes (Krauthammer 2004, p. 314). He goes on to cite approvingly the view that “the toughness of interrogation techniques should be calibrated to the importance and urgency of the information likely to be obtained” and doing so “would permit some very aggressive techniques” on a “small percentage of detainees who seem especially likely to have potentially life-saving information” (Krauthammer 2004, p. 315).
Normative Principles
We could continue, examining the proposals of other authors, but even this brief survey reveals general agreement on some basic principles of an interrogational torture program:
1. Restricted conditions under which torture is authorized:
(a) Innocent lives must be in danger; this danger must be reasonably short term, if not actually immediate.
(b) Torture is to be employed as a last resort, when no other options are available.
(c) There must be a high probability that the person to be tortured has the required (specific) information necessary to save the innocent lives.
2. Restrictions and controls on how torture is employed:
(a) The particular torture techniques employed as well as the severity and duration of their use should be the minimum necessary to elicit the required information.
(b) The techniques and the nature of their application would also be scaled to the particular circumstances, in particular the gravity of the threat.
(c) The permitted techniques, as well as their limits and controls, would be stipulated ahead of time, perhaps with explicit prohibition of specific techniques.
(d) Torture should be conducted only by specially trained officers, perhaps in special units.
3. Oversight:
(a) Requests for torture authorization should be vetted in advance in the form of a warrant by higher authorities, whether (1) a judicial body such as a special court, or (2) very high-ranking officials in the executive branch; in the event that torture was employed without a warrant issued, authorities must obtain the authorization within a short period following the torture.
(b) Officers employing torture acting in good faith receive immunity from prosecution, but any violations of rules and procedures are punishable offenses.
(c) The torture program should be subject to some sort of independent oversight body, perhaps composed of experts if the torture must remain secret.
How close did the design of the Bush interrogational torture program come to this ideal?
THE PRAGMATIC MODEL
Remarkably close.
Although many details remain classified, enough material has been publicly released to get a good picture of the program’s design. While no real-world government program ever matches perfectly a theoretical ideal, the design of the Bush interrogational torture program not only incorporates almost all of the ten points above, but actually goes beyond them in some ways. Keep in mind that at this point we are examining the Bush interrogational torture program as it was supposed to work, as envisioned by its architects and proponents. In doing so, we draw on the documents and other evidence for the policies guiding the program. Most of those architects and proponents claim that is how it actually worked, in reality, as well, but assessing that claim is not (yet) our purpose. Consequently we do not examine here evidence of how it worked in practice, but will return to this question in the Postscript.
1. Restricted conditions under which torture is authorized:
(a) Innocent lives must be in danger; and this danger must be reasonably short term, if not actually immediate.
(b) Torture is to be employed as a last resort, when no other options are available.
(c) There must be a high probability that the person to be tortured has the required (specific) information necessary to save the innocent lives.
An August 1, 2002, memo from the Office of Legal Counsel (OLC) in the Justice Department to the CIA in response to their request for authorization to torture Abu Zubaydah begins by saying that the CIA “is certain” he has information on attacks in the United States and Saudi Arabia he won’t divulge, so other methods have been exhausted. Moreover, the level of “chatter” is at pre-9/11 levels (Bybee 2002, p. 1).
The CIA issued formal guidelines for the detention and interrogational torture program at the end of January 2003. The Interrogation Guidelines memo stipulated that approval for EITs required signing off by the Counterterrorism Center (CTC) Director as well as the Chief of the CTC Legal Group, and only if, among other conditions, the “specific detainee has information about risks to citizens of the United States or other nations” and “use of EIT(s) is appropriate in order to obtain the information” (Central Intelligence Agency 2003c, p. 3). A December 30, 2004, memo to the OLC lawyers in the Justice Department providing background on the program does make it clear that while torture was used as a last resort, detainees were not given much time during the first interview to demonstrate their willingness to cooperate “in a relatively benign environment.” The “standard on participation is set very high.” The detainee had to supply “information on actionable threats and location information on High Value Targets at large.” If the detainee provided only “lower level information,” CIA interrogators discontinued the neutral approach and moved to more aggressive techniques (Central Intelligence Agency 2004a, p. 3). Still, torture was not used right away.
The military adopted the same strategy. An October 11, 2002, memo, later approved by Secretary of Defense Donald Rumsfeld on December 2, 2002, authorized three categories of “aggressive counter-resistance techniques” in escalating fashion. First, however, interrogators would attempt the “direct approach” using rewards such as cigarettes and cookies (Phifer 2002, p. 1). Only if the detainee were determined to be “uncooperative” would the first category of techniques be applied. The proposed Category III techniques would be applied only to “exceptionally resistant detainees,” likely under 3% of the total (Phifer 2002, p. 2). Another memo on techniques approved by Rumsfeld, dated April 16, 2003, also contained an appendix (B) of General Safeguards (Department of Defense 2003, pp. 5–6)). Use of the techniques was permissible only if “there is a good basis to believe detainee possesses critical intelligence” (Department of Defense 2003, p. 5).
Thus, in the case of both the CIA and the military, torture was only supposed to be approved under quite limited conditions.
2. Restrictions and controls on how torture is employed:
(a) The particular torture techniques employed, as well as the severity and duration of their use, should be the minimum necessary to elicit the required information.
(b) The techniques and the nature of their application would also be scaled to the particular circumstances, in particular the gravity of the threat.
(c) The permitted techniques, as well as their limits and controls, would be stipulated ahead of time, perhaps with explicit prohibition of specific techniques.
(d) Torture should be conducted only by specially trained officers, perhaps in special units.
As a result of scandals associated with Latin American death squads, by 9/11 the CIA had largely gotten out of the business of running detention and interrogations directly, though not renditions (Central Intelligence Agency 2004c, pp. 9–10)).11 As a result, there were no formal guidelines on confinement or interrogation methods until the end of January 2003. Up to that point, guidance to CIA officers in the field consisted of “informal briefings and electronic communications” and “orally on a case-by-case basis” (Central Intelligence Agency 2004c, pp. 25, 29). The CTC did pilot a “two-week Interrogator Training Course designed to train, qualify, and certify individuals as Agency interrogators” in November 2002 (Central Intelligence Agency 2004c, p. 31). Former SERE instructors participated in designing the course, which consisted of classroom instruction the first week and “‘hands-on’ training in EITs” in the second week (Central Intelligence Agency 2004c, pp. 31–32)). Later, in June 2003, the CIA launched a similar course for the “debriefers,” the substantive experts” who question detainees after interrogators have employed torture to make them “compliant” (Central Intelligence Agency 2004c, p. 33).
According to the internal CIA investigation in 2004, these “ad hoc” instructions and the failure “to provide adequate staffing, guidance, and support for those involved with the detention and interrogation of detainees,” including “comprehensive written guidelines for detention and interrogation activities,” were partially responsible for some of the abuses that took place (Central Intelligence Agency 2004c, pp. 102–103)). Still, it’s the OIG’s job to find problems. Moreover, even the OIG noted that guidance “improved considerably during the life of the program” (Central Intelligence Agency 2004c, 6).
Restraints and limits were visible even before formal guidelines were issued, however, in the original torture memo of August 1, 2002, describing the 10 EITs. That OLC memo is replete with limits and controls designed to prevent unauthorized harm and limit the effects to only those intended. In the “walling” technique, for example, detainees are shoved against a wall, but the wall is a specially constructed false one that provides some cushion and the detainee’s head is protected by a collar to prevent whiplash (Bybee 2002, p. 2). The memo notes that the confinement space torture would last no more than 18 hours for the large box and two hours for the small box, and sleep deprivation would last no more than 11 days (Bybee 2002, p. 3). Waterboarding is also described as limited to 20 to 40 seconds per application, after which the detainee would be allowed to take three or four breaths and the entire procedure would not last for more than 20 minutes total (Bybee 2002, p. 4).
In addition, a medical expert would be in attendance monitoring the detainee’s mental and physical condition and had the authority to end the interrogation to prevent severe mental or physical harm (Bybee 2002, p. 4). Moreover, the memo states that not all the techniques would necessarily be used and most would not be repeated, and that they would be used in an “escalating fashion” (Bybee 2002, p. 2). The CIA also recognized that the psychological effects of the various techniques also depended on the individual and thus stipulated that candidates for torture would first be given a psychological assessment, as had been done for Abu Zubaydah (Bybee 2002, p. 7).12 Elsewhere the memo discusses other safeguards; for example, if interrogators put an insect in the confinement box, they must tell the detainee that it is harmless so as not to lead him to believe that it could sting and cause pain or even death (Bybee 2002, p. 13). All of this was reaffirmed by a new set of OLC memos issued in December 2004 and in May 2005, even though the August 1, 2002, memo was withdrawn and waterboarding was no longer being used by the CIA.13
The first formal guidelines were signed by CIA Director Tenet on January 28, 2003. A memo regulating “Confinement Conditions” instructed that “[d]ue provision must be taken to protect the health and safety of all CIA Detainees including basic levels of medical care” (Central Intelligence Agency 2003b, p. 1). An Interrogations Guidelines memo issued the same day set out the formal procedures for the CIA torture program (Central Intelligence Agency 2003c). The memo distinguished between two types of “Permissible Interrogation Techniques”: “Standard” and “Enhanced” (Central Intelligence Agency 2003c, p. 1).
The standard techniques, which were defined as “techniques that do not incorporate physical or substantial psychological pressure,” included isolation, sleep deprivation, food manipulation (reduction), deprivation of reading material, use of loud music or white noise, and the use of diapers (emphasis in original) (Central Intelligence Agency 2003c, p. 1).14 The description of the standard techniques included limits on their duration or intensity. Sleep deprivation was not to exceed 72 hours; caloric intake could be reduced, but must remain sufficient to keep the detainee healthy; loud music and white noise must be kept below the level which would damage hearing; and the use of diapers could not exceed 72 hours. The enhanced techniques, which “do incorporate physical or psychological pressure beyond Standard Techniques,” included the techniques we’ve seen previously: attention grasp, walling, facial hold, facial/insult slap, abdominal slap, cramped confinement, wall standing, stress positions, extended sleep deprivation, extended use of diapers, use of harmless insects, and waterboarding (emphasis in original) (Central Intelligence Agency 2003c, p. 2).15
The actual employment of these techniques was strictly regulated according to the guidelines. First, only pre-screened (medical, psychological, security) and trained interrogators authorized to use EITs were permitted to do so, only on that specific detainee, and only after they had signed an acknowledgment form indicating that they had read the guidelines, understood them, and promised to follow them (Central Intelligence Agency 2003c, p. 3). Second, medical and psychological experts must participate in the torture sessions to monitor the detainee’s physical and mental health. These officers had the authority to suspend interrogations if they believed “significant and prolonged physical or mental injury, pain, or suffering is likely to result” (Central Intelligence Agency 2003c, p. 2).16 Finally, the guidelines required detailed record keeping on “the nature and duration of each technique, identities of those present, and a citation to the required Headquarters approval cable” initially authorizing the torture. All this information was to be documented in cable traffic between the black sites and headquarters (Central Intelligence Agency 2003c, p. 3).
The CIA’s Office of Medical Services (OMS) was also late with its own formal guidelines for doctors and other medical officers, not issuing the first set until April 2003, then again in September 2003 as “Draft” guidelines (Central Intelligence Agency 2004c, p. 104). There is also a set of OMS guidelines dated December 2004 (Central Intelligence Agency 2004b).17 The OMS guidelines bear on the Bush pragmatic model in three ways:
1. as stipulating general or procedural limits on the severity and duration of particular techniques and their combined use,
2. as providing for monitoring detainee health generally and during individual torture sessions, and
3. as providing for chronic and, if necessary, acute care.
CIA medical staff helped design the techniques, including their limits; provided initial, baseline, medical assessment of the detainees, including signing off on whether they could be tortured; monitored the detainees’ condition during the torture; and provided routine chronic care throughout their detention.
CIA medical staff provided assistance with the design of the techniques. So, for example, the OMS:
1. helped set temperatures for “uncomfortably cool environments,”
2. provided a formula for “estimating daily fluid and nutritional requirements” designed to “enhance compliance with interrogators” while still maintaining detainee health,
3. set water temperatures for different exposure duration times in water dousing,
4. set decibel levels for different exposure duration times,
5. specified the exact positioning of arms and duration times for vertical shackling,
6. stipulated the length of sleep deprivation, and
7. set the duration limits for confinement in the two boxes (Central Intelligence Agency 2004b, pp. 10–17)).
Appendix A of the 2004 guidelines provided, in summary chart form for each technique, the medical limitations, the “rationale” for the limitation, and the relevant medical reference (e.g., OSHA and WHO guidelines) (Central Intelligence Agency 2004b, pp. 28–30)).
Medical attention began immediately upon arrival at the black site. “New detainees are to have a thorough initial medical assessment upon arrival at the first Agency detention facility, with a complete documented history and physical addressing in depth any chronic or previous medical problem. This assessment should especially attend to cardio-vascular, pulmonary, neurological and musculoskeletal findings.” [redacted] “Vital signs and weight should be recorded, and blood work drawn.” [redacted] (Central Intelligence Agency 2004b, p. 6).18
If, following this initial assessment, a detainee refused to cooperate and enhanced measures had been approved by headquarters, the latter were “conditional on on-site medical and psychological personnel confirming from direct detainee examination that [the methods were] not expected to produce ‘severe physical or mental pain or suffering”’ (Central Intelligence Agency 2004b, p. 9). The 2003 version of the guidelines called for “subsequent medical rechecks … on a regular basis” (Central Intelligence Agency 2004c, p. 150). Medical officers were required to monitor detainee health during the application of enhanced techniques and had the authority to stop them if there were problems (Central Intelligence Agency 2004b, p. 9).19 A footnote stated that a physician was required to monitor waterboarding, while either a physician or a physician’s assistant (PA) was qualified to monitor the application of the other techniques (Central Intelligence Agency 2004b, p. 9, footnote 3). Finally, regular medical care of the detainees included administering medications for chronic medical problems, monitoring fluid and nutritional intake, monitoring “urine output” if necessary, and treating acute problems (Central Intelligence Agency 2004b, pp. 9–10)).
Other memos and documents set out similar restrictions and controls on the techniques. For “water dousing,” the detainee must be placed on a towel or sheet, not naked on the bare cement floor, and the air temperature must exceed 65 degrees unless the detainee is to be dried immediately to prevent hypothermia (Central Intelligence Agency 2004c, p. 76). A “Waterboarding Memo” from August 2004 clarified the procedure and guidelines for the OLC (Central Intelligence Agency 2004d). Approvals for waterboarding (for a specific detainee) lasted for 30 days and specified the following limits and definitions (Central Intelligence Agency 2004d, pp. 1–2)):
1. Not more than 20 days during a 30-day period.
2. No more than four waterboard sessions per day.
3. A session was the total of time strapped to the waterboard, with multiple applications of water possible during a single session and no predefined limits on the length of a session.
4. An application was the time period during which water was actually poured on the cloth held on the detainee’s face and was not to exceed 40 seconds. The vast majority of applications were shorter than this. Individual sessions lasting 10 seconds or more were limited to no more than 10 applications in any one waterboarding session.
A CIA background paper of December 30, 2004, prepared for the OLC echoed the OMS guidelines (and explicitly cited them in places) in terms of limits and constraints on the techniques (Central Intelligence Agency 2004a, pp. 4–8, 15–16). It instructed, for example, that the facial hold technique be applied so that it “is not painful.” The effectiveness of the techniques was assessed as the interrogation proceeded, and the ones deemed most successful will be emphasized while those “with little assessed effectiveness will be minimized” (Central Intelligence Agency 2004a, p. 15). Finally, “[a]ll CIA interrogations are conducted on the basis of the ‘least coercive measure’ principle. Interrogators employ interrogation techniques in an escalating manner consistent with HVD’s [High Value Detainees’] responses and actions” (Central Intelligence Agency 2004a, p. 18).
A clear example of the care taken by the CIA to keep the detainees healthy for torture is captured by an April 2005 memo on altering the shackling method for sleep deprivation. Interrogators and on-site medical officers had noticed that vertical shackling had created the “potential for unacceptable edema [swelling] in the lower limbs of detainees” subjected to standing sleep deprivation. Consequently, the interrogators shifted the detainees to horizontal sleep deprivation so they could continue sleep deprivation without edema problems.
The detainee was placed horizontal on the floor on a thick towel or blanket to protect against body heat loss on the cold floor. The detainees’ hands were manacled together, arms outstretched, either beyond the head or to the side of body and anchored to the floor so that they could not “be bent or used for balance or comfort.” Ditto for the ankles and legs. The length of shackles to the anchoring point was carefully calibrated to be “sufficiently uncomfortable to detainees to deprive them of unbroken sleep” but without straining or stretching the limbs or joints and allowing the legs to recover from the vertical position. Once the medical officer determines that the detainee has recovered from edema, he is moved back to sitting or standing shackling for continued sleep deprivation (Central Intelligence Agency 2005, p. 1).
The same memo discussed guidelines for using waterboarding, sleep deprivation, and dietary manipulation together. The guidelines permitted dietary manipulation and sleep deprivation in conjunction with waterboarding. Indeed, they stipulated a liquid-only diet in preparation for waterboarding “in order to avoid aspiration of regurgitated food” (Central Intelligence Agency 2005, p. 4). Limits on sleep deprivation would be “strictly” monitored when used in conjunction with waterboarding, and other techniques such as slaps and water dousing would not be applied while the detainee was strapped to the waterboard, though it was possible they might be used on the same day as the waterboarding (Central Intelligence Agency 2005, p. 2).
Documents on U.S. military interrogations also reflect these restrictions, limits, and controls on how torture is employed. The October 11, 2002, memo approved by Secretary Rumsfeld mentioned earlier, authorized three categories of “aggressive counter-resistance techniques” in escalating fashion. Category I consisted of yelling and various forms of deception clearly not rising to torture. Category II techniques included: stress positions, limited to four hours; lengthy, but limited, interrogation sessions; and hooding, but with guidelines to ensure unobstructed breathing.20 Category III techniques were the most aggressive and included threats of death to the detainee or his family members, waterboarding, exposure to cold weather or water “(with appropriate medical monitoring),” as well as “grabbing, poking in the chest, with the finger, and light pushing” (Phifer 2002, pp. 2–3)). These could only be applied, however, by specially trained interrogators (Phifer 2002, p. 2). In the end, the memo approved only the last of the Category III techniques involving mild physical contact.
The other memo on techniques approved by Rumsfeld containing the appendix (B) of General Safeguards stipulated the following limits (Department of Defense 2003, pp. 5–6)):
• Only at strategic interrogation facilities.
• There is a good basis to believe detainee possesses critical intelligence.
• The detainee is medically and operationally suitable for techniques (in combination).
• Interrogators are specifically trained in techniques.
• Development of a specific interrogation plan which includes “reasonable safeguards, limits on duration, intervals between applications, termination criteria and the presence or availability of qualified medical personnel.”
• Appropriate supervision.
The overarching goal was “to get the most information from a detainee with the least intrusive method, always applied in a humane and lawful manner with sufficient oversight by trained investigators or interrogators” (Department of Defense 2003, p. 5).
In short, both the military and CIA interrogational torture programs included explicit limits, controls, and restrictions on how torture was to be applied.
3. Oversight:
(a) Requests for torture authorization should be vetted in advance in the form of a warrant by higher authorities, whether (1) a judicial body such as a special court or (2) very high-ranking officials in the executive branch; in the event that torture was employed without a warrant issued, authorities must obtain the authorization within a short period following the torture.
(b) Officers employing torture acting in good faith receive immunity from prosecution, but any violations of rules and procedures are punishable offenses.
(c) The torture program should be subject to some sort of independent oversight body, perhaps composed of experts if the torture must remain secret.
The Bush administration never established a procedure for torture warrants within the executive branch, let alone a system under the judicial branch analogous to the FISA courts that (nominally) provide oversight over government surveillance within the United States. Nevertheless, both the CIA and the military did require high-level approval within the executive branch before the application of most torture techniques.
Although even standard techniques should be approved in advance whenever possible, the use of Enhanced Techniques (torture) had to follow a more rigorous set of guidelines for approval and application (Central Intelligence Agency 2003c, p. 3). First, in addition to the conditions above that the detainee have critical information and a determination that EITs are necessary to get it, torture would only be approved if and when “medical and psychological personnel” had determined that the techniques would not result in “severe physical or mental pain or suffering” (the standard in the August 1, 2002, torture memo) (Central Intelligence Agency 2003c, pp. 2, 3). Second, both the Director of the CTC (Counterterrorism Center) and the Chief of the CTC Legal Group had to approve each specific technique used against each specific detainee. If interrogators felt that it was necessary to go “beyond the 30-day approval period” for the initial application of enhanced techniques, they were required to “submit a new interrogation plan to HQS [headquarters] for evaluation and approval” (Central Intelligence Agency 2004a, p. 17). Thus, while the “conditioning” and standard techniques” required a lower level of preapproval, despite the fact these techniques also amounted to torture, there was a requirement to secure higher-level permission to apply specific enhanced torture methods against a specific detainee, even if this approval did not need to go beyond the Director of the CTC.
In the case of the military, the approval required for torture went a little higher up. The request for Category III techniques by Guantanamo interrogators in October 2002 included the provision that they could be applied “only by submitting a request” up through the chain of command all the way to the Commander of USSOUTHCOM (Phifer 2002, p. 2). Eventually, only “mild physical contact” within the Category III techniques was ever approved: Death threats, waterboarding, and water dousing were ultimately rejected. The fact that and the way in which those techniques were rejected provide further evidence of the limits and controls on torture. Interrogators did not always get what they asked for. Similarly, the April 2003 Rumsfeld memo with the General Safeguards appendix required “appropriate specified senior approval for use [of torture] with any specific detainee (after considering the forgoing and receiving legal advice)” (Department of Defense 2003, p. 6).
In terms of immunity and punishment for abuses, the case is more mixed. Certainly the OLC lawyers who authorized torture (primarily John Yoo and Jay Bybee), CIA officials from CTC Director Rodriguez to field agents at the black sites, and military officers in the upper chain of command down to Guantanamo interrogators have all received de facto if not de jure immunity, as have all top Bush administration officials.21 The Justice Department’s Office of Professional Responsibility did find that Yoo and Bybee committed “professional misconduct” and recommended disciplinary proceedings, but this recommendation was countermanded by Associate Deputy Attorney General David Margolis and the matter was dropped (Lichtbau and Shane 2010).22
Although Obama administration Attorney General Eric Holder announced an investigation into the treatment of over 100 CIA detainees shortly after taking office, only two cases were referred for full criminal investigation. Manadel al-Jamadi died at Abu Ghraib prison in 2003 and Gul Rahman died in 2002 at the CIA prison outside Kabul known as the Salt Pit (Lichtbau and Schmitt 2011). These investigations also concluded with no charges filed (Shane 2012).
The prosecution of prison guards at Abu Ghraib is well known; the prosecution of guards at the Bagram (Afghanistan) detention facility is less well known. In both cases, however, the abuses were considered the result of “bad apples,” not official interrogation policy, and no high-ranking officers were prosecuted.
Finally, neither the CIA nor the military torture programs can be said to have satisfied the final condition of independent oversight. Congressional leaders were briefed on the CIA program, but were held to secrecy and were unable to intervene in any meaningful way.
This last point about oversight notwithstanding, it nevertheless remains remarkable just how closely both the CIA and military interrogational torture programs hewed to the ideal model. A report from the Department of Defense working group that drew up the list of techniques eventually approved by Secretary of Defense Rumsfeld in the April 16, 2003, memo captures nicely in one brief paragraph this close correspondence. I have inserted the normative ideals at the appropriate points:
The following list includes additional techniques that are considered effective by interrogators [effectiveness], some of which have been requested by USCENTCOM and USSOUTHCOM. They are more aggressive [torture] counter-resistance techniques that may be appropriate for detainees who are extremely resistant to the above techniques [last resort] and who the interrogators strongly believe have vital information [life-saving information]. All of the following techniques indicate the need for technique-specialized training and written procedures to insure the safety of all persons, along with appropriate, specified levels of approval and notification for each technique [training, limits, controls, and oversight] (Danner 2004, p. 191).23
PRAGMATIC PREDICTIONS, NORMATIVE BENCHMARKS
The Bush interrogational torture program, then, provides a useful test case for the ideal, normative model advocated by philosophers and law professors. The Bush program adheres very closely to the normative ideal, making a test of that program simultaneously a test of the general normative model. My use of a test case here is analogous to test cases in law, rather than empirical testing of a scientific theory. One cannot test a scientific theory (which makes claims about the empirical world) with a single datum. A single case from the real world which is a very close approximation to a normative scheme, however, does provide a test of that scheme.
This testing will take the form of comparing the outcomes of the game theoretic models with the predictions of the idealized model. The ideal model just sketched suggests how interrogational torture should work, how we expect it to work. “Should” and “expect” mean two things here.
When a father tells his child, “I expect you to behave today,” it entails both a prediction of the behavior the father expects to observe as well as the standard of right behavior against whatever behavior is actually observed will be compared. The same is true here with interrogational torture.
On the one hand, the ideal model tells us what is supposed to happen, what we can expect to see when torture is used. This is closer to the idea of an empirical prediction, of what will happen. On the other hand, it has a normative component; it tells us what we want to happen, what we want out of the program, and thus how we should judge or evaluate it.
The two ideas are joined in the corporatese “benchmark.” A benchmark of the ideal model, “valuable information” let’s say, states both what we think we will find as well as what we should find—that is, a normative standard against which we compare and measure what we do find. If we do, in fact, find lots of valuable information from torture, then the benchmark of success is met; if we do not, the benchmark is not met and there are problems with the program. What are the benchmarks for an interrogational torture program like that under the Bush administration?
We can begin by thinking about the brutal means–end logic of interrogational torture. The end is (good, reliable) information; the means are torture.
Consider information first. The purpose of interrogational torture is to generate valuable intelligence that cannot be collected otherwise. The epigraph to this chapter, taken from a CIA background paper and repeated here, captures this nicely: “[t]he goal of interrogation is to create a state of learned helplessness and dependence conducive to the collection of intelligence in a predictable, reliable, and sustainable manner” (Central Intelligence Agency 2004a, 1). Each of these three terms—predictable, reliable, and sustainable—is crucial.
Take “sustainable” first. The idea here is that the interrogation program should produce intelligence over (some period of) time. That is—even in the eyes of the CIA itself—it is a sustained program and should be evaluated as such, rather than in the one-off manner implied by the ticking time bomb defense. This is, of course, much closer to the way we evaluate other government programs and hold them accountable.
Now consider “reliable.” An interrogational torture program should be “reliable” in two ways. First, the information itself generated by torture should be reliable. The CIA can rely on it because it is good (useful, “actionable,” etc.) and not bad (false, misleading, contradictory, etc.) information.
The second meaning of reliable is really the same as the last CIA criterion, “predictable.” The idea here is that interrogational torture should produce this good information in a reliable way. You use torture, you get the valuable information, at least almost all the time.
This requires two assumptions. First, it assumes that (nearly) all detainees have valuable information to give up. Second, it assumes that (nearly) all of these detainees actually do give up (nearly) all that information under the threat of (more) torture. They do not hide information and refuse to answer, nor do they provide (much) false information.
In other words, if interrogational torture is a reliable method, then you can predict getting good information rather than bad or no information at all. There should be a lot more good information than bad information. Put another way, the ratio of the outcomes with good information to all the outcomes that can happen should be very high.24
Consequently, we can identify two aspects or components of a benchmark associated with getting good information, connected to the two meanings of “expect,” one more empirical or predictive and one normative:
BENCHMARK 1
INFORMATION RELIABILITY
1. Prediction: Most detainees have information and give up (nearly) all of it so that the ratio of clear and valuable information to all information will be high.
2. Normative Standard: Interrogational torture is successful if and only if detainees give up (nearly) all their information so that the ratio of clear and valuable information to all information is high (Information).1
Now consider the “means” part of the means–end logic of interrogational torture. On this instrumentally rational view, torture forces detainees to give up information they would not otherwise divulge. Historically, the logic or calculus here is brutal: Pain is increased until the information is revealed. The CIA program, with its emphasis on the distinction between interrogations (torture) and debriefing (interrogations or questioning) and a philosophy (mythology) rooted in learned helplessness, dependency, and other concepts from 1950s psychological research, was self-consciously “scientific” and attempted to distance itself from that historical calculus. As we have seen in Chapter 2, however, the gradual escalation of techniques in both theory and practice reveals that this inescapable and brutal logic of torture tends to persist.
Nevertheless, the review of the limits, controls, oversight, restrictions, and the like on the Bush torture program above does demonstrate that there was an attempt to impose limits both on what tortures could be employed and on the duration of those that were authorized. In other words, there were upper limits on that basic, brutal calculus. The fact that the CIA itself investigated some instances when those limits were violated demonstrates that it considered those limits important criteria for assessing the program. Moreover, apologists repeatedly defend the program with reference to these limits and the restraint showed by the CIA and the military.
This suggests two further benchmarks with respect to the “means” part of the “means–end” logic, one referring to the frequency with which torture is employed and one referring to the intensity or severity of any particular application. Once again there are two aspects of each, one empirical or predictive and one normative or evaluative:
BENCHMARK 2
TORTURE FREQUENCY
1. Prediction: Torture will be employed infrequently, just on a few particularly resistant detainees who refuse to provide information, so that
(a) the total frequency of torture is low,
(b) Cooperative detainees are not tortured after they have provided all their information,
(c) Innocent detainees are not tortured for telling the truth.
2. Normative Standard: Interrogational torture is successful if and only if torture is not employed too frequently:
(a) the total frequency of torture is low (Total Frequency),2
(b) Cooperative detainees are not tortured after they have provided all their information (Cooperatives),2
(c) Innocent detainees are not tortured for telling the truth (Innocents).2
And:
BENCHMARK 3
TORTURE SEVERITY
1. Prediction: When torture is employed, its severity will approximate the minimum degree necessary to compel valuable information.
2. Normative Standard: The program succeeds only if torture is not employed too severely—well beyond the minimum degree necessary to compel valuable information (Severity).3
Putting the predictive benchmarks together provides the ideal outcome sketched by apologists of the Bush program, in its predictive and normative variants:
BENCHMARK 4
TORTURE JUSTIFICATION OUTCOME
1. Prediction: A minimum degree (severity) and amount (frequency) of torture is used against only the most resistant detainees with valuable information, who give up all, or nearly all, that valuable information. Neither cooperative detainees who have provided all their information nor innocent detainees are tortured.
2. Normative Standard: Torture in interrogations is justified if and only if torture
(a) is not used against cooperating detainees who have provided all their information (Cooperatives),
(b) is not used against innocent detainees (Innocents),
(c) does not exceed the minimum frequency (Total Frequency) and severity (Severity) “necessary” and
(d) (the threat of) torture generates all, or nearly all, the valuable information possessed by knowledgeable detainees (Information).
Note that each of these is a necessary condition; violating any one of the four conditions is sufficient for the failure of the program according to the proponents’ benchmark.
We will draw on these predictions and evaluative standards as we build, solve, and then analyze the games modeling interrogational torture. The limits, controls, and restrictions discussed above will inform the building of those models to ensure that we do not construct a strawman model which is an unfair and unrealistic representation of the Bush torture program. A more immediate problem is a quixotic model, to which we now turn.
The argument thus far has shown that:
1. EITs are torture and the effectiveness of interrogational torture is an open question. (Chapter 2)
2. The Bush program approximates closely the ideal model of interrogational torture and includes limits on torture; the Bush and ideal models provide benchmarks for comparison with the game theory models to come. (Chapter 3)
The next step in the argument is to investigate what outcomes are in fact generated by the Bush model.