10

Valuable Information, Selective Torture

Ennigkeit’s … forceful and intensive threat … caused [Gäfgen] to fear even harsher measures. He did not want to expose himself to those ….

—GERMAN COURT DOCUMENT IN THE CASE OF MAGNUS GÄFGEN

There are three additional pairs of equilibria in the RIT game, one each under objective and leading questioning. They are “pairs” because each member of the pair calls for the same set of actions in both questioning types, but the beliefs supporting those actions differ across the questioning types. Since substantively they call for the same set of observable actions, we will consider each pair together. This chapter considers the first of these pairs: the valuable information, selective torture equilibrium. The next chapter examines the two pairs in which no information is provided, one in which the detainees are tortured and the other in which they are not.

EQUILIBRIUM FEATURES

In the valuable information, selective torture equilibrium, the Cooperative Detainee provides information satisfying the Interrogator, who therefore does not torture afterward. Just as in the valuable information, surprise torture equilibrium, the Cooperative Detainee is confident both that the Interrogator is pragmatic () and that the Interrogator will believe that the Detainee has told everything he knows () and so decides to play “information.” In the leading questioning version, the Innocent Detainee also has the move “information” but does not believe that it is sufficiently likely that the Interrogator is Pragmatic, and thus refuses to lie and confirm what the Interrogator wants to hear.

The Interrogator chooses not to torture after “information” in this equilibrium because she believes that the Detainee has in fact revealed everything he knows (, under objective questioning) or she wants him to confirm (, under leading questioning). The Interrogator tortures after “no information” because she thinks that it is unlikely that the Detainee is Innocent rather than Resistant when a Detainee fails to reveal anything valuable (). In other words, the equilibrium depends on torturing an Innocent Detainee for telling the truth of his innocence.

Thus, the only differences between the objective and leading question variants of this equilibrium are the addition of the Innocent Detainee’s beliefs and the change in the Interrogator’s information hiding threshold from under objective questioning to under leading questioning. The latter change is due to the fact that, under leading questioning, the information clarity parameter u drops out from the Interrogator’s payoffs and so the resulting threshold derived from them. That makes it identical to the Detainee’s version under objective questioning, .

Figure 10.1 presents the equilibrium in the parameter space, with the dots again representing the region covered by torture. The thresholds for both questioning variants are in their usual places, but the gray-shaded region in the northwest corner is the equilibrium under objective questioning only. In the leading questioning variant, the shaded volume would expand to the left, colonizing the region between and but still staying north of and to the front of . Given that the pragmatic justification for interrogational torture rests on the claim that it provides actual intelligence—that is, new, valuable information unknown and unknowable prior to questioning and torture—we concentrate on the objective variant of the equilibrium in what follows. Note that under objective questioning the dots fill the entire volume to the front of , including the valuable information region, because even in that region an Innocent Detainee is tortured for telling the truth.

Figure 10.1 Valuable Information, Selective Torture Equilibrium

Information Revelation Threshold

The equilibrium is bounded by the Cooperative Detainee’s information revelation threshold , the information hiding threshold , and the Innocent Detainee recognition threshold . Take the information revelation threshold first. As drops, as the Cooperative Detainee’s willingness to talk increases, the equilibrium region expands, with “information” occupying more volume and “torture” (of a Cooperative Detainee) occupying less volume. Conversely, a Cooperative Detainee with a higher threshold is less willing to divulge information, even under threat of torture, and so the equilibrium volume shrinks, opening up more space for torture of a Cooperative Detainee.

So we have four elements at work here: the value of the Cooperative Detainee’s information v, the intensity or severity of the torture k, the size of the region occupied by (and so the likelihood or frequency of) information (of some quality) above , and finally, the complementary region on the q axis below , which is the size of the volume taken up by (and thus the frequency of) torture of a Cooperative Detainee who does not provide information. Considering only this q dimension, then, and holding the amount of torture constant, an expansion of the equilibrium region along the q axis perforce means less valuable information. That is, for a given level of torture (k), and , , and held constant, the only way for the region to expand, for the equilibrium to occupy a greater proportion of the parameter space, is for the value of information v to drop. If we hold v constant instead of k, then the only way for the region to expand to occupy a greater proportion of the parameter space and thus be more likely is for the costs (brutality) of torture k to increase. In other words, torture must become more severe. There are, in other words, multiple trade-offs:

1. As information value increases,

(a) the likelihood of getting information decreases and

(b) the frequency of torture increases

2. As information value decreases,

(a) the likelihood of getting information increases and

(b) the frequency of torture decreases

3. As torture severity increases,

(a) the likelihood of getting information increases and

(b) the frequency of torture decreases

4. As torture severity decreases,

(a) the likelihood of getting information decreases and

(b) the frequency of torture increases

Trade-off 1a and its complement trade-off 2a are restatements of Implication 7.2: Everything else being equal, the more valuable the information possessed by the Cooperative Detainee, the higher the value of and the less likely he is to reveal it. The less valuable the information, the lower the value of and the more likely he is to reveal it. As we said in Chapter 7, this is of course true for interrogations without torture. The point bears repeating that torture does not do any better.

Now consider trade-offs 1b and 2b. Higher-value information (a higher v) pushes up and thus the lower bound of the equilibrium. As it does, as the darker-shaded region moves toward the top, the volume below it is replaced by torture of the Cooperative Detainee. (Remember that an Innocent Detainee as well as a Resistant Detainee is tortured in the entire space to the front of .) Thus, as accords with intuition, more valuable information is accompanied by more torture.

Finally, take a look at the second set of trade-offs 3 and 4 associated with torture severity. Consistent with the assumption granted to proponents, namely that torture compels information from a Cooperative Detainee, more brutal torture means that information is more likely, whereas less brutal torture reduces that likelihood (trade-offs 3a and 4a). Notice, though, that there is also a less obvious trade-off. All else being equal, an increase in torture severity means less frequent torture of Cooperative Detainees (i.e., more space occupied by information, less by torture), while a decrease in severity results in less information and more torture (trade-offs 3b and 4b). But this means that the pragmatic goal of infrequent and minimally brutal torture in obtaining information is unobtainable, with the following important implication:

Implication 10.1 (Torture Frequency–Brutality Trade-off). Everything else being equal, increasing the likelihood of information requires either less frequent torture or less brutal torture; it is not possible to minimize both.

Putting these multiple but related trade-offs together, we have the following implication:

Implication 10.2 (Information Value–Brutality Trade-off). Everything else being equal, eliciting information is more likely (the equilibrium region expands to a greater proportion of the parameter space) only if the information becomes less valuable or the torture becomes more brutal, not both.

Information Hiding Threshold

Now take a look at . The lower the value of , the more room there is for an f satisfying the Interrogator, making her willing to forgo torture, but also increasing the possibility of the Detainee getting away without revealing some information. That is, the more “lenient” she is, the lower her expectations of what she expects the Detainee to reveal to her, the larger the equilibrium region, and the less torture is employed. The more suspicious the Interrogator is that the Detainee has not revealed all of his information, the higher the value of ; that is, the more it moves to the right and the less likely it is that there is an f exceeding it. In other words, the less lenient she is, the higher her expectations for what counts as “cooperation,” the smaller the equilibrium region, and the greater the space supporting torture.

Imagine grabbing and sliding it to the left (i.e., lowering it). As you did, the gray-shaded region would grow and the region to the left would shrink, resulting in:

1. an expansion of the information region

2. more information hiding

3. less torture of a Cooperative Detainee

Now slide it back to the right. As you do, the gray-shaded region shrinks and the region to the left grows, resulting in:

1. a reduction of the information region

2. less information hiding

3. more torture of a Cooperative Detainee

There is thus another set of trade-offs, this time between the frequency with which information is obtained, the frequency of information hiding, and the frequency of torture:

Implication 10.3 (Information–Torture Trade-off). Everything else being equal, eliciting information is more likely (the equilibrium region expands to a greater proportion of the parameter space), and torture of a cooperative detainee less frequent, only if the standard of detainee cooperation is lowered, thereby increasing information hiding.

From Proposition 7.3 in Chapter 7 we know that is greater than or equal to one-half; the Interrogator will have a high threshold. The point at which she says to herself “ok, that’s all he’s got, I can stop now” will be a high hurdle. This pushes the equilibrium region to the right, shrinking it along the f axis.

Innocent Detainee Recognition Threshold

Finally, consider , the Interrogator’s Innocent Detainee recognition threshold. If the updated probability the Detainee is Innocent upon observing “no information” is above , the Interrogator chooses “not torture.” For probabilities below , the Interrogator chooses “torture.” The valuable information, selective torture equilibrium requires the latter. Thus, the higher this threshold (the more the Interrogator thinks that a Detainee who fails to provide information is Resistant, not Innocent), the greater the volume occupied by both “information” and torture of an Innocent Detainee in this equilibrium. As goes down, as the space supporting the belief that a “no information” Detainee is Resistant shrinks (i.e., the space supporting the Detainee is Innocent grows), both the “information” and “torture innocents” regions shrink toward the face of the parameter space. In other words, it is not possible to have one without the other:

Implication 10.4 (Information–Torture of Innocents Trade-off). Everything else being equal, an increase in information is inevitably accompanied by an increase in innocent torture; a decrease in innocent torture is accompanied by a decrease in information.

From Proposition 7.2, is less than one-half, restricting the equilibrium region to something under the front half of the parameter space. The lower the prior probability p_R, the closer the value of to 0, leaving less room for “information” and “no torture” of a Cooperative Detainee. The more the Interrogator believes everyone breaks—as is likely—the lower the value of .¹

In other words, the assumption that real-life interrogators will believe that almost no innocent detainees end up manacled to a chair in a secret prison (i.e., p_I is vanishingly low) and that “everybody breaks, bro” (i.e., p_R is very low and p_C is high) decreases the volume of the equilibrium region. We have, in other words, a strange and even paradoxical implication:

Implication 10.5 (Torture–Information Paradox). Everything else being equal, the very assumptions driving an interrogational torture program—the most likely detainee has information and will reveal it under (the threat of [more]) torture, very few detainees cannot be “broken,” and the chances of a detainee being innocent are virtually zero—all make valuable information less, not more, likely.

With these general considerations in mind, we now turn to the case studies. We noted above that this equilibrium covers two cases in the real world, one in which the detainee hides some information, but it is enough to satisfy the interrogator, and one in which the detainee truly does reveal all of his information and satisfies the Interrogator. We consider examples of each.

INFORMATION HIDING: PACHA WAZIR

For a case in the real world to match the information hiding outcome pertaining to a Cooperative Detainee, it must meet the following conditions:

1. The detainee was cooperative (i.e., had information).

2. The detainee was subjected to objective or leading questioning.

3. The detainee gave up some but not all of his valuable information.

4. The cooperative detainee was not tortured afterward.

The ideal case illustrating this equilibrium is one in which there is a Detainee known to possess some valuable information. He is not tortured, but knows that torture is likely if he does not give up his information. He is questioned, without torture, and provides information. The Interrogator believes that the Detainee has provided all the information he has and so does not torture him. We know in hindsight that, in fact, he was hiding still more information which he never divulged.

These criteria, particularly the last one, make it difficult to find a case in the public domain which fits this bill perfectly. But we can come close with the story of an Afghan banker named Pacha Wazir.²

Wazir, known also by the honorific Haji Wazir since he had made the pilgrimage (haj) to Mecca, came from one of the most prominent families in Jalalabad, Afghanistan. Wealthy, portly, and sporting a “fastidiously trimmed beard,” he maintained a “palatial” home in Dubai and was married with seven children (Suskind 2006, p. 160). Wazir ran a successful currency exchange business, several banks, and wire transfer stations, as well as a chain of informal money transfer stations called hawalas with his brother. His operations covered South Asia and Europe and he maintained additional offices in Dubai and in Pakistan.

One of his clients was Al Qaeda.

At least that was what both Canadian and U.S. intelligence suspected. The Canadian Security Intelligence Service (CSIS) linked a terrorist suspect they were monitoring, Mohamed Harkat, to Wazir (Canadian Security Intelligence Service 2009, Federal Court of Canada 2010). As for the United States, Wazir was considered a potential source of terrorist financing by several government agencies and departments. The FBI had asked banking authorities in Dubai, whom they were helping to set up tighter auditing and other financial monitoring and controls, to look into Wazir. The information that came back in the summer of 2002 was alarming: The FBI believed that Wazir “was responsible for handling a startling $67 million in assets for al Qaeda in just over two years” (Suskind 2006, p. 145). Nevertheless, the FBI did not feel it had enough evidence of laundering through U.S. banks and institutions in order to prosecute him in the United States.

The CIA, however, had a back-up plan for Wazir: “to own him” (Suskind 2006, p. 146). The FBI was told to back off while the CIA prepared its plans. Those plans were moved up abruptly by the diligence of United Arab Emirates (UAE) central bankers, who froze Wazir’s assets. Wazir inquired why; and when he was told the FBI was investigating him, he traveled to Dubai on his own accord to meet with them and clear up what he claimed was a misunderstanding. The FBI delayed him a day, enough time to get in a CIA snatch team.

The CIA team grabbed him in his own driveway as he was about to leave for his meeting with the FBI. They kept him in the U.A.E. for the initial interrogation, but he wasn’t talking. To up the pressure on him and see if they could get information out of his brother, the CIA kidnapped Wazir’s brother in Germany and spirited him out of the country. Neither, however, provided any information. It wasn’t long before Pacha Wazir was on his way to a CIA black site outside Rabat, Morocco.³

Wazir, of course, didn’t know he was in Morocco. He must have known he was somewhere in North Africa, given the Arabic accents and the occasional French he heard spoken around him. But he discovered quickly that the main interrogator was an American. The American was unhappy with Wazir’s answers and shouted at him in between coughing fits.

Soon, though, the American was replaced with another European-looking interrogator who spoke Arabic and French. “Jacques,” as the man introduced himself, was actually Glenn Carle, a career CIA case officer who had been sent to work with the Moroccans and interrogate Wazir.

At his first interrogation of Wazir, Carle found him “terrified,” though he had not been tortured (Carle 2011, pp. 69, 70). As we have seen, in mid-2002, the full-blown EIT program, with its conditioning, standard, and enhanced techniques, had not yet been established. Thus, though Wazir was no doubt hooded and shackled during his rendition flight, he had apparently not been subjected to the full panoply of conditioning that would become standard in short order: sleep deprivation, nakedness, stress positions, etc.

Still, Wazir would have been told that no one knew where he was and he would have known the reputation for torture that most of the intelligence services had throughout North Africa and the Middle East. Thus, although he had not been tortured, he certainly feared it could happen. Though Carle himself opposed torture, and never used it during his interrogation of Wazir, he did not disabuse Wazir of this possibility. In other words, Wazir knew that torture was a possible “move” by his interrogators, even if it might not be Carle doing it himself.

Carle recalls that when he first started working on the Wazir case, he found that “the assessment that was made of [Wazir] was quite compelling and … accepted it … I knew my colleagues to be hard-working and careful and that they reviewed their assessments regularly and the assessment was that [Wazir] was one of the top players in Al-Qaeda” (Leopold 2011). “He had been involved in activities of legitimate concern to the CIA, because they did touch upon al-Qa’ida activities. That’s a fact” (Ackerman 2011). Consequently, in Carle’s view, “[a]t the moment he was rendered it seemed the right decision to interrogate him and bore down into his activities. Huge effort had gone into his case, and the information appeared strong. It looked as though we had a chance to strike a truly damaging blow to Al Qaeda” (Horton 2011a).

To this day, Carle believes that Wazir “was not a random individual who knew nothing … he knew information relevant for our counter-terrorism operations—he was not a complete innocent” (Horton 2011a). Certainly the CIA and other government officials (FBI, Treasury) believed that Wazir possessed valuable information. In short, it appears fairly clear that Wazir was no Innocent Detainee; the question was whether he would cooperate or not.

Carle attempted to induce cooperation using the same techniques he had used as a case officer to get people to betray their countries: a combination of rapport, assessing Wazir’s motivations, and manipulating, stressing, and disorienting him (verbally) (Carle 2011, p. 67). The goal was to help Wazir reason his own way to wanting to cooperate. Although he was initially a little ambivalent about whether some of the disorienting methods were torture (e.g., disrupting sleep patterns), Carle ruled out physical torture from the beginning and the “milder” disorienting methods early on (Carle 2011, pp. 26, 67, 138–139). As Carle later put it in an interview, “from my first second of involvement in the [Wazir] operation I simply would not allow or have anything to do with any physical coercive measure. I would not do it” (Ackerman 2011).

This did not prevent Carle from “keep[ing] him fearful that [Carle] controlled his fate” (Ackerman 2011). Indeed, in places he went further. He once warned Wazir that “others were not as nice, or as patient, as [Carle] was” (Carle 2011, p. 74).

The result was, according to Carle, successful. “The information he provided was useful. … [Wazir] knew things we did not, and needed to know” (Carle 2011, p. 85). Indeed, Carle estimated that he provided “85–90 percent of what he knew” (Carle 2011, p. 109). “He answered truthfully—most of the time. He continued to pretend ignorance on some topics where my colleagues and I believed he had information” (Carle 2011, p. 109). “He withheld information about certain aspects of al-Qa’ida operations” (Carle 2011, p. 115). Still, “in the end, … I decided he was, fundamentally, straight with me. Never totally, but fundamentally, yes.” Wazir “was truthful, innocent, disingenuous, and complicit simultaneously” (Ackerman 2011).

Carle attributes extracting the information to his rapport-based technique, “rather than seeking to cow or humiliate him” (Horton 2011a). While it is clear that Carle did not take the humiliation/psychological “regression” tack that would come to characterize the EIT program, it is impossible to know whether his success was due to rapport alone and/or the possibility things could get worse. Given Wazir’s initial reluctance to talk, it seems reasonable to think that the possibility of torture had some influence.

Carle also came to believe that Wazir was not the high-level Al Qaeda operative the CIA had believed, let alone bin Laden’s banker (Canadian Security Intelligence Service 2009, pp. 23–24)). The CIA had “erroneously inflated [Wazir]’s importance and role.” He “was more like a train conductor who sells a criminal a ticket” not “part of the al-Qa’ida network” and so “not quite what we believed” (Carle 2011, pp. 133, 135).

He simply did not “have the critical information, or close ties with Al Qaeda, that the Agency believed he had and that had justified his rendition. … he was less directly involved with Al Qaeda than we had assessed him to be; and he was no terrorist. … [H]e was fundamentally a businessman; I do not believe he was actively colluding with Al Qaeda. I came to believe him when he said that he vehemently opposed Al Qaeda’s theology and acts. … It was the wrong decision to continue to hold him for eight years after we had established that many of our assumptions were simply wrong” (Horton 2011a).

The men and women nearly 4,000 miles away on the plushly carpeted, wood-paneled seventh floor of CIA headquarters could not have disagreed more. They were convinced that Wazir was hiding valuable information: They wanted the remaining “10 to 15 percent” (Carle 2011, p. 109). The exact same information which Carle interpreted as demonstrating Wazir’s lack of knowledge “convinced Headquarters that he was dissembling” (Carle 2011, p. 145). They pressured Carle at every step of the interrogation; “the order was to do whatever it took to get him to talk … and to do so now” (Carle 2011, p. 64). Finally, they became fed up and decided to render him again, this time to the notorious “Salt Pit,” the CIA black site prison in an old brick factory on the outskirts of Kabul.

This time he got the full treatment by the ninja-clad CIA snatch team: forcible sodomy, shackles, and hood. There he would end up being tortured by other CIA officers under orders from Langley. Whether or not he gave up further information is unknown. He was released in February 2010, eight years after his capture. No charges against him were ever filed.

NO INFORMATION HIDING: MAGNUS GÄFGEN

For a case in the real world to match the no information hiding outcome, it must meet the following conditions:

1. The detainee was cooperative (i.e., had information and was willing to give it up to avoid torture).

2. The detainee was subjected to objective or leading questioning.

3. The detainee gave up all of his valuable information (as far as we can tell).

4. The cooperative detainee was not tortured afterward.

While there may very well be cases from the CIA or other terrorist interrogations, none to my knowledge has been made public. There is, however, a case from Germany which matches the no information hiding version of the valuable information equilibrium perfectly: that of Magnus Gäfgen, a German law student and devout Catholic.⁴

On Monday, September 29, 2002, at 4:25 in the afternoon, Gäfgen and his girlfriend had just parked their car in a basement parking garage at the Rhein-Main airport in Frankfurt when they were yanked from their seats by police commandos in plainclothes. The commandos threw them to the ground and cuffed them, yelling, “Where is the boy?!”

The boy was Jakob von Metzler, the toothy, smiling 11-year-old son of one of the oldest banking families in Europe. On the morning of September 27, 2002, he had said good-bye to his buddies and got on bus number 35 near his school in Frankfurt am Main. It was the last day of school before the fall break and he was looking forward to going on vacation with his family. Normally he got off at a bus stop just 100 meters from the family villa next to Park Louisa in the Frankfurt City Forest, to walk the rest of the way home. He never made it.

A ransom note did, however. It was discovered on the villa property at 12:40 pm the same day by a domestic servant. The Metzlers complied with the demands for a million euros in used, unmarked, small denominations stuffed in Aldi shopping bags. Precisely as the note demanded, they put the bags next to the signpost at the northbound tram stop “Oberschweinstiege” of line 14 at 12:49 am on Monday, September 29. They had also, however, contacted the police, and the drop-off location was under surveillance when the money was picked up at 1:00 am.

The police followed the young man who picked up the money. Hoping he would lead them to the boy, they continued to shadow him on Monday, as he drove around Frankfurt with his girlfriend, depositing the ransom money into his account at various bank branches, buying a vacation for two to the Canary Islands, and taking a test drive in a C-class Mercedes and then ordering it, using some of the ransom money for the down payment. When, however, he drove that afternoon to the airport at Frankfurt, they arrested him.

The police had already identified him as Magnus Gäfgen. Gäfgen was studying criminal law at Wolfgang Goethe-Universität in Frankfurt. Born and raised in Frankfurt, he had been active in Catholic youth groups since he was 16 and had completed his civil service duty working in a nursing home.⁵ They knew, too, that he knew Jakob through his girlfriend’s circle of friends.

Following a brief stop at a hospital to clean up some abrasions on his face, Gäfgen was taken to police headquarters. Detective M., a veteran investigator of kidnappings, brought Gäfgen to his office and immediately started questioning him about the whereabouts of the boy.⁶ Confronted with the fact that the police had seen him pick up the money and deposit it into his account, Gäfgen first said that someone had offered him 20,000 euros to pick up the bags. He professed, though, to know nothing about the boy or his fate. Detective M. tried to build trust with Gäfgen by sitting next to him and using his first name. Upon being told that the media knew of the kidnapping, Gäfgen said “hopefully the child is still alive” (Landgericht Frankfurt am Main 2003, p. 17). Detective M. showed Gäfgen pictures of the boy and talked about the parents’ fear, but nothing worked. Gäfgen provided no information on Jakob’s location.

Detective M. then wrote down on a slip of paper three questions and put it in front of Gäfgen:

Is the boy alone somewhere?

Is the boy being guarded/under supervision?

Is the boy no longer alive?

He told Gäfgen he was going to turn around and asked Gäfgen to put an “X” next to the question corresponding to the boy’s condition. Gäfgen put an “X” next to “guarded.”

Based on this and the fact that the police had mistakenly thought that half of the ransom money was missing from Gäfgen’s apartment, the police believed that Gäfgen had accomplices. He was asked more questions, but refused to answer them. Between 11:30 and midnight he spoke to an “on call” defense lawyer and afterward said he was prepared to say where Jakob was. He told the police that Jakob was in one of the many little cottages on the Langener Waldsee, a recreational lake about 10 miles south of Frankfurt, describing the cottage and the surroundings in some detail.

The detectives pressed him about his accomplices. Gäfgen then falsely accused a pair of brothers he had known when he was younger and with whom he had fallen out. At 1:00 am, Gäfgen asked to stop and go to bed. The detectives agreed and Gäfgen promised to speak further to Detective M. in the morning.

Meanwhile, overnight, a thousand police and dogs were dispatched to the Langener Waldsee to search for Jakob. SWAT teams raided the houses of the brothers in the early morning darkness, immediately asking where the boy was before taking them to the police station for more intensive questioning. They denied any involvement in the kidnapping.

Deputy police chief Wolfgang Daschner arrived at police headquarters at 6:30 am on Tuesday, October 1. He had slept little the night before, worrying about Jakob and how to rescue him. Having heard an update about where things stood from one of his section leaders at 6:35, Daschner called Detective Ennigkeit and Detective Mü. into an adjoining room and informed them he “intended to order the application of direct coercion” (Landgericht Frankfurt am Main 2005, pp. 9–10)).

Daschner calmly set out his reasoning. He did not expect much from the brothers fingered by Gäfgen because it appeared by then that Gäfgen had lied. The child’s life was in danger, with no food or water for three days. “Gäfgen must therefore be induced to divulge the boy’s whereabouts” (Landgericht Frankfurt am Main 2005, p. 10). If it were available, a truth serum could be administered by syringe. Having explained his reasoning, Daschner ordered that Gäfgen be threatened with torture, given a chance to talk, and then, under medical supervision, be subjected to renewed questioning under the infliction of pain not causing bodily injury. Detectives Mü. and Ennigkeit were told to make the necessary preparations.

Detective Mü. was “perplexed,” believing that the use of force was unlawful even in these “extra-legal circumstances” (Landgericht Frankfurt am Main 2005, p. 10). He also thought that there were alternatives which would work, such as confronting Gäfgen with Jakob’s family members.

Despite these misgivings, Detective Mü. began preparations and at 6:50 called the chief of the SWAT team. Detective Mü. asked whether he had anyone in his unit prepared to “torture” Gäfgen. The SWAT leader was so taken aback by this word and the request that he said nothing and hung up the phone. He immediately discussed the request with his unit leaders, and all agreed such an act and order would be illegal.

Five minutes after calling the SWAT team, Detective Mü. ordered a subordinate to summon the police doctor in the building. At 7:00 am, Detective Mü. called a meeting of the section leaders of the kidnapping command center, including Detective Ennigkeit. Detective Mü. relayed Daschner’s order and began a discussion. Daschner’s order provoked much concern and dismay, with many detectives arguing that there was no legal basis for such an order and that any confession thus obtained would not hold up in court. The SWAT team leader said he could not give such an order to anyone in his unit; if he did, it “would put the good reputation of his unit in doubt” (Landgericht Frankfurt am Main 2005, p. 11). Another said that the use of force wouldn’t work. Moreover, there were better alternatives and they should stick to the current plan to confront Gäfgen with family members. One detective, “R,” did say that his conscience made him support Daschner’s plan—even if it was illegal—as a last resort, if all other avenues had been exhausted. They concluded their discussion with an agreement to stick to their plan, including further questioning of the brothers and the search at the Langener Waldsee. Hurting Gäfgen was to be done, if at all, only when they had taken all other possible steps.

Twenty minutes later, at 8:00 am, Daschner called several of the group into his office, demanding angrily and loudly to know why his order had not been carried out. They relayed their unified misgivings about Daschner’s order and explained their plan and hopes of success. Daschner asked the SWAT team leader why he couldn’t find anyone to carry out the torture. Minimizing the significance, Daschner provided an example of what he was talking about: “twisting over the thumb and wrists” or “pressure” “in particular spots” such as the ear, which would cause “pain, a lot of pain” that would induce Gäfgen to talk “in a very short time” (Landgericht Frankfurt am Main 2005, p. 13; Dahlkamp 2004).

The SWAT leader said that such an order could not be given and that no one in his unit was prepared to carry it out. Upon Daschner’s continued pressing, the SWAT leader said that there was one person who might do it, but he was on vacation. Daschner then demanded that he be flown by helicopter back to Frankfurt, and Daschner would speak to him personally.⁷

Simultaneously, the detectives had confronted Gäfgen with his mother. Gäfgen told her that he had been blackmailed and said nothing about the whereabouts of the child.

At 8:30 am Daschner called Ennigkeit into his office. Daschner instructed Ennigkeit to question Gäfgen again, to appeal to his conscience and the acute mortal danger Jakob was in. If Gäfgen still refused to provide any information, then Gäfgen should be told that he could expect direct force to be used against him. He should be threatened that he will be questioned again while being inflicted with pain under a doctor’s supervision. He would also be informed that the torture would cause no permanent damage. It was clear to both Daschner and Ennigkeit that the threat must be “intensive and intimidating” given both the urgency of the situation and Gäfgen’s “stubborn resistance” (Landgericht Frankfurt am Main 2005, p. 14). Daschner also let Ennigkeit know that the man willing and able to carry out the torture if necessary was being flown by helicopter from his vacation. The entire conversation lasted only a few minutes.

About ten minutes later, the police doctor arrived at the command center, was informed by Detective Mü. of the situation and Daschner’s plan, and asked whether he was prepared to participate medically. The doctor said yes. When Detective Mü. notified Daschner by telephone about the doctor’s readiness, Daschner responded that he had already ordered Ennigkeit to threaten Gäfgen.

Indeed, by that point, Ennigkeit was already sitting alone with Gäfgen in the interrogation room. He sat down across from the law student, face to face, a foot away, and told him that the police were going on the assumption that Jakob’s life was in danger and that it would be to Gäfgen’s “advantage” to reveal the location of the child. If he continued to remain silent or provide false information, the police were prepared to inflict pain on him, with a doctor present and with no lasting bodily harm, in order to get him to reveal details that could save the child’s life (Landgericht Frankfurt am Main 2005, p. 16). They were also considering administering a truth serum.

Ennigkeit spoke “intensively and forcefully” (Landgericht Frankfurt am Main 2005, p. 16). The detective added that a special officer was on his way by helicopter “who could inflict pain on him that he would never forget” (Landgericht Frankfurt am Main 2005, p. 16). As he said this last part, he made a circling motion with his hand, imitating the rotors spinning. Ennigkeit concluded by saying that he hoped that this and Gäfgen’s conscience would cause him to change his mind and “provide truthful information” (Landgericht Frankfurt am Main 2005, p. 16).⁸

The threat made an impression on [Gäfgen]. He had not been moved by the appeal to his conscience and the possibility for leniency …. The threat scared him … Ennigkeit’s … forceful and intensive threat … caused him to fear even harsher measures. He did not want to expose himself to those (Landgericht Frankfurt am Main 2005, p. 17).

It took fewer than ten minutes for Gäfgen to say where they could find the boy.

Gäfgen himself was driven to the location, a small pond near the town of Birstein, to point out exactly where Jakob was. The police found Jakob precisely where Gäfgen said he would be: in shallow water under a dock, wrapped in a plastic bag and duvet cover.

No one ever ended up laying a hand on Magnus Gäfgen; the threat had worked. He had revealed all his information under objective questioning and the threat of torture, but it was too late for Jakob von Metzler. Gäfgen later confessed (without torture) that after “intensively thinking things through” in the planning stages of the kidnapping, he had decided that Jakob “must die” if his plan was to succeed (Landgericht Frankfurt am Main 2003, pp. 10, 11). The boy already lay murdered in Gäfgen’s trunk while the ransom note was being thrown onto the Metzler property.

BACK TO THE MODEL

In both the Wazir and Gäfgen cases, was low. In the Wazir case, this was true for both Carle and his CIA bosses back in Langley; neither thought that Wazir was an innocent somehow accidentally caught up in the program. Canadian intelligence and the United States’ own investigation linked him to Al Qaeda financing. In the Gäfgen case, the fact that Gäfgen picked up the money made it clear to the police right away that he was somehow involved; and after they quickly ruled out other suspects, it became just as clear that he had acted alone and knew Jakob’s location.

The other thresholds differed across the two cases and even within the Wazir case. Wazir’s threshold was apparently high, at least during the time spent with Carle at the CIA’s Moroccan black site. While he revealed a lot of information—as much as 90% of what Carle believed he possessed—he did not reveal it all. For Carle this was enough; it satisfied his threshold .

It was not enough for CIA officials back in Langley; their version of was higher than Carle’s. They expected more out of Wazir. One reason may have been related to the clarity parameter u. Wazir’s exact role in Al Qaeda financing was just never that clear. A Canadian federal court looked into the Wazir case during a related proceeding and found not only that “the public evidence concerning Hadje Wazir being linked to bin Laden is also inconclusive” but also that it couldn’t be sure that a Wazir associated with Al Qaeda financing was even the same person as Pacha Wazir (Federal Court of Canada 2010, pp. 345, § 259). In the court’s words:

As to whether Hadje and Pacha Wazir are one and the same or not, the Court notes that these names are common in Pakistan. The evidence is therefore also inconclusive in that regard (Federal Court of Canada 2010, p. 345, § 260).

This lack of clarity may have also figured into CIA estimates of what they thought Wazir might be hiding. They not only wanted the last 10–15%, but also were convinced that the last hidden information was very valuable, an assessment not shared by Carle. In other words, Langley thought that the v Wazir was not revealing must be very high. The only way to bring down Wazir’s to the point of talking was to torture more. And so they shipped him off to the Salt Pit. Whether or not he provided more valuable and accurate intelligence, we just do not know.

In Gäfgen’s case, was also very high, if only because the police wanted only one very specific piece of information from their suspect: Jakob’s location. Gäfgen’s threshold , however, was apparently very low. All it took from Detective Ennigkeit was talking about a “specialist” in administering pain and a whirring of his fingers to cause Gäfgen to tell the truth.

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)

3. The Bush model generates strange, quixotic outcomes. (Chapter 4)

4. The Bush interrogational torture program is more realistically modeled as objective and leading question variants of an incomplete information game, with three types of detainees, two types of interrogators, and uncertainty about the amount and value of information provided. (Chapter 5)

5. By positing a set of Detainee strategies, calculating the Interrogator’s expected utility using Bayes’ Rule to identify her best response, and checking for incentives to deviate by any of the Detainee types, it is possible to derive a perfect Bayesian equilibrium in which a Detainee is tortured after providing information. (Chapter 6)

6. The RIT model generates nine perfect Bayesian equilibria, the formal and empirical characteristics of which generate important observations, propositions, and implications, including:

(a) The Interrogator’s thresholds for believing that a Detainee is Innocent after “no information” are less than one-half, with close to one-half and closer to zero.

(b) The Interrogator’s information hiding threshold under objective questioning is greater than or equal to one-half, whereas her information hiding threshold under leading questioning as well as the Detainee’s version of are a little less than one-half.

(c) Objective questioning (potentially) provides better information, but is necessarily accompanied by more torture, than leading questioning, which, however, provides less valuable information.

(d) All things being equal, interrogators are more likely to get less valuable information than highly valuable information. (Chapter 7)

7. Surprise torture of a Cooperative Detainee—even if he has provided all his information—is not only likely, but perversely more likely

(a) the more willing the Detainee is to divulge information and

(b) the more important the information is in terms of “unknown unknowns.” (Chapter 8)

8. The perversity under objective questioning persists under leading questioning, namely

(a) the more the Innocent Detainee believes Interrogator promises of no torture in exchange for confirmation, the more likely is ambiguous and false information.

(b) the more brutal the torture, the more likely is ambiguous and false information.

(c) the more important the confirmation is to the interrogator and the more difficult it is for the detainee to understand what is being asked, the greater the likelihood of torture.

(d) leading questioning will not eliminate torture; the higher the interrogator’s demand for confirmation, the more torture. (Chapter 9)

9. The valuable information, selective torture equilibrium reveals four further trade-offs and a paradox:

(a) Everything else being equal, eliciting information

i. requires either more frequent or more brutal torture,

ii. is more likely when the information is less valuable or the torture is more severe, but not both,

iii. is more likely when the standard of cooperation is lowered, even though it increases the likelihood of information hiding,

iv. is necessarily accompanied by innocent torture; decreasing the likelihood of innocent torture requires decreasing the likelihood of information.

(b) the very assumptions justifying interrogational torture paradoxically make it less reliable in terms of the likelihood of getting valuable information. (Chapter 10)

The next step in the argument is to examine and interpret the no information, torture and no information, no torture equilibria.