Bargaining with criminals: The morality of witness collaboration in Mexico's “war on drugs”

Abstract

Public authorities take considerable and oftentimes controversial steps in their efforts to dismantle criminal organizations involved in drug trafficking and related crimes in Mexico. Among other things, they recruit offenders who abandon their criminal organization and strike a deal with law enforcement agents and prosecutors to share information about their co-perpetrators in exchange for leniency in sentencing as well as of protection from retaliation. This article explores whether the deployment of collaborators is morally permissible in view of the significant risks it exposes them to, most notably retaliatory aggressions. The article examines the underlying philosophical problem regarding the justifiability of deploying collaborators in the social and political circumstances prevailing in the country. The normative framework I advance to explore the Mexican case can be useful in examining the ethical implications of using collaborating witnesses elsewhere.

Public authorities take considerable and often controversial measures in their efforts to confront drug trafficking and related crimes in Mexico. Among other measures, they recruit, deploy, or “turn” significant numbers of individuals to obtain information about criminal organizations and their operations. Collaborating witnesses, or testigos colaboradores, fall within this category. They are offenders who abandon their criminal organization and strike a deal with Mexican law enforcement agents and prosecutors to share information about their co-perpetrators in exchange for leniency in sentencing. Similarly, the Drug Enforcement Agency (DEA) heavily relies on collaborating witnesses in its operations in Mexico and elsewhere. In both cases, which are the focus of the article, public authorities—in Mexico and the USA—have determined that enlisting collaborators in the service of the criminal justice system is worth the risk of retaliatory aggressions and other consequences. This article calls into question the morality of that determination.

Even if it turned out that the abuse of the system for collaborators can be eradicated and the practice reformed, there is still room to question the morality of the practice. Above all, this article argues that the use of collaborators ought to be regulated and constrained according to clear and defensible principles, in particular, principles of proportionality and necessity, as they are commonly understood in the literature on just intelligence. As I shall argue, intelligence involving collaborating witnesses in Mexico commonly fails to honor these principles, however, in part because of the institutional and socioeconomic conditions in the country, which include extraordinary violence and pervasive corruption, exacerbate the potential risk of harm to collaborators themselves and to ordinary citizens.

To anticipate the gist of the argument in the following sections, I shall contend more specifically that in light of the harm inflicted on collaborators, and in some cases on the citizenry, the use of collaborators in Mexico is a disproportionate and sometimes even an unnecessary intelligence measure. Although intelligence work inherently entails uncertainty, particularly of the epistemic kind, for collaborating witnesses and civilians, the degree of uncertainty varies depending on the context. In a constitutional democracy, with low levels of corruption, strong security and justice institutions, and moderate level socioeconomic deprivation, the uncertainty diminishes considerably. Conversely, in countries such as Mexico, where authoritarian enclaves persist within formally democratic institutions, and where corrupt and unaccountable public officials, weak security and public institutions, rampant violence, and precarious socioeconomic conditions prevail in many regions of the country, the margins of uncertainty for intelligence work widen considerably, putting both collaborators and civilian population at grave risk. In such settings, intelligence interventions that rely on the work of collaborating witnesses and informers will tend to be disproportionate, even if necessary, to the detriment of intelligence agents themselves as well as civilians. In situations of epistemic uncertainty arising from institutional deficits or social violence, a restrictive view of proportionality is the morally appropriate path for intelligence operations.

In considering this topic, several questions arise. Why focus on collaborators and the perils they go through as they become state witnesses, for example? Are they not, after all, secondary figures in the landscape of drug violence in Mexico? What does the fate of a handful of collaborators (most of them morally and legally culpable to some extent) matter compared to the plight of the hundreds of thousands of the innocent victims of drug violence, some of it perpetrated by these very same collaborators?

For theoretical purposes, exploring the Mexican case illuminates the moral implications of using collaborating witnesses elsewhere. Scholarly work on the ethics of knowledge acquisition in law enforcement contexts has advanced some guidelines to orient the work of intelligence-led policing (Kleinig, 2009) but has paid little attention to the more specific case of rewards for cooperation in circumstances of extraordinary violence (see, for instance, Dunnighan and Norris, 1998; Harfield, 2012). Similarly, some academic studies on plea-bargains focus on their corrosive impact on the criminal justice system; others allude to the problem of “witness intimidation” (Lippke, 2011: 154; Natapoff, 2009: 42–43, 131–134). Yet, most fall short of theorizing more deeply the relationship between, on the one hand, witness collaboration and, on the other, debilitated institutions and social and political violence, particularly massive violence. While the just intelligence theory will be used as a lens to offer an ethical examination of the case of collaborators in Mexico, the case itself can shed light on some of the limitations of the theory in complex conflict settings, typical of the Global South. In this sense, the article relies on the just intelligence theory while at the same time presenting a critique of it.

The article unfolds as follows. Section one sketches the philosophical framework that sets moral parameters for subsequent sections—the just intelligence paradigm. This framework guides the inquiry in the next sections. Section two focuses on the case of testigos colaboradores of Mexican authorities. Section three then turns to the Drug Enforcement Agency's use of plea-bargains in the American criminal justice system to bring offenders to cooperate with their intelligence operations in Mexico. The final section provides a brief conclusion.

Witness collaborators and just intelligence

Many criminal justice systems rely on witness collaborators including the USA (Lippke, 2011; Natapoff, 2009). For their critics plea-bargains distort punishment, since similar crimes receive different penalties, depending on how “useful” the accused turns out to be to authorities. Penal rewards of this kind also undermine due process, since they set incentives for defendants to self-incriminate. If these agreements are informal and secretive, as they usually are, they erode transparency of the criminal justice system because they hide the nature of agreements between law enforcement agencies and witnesses from the public eye. In this article, I focus on a different problem: the potential risk of retaliation against collaborators, which state authorities promoting rewards for cooperation, as well as the contexts in which they take place, may exacerbate, or contain. Since the use of collaborators by law enforcement agencies must be understood as a form of intelligence collection, the ethical framework of just intelligence theory can be productively harnessed to address this problem.

A just intelligence framework sets out principles that “limit the harm intelligence collection causes while outlining what circumstances would be required to justify the harm caused” (Bellaby, 2012: 108). Bellaby (2012) and other scholars argue that the just war tradition in philosophy can be a useful starting point to examine the proper use of intelligence (Quinlan, 2007). Just war theory seeks to elucidate two dimensions of the ethics of war—the proper ends justifying a war (jus ad bellum), and the morally appropriate means to conduct it (jus in bellum) (Macnish, 2015). By analogy, jus ad intelligentiam is concerned with the proper motives to employ intelligence services, and jus in intelligentia posits the morally adequate ways to do so. ¹

The analogy between just war and just intelligence is of course not perfect (Omand and Phythian, 2013), and some scholars are skeptical about the fruitfulness of deploying the former theory in the context of intelligence, particularly with respect to preventive work as opposed to more targeted forms of intelligence against suspects (Diderichsen and Rønn, 2017). These reservations notwithstanding, for the purposes of this article, the general gist of the theory can be employed to demarcate the parameters of morally permissible intelligence work.

Jus ad intelligentiam, which typically applies to elected representatives or bureaucrats, falls beyond the scope of the article. ² Jus in intelligentia principles include that of discrimination (liable or non-liable targets must be clearly discerned one from the other); of proportionality (the harm that is perceived to be caused by specific intelligence interventions should be outweighed by the perceived gains); and of necessity (less harmful interventions should be attempted before more harmful ones are chosen).

The principle of necessity posits that in choosing between two harmful means of collecting intelligence, the least harmful must always be selected—harm minimization is critical for this principle. Proportionality has three components: the benefits deriving from the intelligence activity in question (e.g. gathering evidence to convict a corrupt public official complicit with criminal organization, or thwarting an operation by one of these organizations), its costs (e.g. the harm they cause to those who conduct it, or to civilians), and the appropriate balancing between benefits and costs (Bellaby, 2012: 114; Hurka, 2005: 38; Macnish, 2015: 532; Uniacke, 2011). While the relevant benefits must be clearly connected to the just cause, there is no parallel restriction with respect to the cost of intelligence. Every harm counts, although some count more than others. The “risks to agents and those to whom a duty of care is owed”; “risks of collateral damage to others”; and “risk to future operations and to institutional reputations if the operation were to go wrong” should be front and center (Omand and Phythian, 2013: 54). By contrast, the harm experienced by members of criminal organizations—and this includes the witness collaborator—matters but has diminished weight, since through their actions they made themselves liable to be affected by intelligence seeking to confront the crimes they are or were involved in. Factors such as their complicity with, and their moral responsibility for, the crimes in question render it permissible to put them at risk of harm to their lives and integrity; the greater their complicity and their responsibility, the more justified it will be to impose such risks on them. Put succinctly, then, the kinds of harm that count towards the proportionality assessment are “morally weighted”, to use Seth Lazar's (2012: 6–7) term: some are more morally significant than others.

Tallying up moral costs and benefits in the realm of intelligence is of course extremely difficult. One challenge, on the epistemic level, involves accurately identifying the significance of potential benefits of an intelligence intervention as well as its unintended consequences, with their unintended costs. Embarking on an intelligence operation commonly has as its goal collecting information to determine the nature and level of a security threat, or whether there is one at all. If it is found that no threat was forthcoming, but the cost of the operation turns out to be high, the intelligence operation can be judged to be disproportionate; but this judgment can only be made retrospectively (Miller, 2021). Similarly, assessing alternative intelligence operations presupposes foresight of the consequence of each alternative. Yet such foresight is oftentimes unreliable, as unexpected events might derail intelligence operations or even throw them into complete disarray. ³ For all these reasons, intelligence operations involve a high degree of epistemic uncertainty. While the just intelligence theory acknowledges the epistemic uncertainty around intelligence interventions, it may not duly consider the extent to which the uncertainty is heightened in nations with institutional deficits in law enforcement (since one of the essential functions of institutions is precisely to help reduce unpredictability); with large socioeconomic disparities; and engulfed in massive social violence. These factors undermine the ability of law enforcement agencies to anticipate and minimize disruptions to their operations.

At any rate, epistemic uncertainty makes balancing costs and benefits all the more challenging, particularly with intelligence cases in the “twilight zone”, by which Kevin Macnish (2015: 539) means those cases that are neither clearly disproportionate nor clearly proportionate. Most cases are of this kind. In balancing harms and benefits, according to this view, the theorist making a proportionality assessment can either take a restrictive view, which presumes disproportionality unless there are strong reasons to the contrary; and a permissive view, which presumes proportionality unless there are strong reasons to the contrary.

“Don’t bother calling the witness, your honor—he’s dead”: ⁴ Using and abusing testigos colaboradores

In the last 20 years, Mexican federal authorities have increasingly recruited testigos colaboradores in a bid to reduce the dramatic surge of drug-related violence, which just in 2020 greatly contributed to Mexico's homicide rate of 29 homicides per 100,00 inhabitants (INEGI, 2021) as well as to the country's crisis of enforced disappearance (49,581 people were forcibly disappeared between the end of 2018 and mid-2021) (Secretaría de Gobernación, 2021). The use of collaborators became formally acknowledged in 1996, before rising steadily but significantly during the tenure of Vicente Fox (2000–2006) and Felipe Calderon (2006–2012), and then declining during Enrique Peña Nieto's (2012–2018) presidency, when the interest and resources allocated to them diminished considerably (Méndez, 2013; more generally, see López Benitez, 2011; see also the Appendix for an official calculation of their number; this calculation is probably an underestimation). It is unclear whether testigos colaboradores will gain saliency again during the administration of Andrés Manuel López Obrador (2018–2024), but they are present in a few high-profile cases in the current criminal justice agenda, including the enforced disappearance of 43 students of the rural teachers college of Ayotzinapa, discussed below. There is also some indication that informal programs of testigos colaboradores are in operation at the local level (Ahmed and Villegas, 2019). During the last two decades, the figure of the witness collaborator has received increasing (albeit insufficient) legal and constitutional grounding in Mexico, first in the Ley Federal de Contra la Delincuencia Organizada (Federal Law against Organized Crime) of 1996 ⁵ and then in 2012, with the Ley Federal para la Protección a Personas que Intervienen en el Procedimiento Penal (Federal Law for the Protection of Individuals who Intervene in Criminal Proceedings), among others.

Despite these regulations, as we shall see, collaborators face life-threatening conditions as a consequence of their cooperation with authorities. Is it morally acceptable to subject them to such risk, or is it too high a moral price to pay in the quest for useful intelligence? Following the just intelligence paradigm sketched in the previous section, the principles of necessity and proportionality help us answer this question. The section will argue that, taking into account the institutional and socioeconomic background conditions (i.e. weak political and criminal justice institutions, corruption, violence, social disparities, and so on) in which the practice is embedded, as well as other considerations discussed below, collecting intelligence through collaborating witnesses may be a necessary measure, but it remains disproportionate.

A matter of necessity?

Intelligence collection through the use of witness collaborators occurs in many criminal justice systems. Often, law enforcement cannot fathom the details of the activities of criminal organizations without a knowledgeable “source” within them, since these organizations operate clandestinely and commonly have enough financial or logistic resources to elude public authorities. The Mexican criminal justice system in particular might be more dependent on collaborating witnesses than other systems because of its insufficiently developed capabilities to conduct rigorous investigations. In his analysis of police forces in Mexico, Gustavo Fondevila rightly argues that owing to “its inadequate professionalization (the use of scientific investigation methods), the lack of a professional police infiltration system (undercover agents) and citizens’ lack of trust in the police, almost all the relevant information received comes from informers”. As a result, he adds, the police have been forced to “hyper-develop its informer system as its sole source of information” (Fondevila, 2013: 117–118). Thus, notwithstanding the Mexican state's responsibility to develop investigative capabilities, the dependence on informers or collaborating witnesses seems to be a necessary counter-narcotic strategy.

Compounding matters further, many criminal organizations, running profitable businesses that are not limited to drug trafficking, have financial resources that match or exceed those of local and sometimes even federal authorities; the militarization of drug-trafficking combat, meant to assist and sometimes replace these authorities, does not redress this imbalance. Finally, for the purposes of crime prosecution, collaborators are the only avenue for authorities to go beyond the lower-level “operators” of criminal organizations and reach their higher echelons (in administrative and/or authoritative positions). ⁶ In light of these considerations, it seems legitimate for authorities to argue that they rely on collaborators as a matter of necessity.

A matter of proportion?

Are the interventions proportionate, however? That depends, as we have seen, on the assessment of costs and benefits. Let us consider the benefits first. Testigos colaboradores have a grim record in terms of the benefits they bring to the criminal justice system. Accusations based on their testimony have produced little in the way of successful prosecutions. According to a federal judge interviewed for this article, the use of the testigos colaboradores became so vitiated that often the same individual would offer testimony in several, unconnected prosecutions. The result was a contaminated prosecutorial process where convictions rested on these dubious testimonies. Unsurprisingly, many investigations relying exclusively on the testimony of these witnesses have ended up crumbling. ⁷ The perverted use of testigos colaboradores dovetails with two ingrained practices in the Mexican criminal justice system—the “fabrication” of witnesses for the legal inculpation of innocent citizens and the persecution of political enemies by federal authorities.

But these results arise, as stated, from a perversion of the system. There are some instances in which the use of testigos colaboradores does yield benefits. These are difficult to appreciate because information about the role of collaborators in criminal prosecutions is not always publicly accessible; their contributions are sometimes informal and thus fail to make it into legal records. While the Federal Prosecutor's Office has extensively relied on witness collaborators (see Appendix), a former district judge interviewed for this article affirmed that s/he is unaware of a single sentence in which penal rewards have in effect been granted in return for cooperation. ⁸ Nonetheless, the case of Gildardo N furnishes an example of these potential benefits. This witness collaborator participated in a criminal case that caught international attention: the enforced disappearance of 43 students from a local teachers’ college (Ayotzinapa), who were abducted by a criminal organization with the complicity of local and federal security forces in the state of Guerrero. Gildardo N was apprehended in connection with these crimes but later released when a judged considered that the evidence against him, while incriminating, had been illegally obtained. He was later recruited as a testigo colaborador. The information he shared seems to have steered the investigation away from the impasse to which it had been directed by the previous administration (deliberately some argue), and in a more promising direction (Centro de Producción CEPROPIE, 2021). Let us concede then, for the sake of the argument, that at least in some cases the use of testigos colaboradores can yield benefits such as locating the whereabouts of victims of enforced disappearance or, simply, abating violence. If Gildardo N were put in jeopardy for his cooperation, but as a result Mexican society was closer to finding the location of the Ayotzinapa students, wouldn’t that be worth the risk?

To answer questions of this sort, we need to turn to the costs of using witness collaborators, which include the harm to collaborators themselves. As was mentioned above, due to their complicity and responsibility with wrongdoing, the harm they might be subjected to because of their collaboration, or the risk thereof, is a source of moral concern but has diminished moral weight, compared to what the actors might suffer. But just how much should the harm suffered by collaborators be discounted, morally speaking? The answer to this question depends on two considerations. An institutional consideration is whether the criminal justice system possesses the tools to protect them. The second consideration refers to the varying criminal and socioeconomic profiles of collaborators, which entail different degrees of culpability and responsibility, as well as different bargaining powers. I now turn to discuss both matters in greater detail. As an illustration of them I provide two contrasting examples—the case of “Venus”, a low-level perpetrator, and the case of Edgar Enrique Bayardo del Villar, aka Tigre. These examples shed light on the richly textured considerations that must be made in assessing the costs of intelligence work.

Despite attempts at institutionalizing the practice in Mexico, for collaborators the risk involved in cooperating with authorities has always been and remains extraordinarily high. The laws that were put in place to regulate the practice were either vague or never fully enforced. As a consequence, security guarantees for testigos colaboradores, from recruitment stage to more advanced stages in their collaboration, were poorly implemented, if at all.

Venus was a testiga colaboradora who began cooperating in 2010 (García and García, 2019), two years prior to the enactment of the Federal Law for the Protection of Individuals who Intervene in Criminal Proceedings. She is believed to have been a collaborator for the Zetas, a criminal organization described in further detail in the next section. Venus and others testified against the main leaders and some other members of the organization (Espino, 2019a, 2019b). According to her, she was disenrolled as a testigo colaboradora in 2015 (García and García, 2019) but was summoned to testify the same year, without protection. Despite the fact that a federal judge ordered the Federal Prosecutor’s Office to extend safety guarantees to her, the Office declined to do so (Espino, 2019a, 2019b). She refused to testify on that occasion, claiming she had been intimidated against doing so several times (Rainsford, 2019). Venus reports that as a testiga colaboradora she was offered housing in overcrowded and unsanitary conditions, in a house guarded by the military and federal agents. She lived there with another 13 testigos colaboradores and the families of some of them.

This case is representative of a pattern of cases where collaborators find themselves in the grips of an inefficacious criminal justice system that is unable to properly guarantee them a minimal level of protection in exchange for their cooperation but recruits them into cooperating nonetheless. ⁹ But the harm the likes of Venus experience is not only the result of the institutional precarity of witness protection. It is also caused by the socioeconomic precarity of collaborators.

Collaborators are not a uniform class of people. They differ in their socioeconomic status, their degree of involvement in crime, and the extent to which they can leverage an advantage over the criminal justice system. In the case of Venus, there is every indication that she stemmed from a disadvantaged socioeconomic background. Criminal offenders of this socioeconomic profile tend to be recruited into criminal organizations under conditions that cannot be described as fully voluntary. With some exceptions, they occupy indefinitely the lower rungs of the criminal rackets they serve and have a modest if not insignificant role in its decision-making processes. In a sense, they are victims of drug violence as well. Similarly, it is not even clear that Venus participated voluntarily as a collaborating witness, given that she was enrolled into the program of witness collaborators, then disenrolled, then summoned to testify again, all the while lacking a clear sense of the outcome of her criminal case. Her participation seems to have been based on manipulation or even coercion, and obtained without guaranteeing her access to legal assistance. Because her gains in terms of potential sentence reductions are anything but secured, the bargain is unfair to her. In subjecting Venus to the risks entailed in serving as a state witness, the state unjustifiably coerces her.

Contrast the criminal and socioeconomic profile of Venus to that of Bayardo—the second example I shall discuss. Before turning into a testigo colaborador, Bayardo held several high-ranking positions in the federal and state police forces while at the same time cultivating ties to drug-trafficking organizations such as the Sinaloa cartel and even informing for some of them for 20 years. ¹⁰ Bayardo provided a great deal of intelligence (not all of it reliable) to authorities, until he was executed in a Starbucks a year after he began his collaboration. Upon close inspection, it is clear that his execution was possibly owing in part to some of the same failures of protection of collaborators described in the case of Venus. Not only was the presence of Bayardo at such a public venue a blatant security failure, an investigation of his assassination revealed that the security agents in charge of his protection were poorly qualified as bodyguards (they were low-ranking agents), working 48 hour shifts; in fact, they were subsequently imprisoned for their negligence (Proceso, 2011). Bayardo's case is not an isolated one. Several high-ranking leaders of criminal organizations have received lenient sentences in return for some sort for cooperation.

Notwithstanding the similarities in relation to the deficiency of protection guarantees, which alone ought to dissuade law enforcement agencies from using collaborators, the cases of Venus and Bayardo are different in other regards. Bayardo's criminal profile warranted subjecting him to a greater level of risk than would be appropriate for collaborators like Venus and similar offenders. Contrary to low-level perpetrators, who must negotiate in lopsided and asymmetric conditions, collaborators like Bayardo have a significant bargaining advantage. Being affluent, they gain access to expensive lawyers and sophisticated legal advice, which allows them to exploit the weaknesses of the criminal justice system.

In any case, many of them, again like Bayardo, have a deep involvement with the criminal racket they worked for and subsequently informed against. They are, therefore, also morally blameworthy. Such matters suggest that the state is then more justified in putting these collaborators under retaliatory risk given their responsibility in the crimes under investigation. Lippke (2011: 154) is right in arguing that offenders “can reasonably be expected to make more sacrifices in helping to bring their fellow wrongdoers to justice”. The kind of protection we can extend to them, he grants, is imperfect. “However, those directly involved in the wrongdoing might be thought to have a stronger duty to take the relevant risks. Hence, having to do or risk more is precisely what we should demand and expect of criminal wrongdoers” (2011: 154). Since they helped “create the mess”, as Lippke (2011: 154) succinctly puts it, they now bear the brunt of the responsibility to fix it. A corollary of this idea seems to be that the greater their involvement and the responsibility in the commission of crimes, as was Bayardo's case, the greater the risks of reprisal they should be expected to assume.

In sum, while for Mexican law enforcement authorities conducting intelligence efforts through the use of collaborating witnesses may be necessary, those efforts are not always proportionate. As we saw, assuming that some significant benefit in intelligence collection arises out of the witnesses’ contribution, the assessment of costs varies depending on how morally justified authorities are in exposing collaborators to harm. This is turn depends on whether witnesses collaborate voluntarily, and with some advantage in the bargain; on whether they are complicit and responsible with respect to the crimes in question; and especially on whether the state has the ability to protect the collaborator against retaliation from former accomplices. When it is not clear that the costs of witness collaboration are outweighed by its benefits (the paradigmatic case of Venus being one such instance of this), the state must adopt a restrictive view of proportionality.

A “frog hunt”: The morality of the DEA's bargaining

Mexican law enforcement agencies are not the only ones leading intelligence efforts that involve the use of witness collaborators. Since the 1970s, when the Ford and Carter administrations ramped up their counter-narcotic efforts, the DEA has “engaged in aggressive, unregulated policing and intelligence gathering in Mexico” (Teague, 2019: 803). There are many problems with the way in which the DEA routinely conducts its operations within Mexican territory, most notably, the violation of national sovereignty. I will not go over these problems in any detail here however. Instead, I shall focus on how the DEA's use of collaborating witnesses may take a toll on ordinary citizens. While there might be gains from this intelligence, the harms to civilians resulting from them may be significant; some of these harms may have not been intended but they were certainly foreseeable, given the background conditions (political, social, institutional) that prevail in Mexico, particularly in areas under control of criminal organizations. All things considered, in this context, the use of collaborating witnesses turns out to be both unnecessary and disproportionate.

Examining the use of collaborating witnesses by the DEA is warranted for several reasons. The agency is one of the most important counter-narcotic actors operating in the country. In addition, the intense scrutiny of its actions on the part of the media and non-governmental organizations offers a glimpse of its modus operandi. In the realm of intelligence services, where public information is scarce, this is a significant advantage. Finally, the DEA furnishes a useful contrast to the cases discussed before. While Mexican law enforcement agencies are, as we saw, highly unprofessional, budget-strapped, and relatively modest in the aims they pursue by recruiting collaborators, the DEA is a professionalized, well-resourced, and ambitious agency. For these reasons, its actions have wider social repercussions than those previously examined.

I shall begin by discussing in some detail the most egregious atrocity in Mexico in connection to drug violence—the 2011 massacre in Allende, in the northern state of Coahuila—in which the DEA played a decisive, if unintended, role. ¹¹ The massacre was perpetrated by the Zetas, one of the most violent criminal organizations at the time and which had once been the armed branch of Gulf Cartel. An intelligence operation in the pursuit of two members of the Zeta leadership—the Treviño brothers (Z40 and Z42)—triggered the massacre. In turn, the operation was part of the so-called kingpin strategy, which the agency enthusiastically endorses, and which touts the fragmentation of criminal organizations and the capture and elimination of their leaders as the ideal mechanism to weaken these organizations (Pérez Ricart, 2020).

A DEA agent (Robert Martínez) had accomplished an intelligence feat that carried the promise of contributing to the decapitation of the Zetas: he had obtained the trackable cellphone identification numbers of two of its kingpins, the Treviño brothers. The leaker was José Vázquez, a cocaine distributor in Texas who was part of the organization. To extract the information, the agent threatened Vázquez with imprisoning his mother and wife (presumably with legal grounds to do so), forcing Vázquez to exert pressure himself on other criminal confederates to obtain the coveted intelligence. Charging a family member in order to pressure a defendant is commonly referred to as a “wired plea” because the outcome of the family member's case is “wired” to the defendant's cooperation (see Natapoff, 2009: 49).

Against the explicit wishes of both the agent and the leaker, DEA supervisors in Mexico City or Texas shared the intelligence with a unit of the Mexican police, potentially the Sensitive Information Unit (SIU). However, immediately after the unit received the information, the Zetas learned about the leak and identified the leakers, presumably because they had infiltrated an informer in the SIU. Ironically, the members of the unit had been trained by the DEA itself and were supposed to be impervious to leaks and corruption. The revelation precipitated the massacre and the Zetas, with the complicity of local authorities, set out to retaliate against “frogs”, as Zetas pejoratively call those who cooperate with the public authorities. When this failed (the suspects managed to flee to the USA), they went after their relatives, friends, and associates in Allende and in Piedras Negras. Atrociously and absurdly, anyone with the last name Garza (a common surname in northern Mexico) was targeted (Thompson, 2017b). The bodies of the victims of the massacre, in the hundreds, were burned clandestinely in a local ranch-cum-crematorium. From the perspective of the capture of the Treviño brothers, the operation was part of a broader effort that bore fruits, as both targets were eventually apprehended in 2013 and 2015, two and four years after the massacre, respectively. Was the intelligence operation in question necessary and proportionate?

A matter of necessity?

While targeting the leadership of Zetas and placing them under surveillance was a legitimate intervention, it does not follow that it required witness collaborators. Unlike Mexican law enforcement agencies, with their chronically meager budgets and unprofessional cadres, relative to the tasks they face, the DEA receives adequate funding (with the ensuing professionalization of its agents and cutting-edge technology) and is part of a productive and powerful intelligence community, which includes other agencies such as the Federal Bureau of Investigation (FBI) and the US Immigration and Customs Enforcement (ICE). These characteristics allow the DEA to pursue alternative, less harmful intelligence methods. As a matter of fact, the eventual capture of Miguel Ángel Treviño, carried out without a single casualty around the US–Mexico border, involved the use of commercial computer spyware, probably placed on the kingpin's computer through phishing or some other form of cyber manipulation; GPS mobile tracking devices; as well as the use of a drone facilitated by ICE (Garcia, 2013). The capture, then, was the result of the coordination between the DEA, with the assistance of other US intelligence agencies, and the Mexican military. Put succinctly, less harmful avenues of intelligence collection were available to the DEA for capturing the Zeta kingpins, and therefore its actions, leading to the Allende massacre, flouted the principle of necessity.

A matter of proportion?

Formally speaking, the DEA abides by a stringent protocol guided, implicitly at least, by the value of proportionality (Office of the Attorney General, 2002). Thus, according to some of its guidelines, the work of a “cooperating witness” must be “be entirely voluntary”, and the agency commits, “but cannot guarantee”, that his or her identity “will not be divulged”. Before engaging him or her, authorities must run a “suitability determination”, which assesses “the risk of physical harm that may occur to the person or his or her immediate family or close associates as a result of providing information or assistance” to the DEA (2002: 8). Yet it seems clear that operations on the ground do not always abide by the protocols in the books. In the case of Allende, obtaining cellphone identification numbers (the benefit) at the expense of a massacre (the cost) was a blatantly disproportionate operation. Before settling this point, however, we must address two possible rejoinders to this view.

First, it may be argued that the massacre was of the Zeta's making, not the DEA's, and therefore one cannot hold the agency responsible and claim that its actions violated proportionality constraints. The DEA unleashed the ire of the criminal organization unintentionally, when its strategy involving the collaborator was leaked. But, so the argument goes, leaks of that sort may occur in any intelligence action, and the same is true of retaliation. Both phenomena are part and parcel of intelligence work. The response to this rejoinder is that while it is true that the origin of the retaliatory violence is the criminal group (with the complicity of some state actors), and that it was certainly not the intention of DEA agents to wreak havoc on the local population, they should have foreseen the potential outcomes of their actions in view of the features of the context in which they operated. Because they turned a blind eye to them instead, DEA agents failed to properly acknowledge the epistemic uncertainty surrounding the consequences of their actions and wrongly discounted the possibility that informational leaks would have a detrimental impact on collaborating witnesses and even innocent bystanders. What were the features of the context that should have led the DEA to a more cautious approach?

One is Mexico's background institutional conditions. As has been discussed throughout this article, the Mexican state is only able to provide sub-optimal security conditions in many regions of the country. As Kelly and Serrano (2012: 1–25) argue, Mexico has become a security failure. This is because its criminal justice system evolved as an authoritarian tool to control political opponents and the prosecution of minor delinquents. After the transition to democracy in the late 1990s and early 2000s, and the rise in drug violence in the last 20 years, the system was unable to build institutional capacities to combat drug crimes. The front lines of the “war” against drug crimes (police forces, courts, and the penitentiary system at the local and federal levels) were unprepared to investigate and prosecute these crimes, and to resist the corrupting power of drug organizations. It is no wonder that global indexes show Mexico to have an alarming level of impunity, as perpetrators tend to evade criminal and civil responsibility (Global Americans, 2021). Thus, local security forces and law enforcement agencies are too burdened or too corrupt to avoid informational leaks, even in the case of intelligence units trained with the assistance of the DEA itself, like the SIU. They cannot be trusted to always process sensitive information confidentially or to protect citizens adequately.

The second feature of the intelligence context under discussion relates to the kind of violence unfolding in the region. Criminal rackets such as the Zetas rely on a combination of selective and indiscriminate violence in their attempt to exercise political and social control over the territory and its population, in defiance of formal authorities. ¹² They recognize how destabilizing informants can be in the achievement of this goal, which explains why they deal with them in a particularly harsh manner. Harsh treatment of collaborators or those in their circle amounts to a deterrence mechanism: a way to discipline the ranks, discourage ordinary civilians from assisting formal authorities (Kalyvas, 2012), and even a strategy to emulate a state that dispenses sanctions when appropriate (Dudai, 2018). The Zetas “punished” collaborators for precisely these reasons (Smulders et al., 2017), ¹³ and such “punishments” are one of the multiple sources of violence in regions like Coahuila. Where these dynamics of control and deterrence are in place, it should not come as a surprise that failed intelligence can go awry in ways that cannot be anticipated.

The second rejoinder to the charge that in the Allende case the use of witness collaborators was a disproportionate measure is that, if one takes a broader, long-term view of the security situation in the region where events unfolded, and if one takes the massacre to be a failed step in a series of actions that eventually led to successfully ending Zeta dominance, then arguably, taken as a part of the whole, the intelligence leading to the massacre was proportionate, the long-term benefits outweighing the immediate costs. This rejoinder is unconvincing. It may very well be true that decapitating the Zetas did have, as part of a broader strategy, an overall positive impact on the abatement of social violence (although this might be disputed, as some believe that decapitation of criminal organizations leads to further violence rather than to its reduction). However, the question is whether the state may permissibly engage in any kind of action, in this case any intelligence interventions, for the sake of a greater good. It may not.

The state certainly has the duty to assist and protect citizens. Yet, as a general rule, it should not provide such assistance and protection by deliberately inflicting harm on other people. This side-constraint ought to limit the range of permissible intelligence actions for law enforcement agencies. Foot (1967) articulated this view in a classical essay in political philosophy. She argued that an agent must not perform a positive duty (providing assistance) by means of breaching a negative duty (refraining from harming others). Thus, in her example, a doctor may deny a life-saving medicine at the highest dosage to one severely ill patient with grim survival prospects and instead administer it at a lesser dosage to five other patients with positive recovery prospects instead; but the doctor must not take the life of the same severely ill patient in order to donate her organs and save five other people who need them. The latter but not the former course of action would amount to breaching a negative duty to perform a positive duty. Returning to the Allende case, imperiling the lives of a few witness collaborators in order to achieve intelligence gains that may save lives down the line just is a case of violating a negative duty (directing harm towards innocent civilians) in order to perform a positive one (ridding the region of Zeta's brutal dominance), and by the same token is morally impermissible. For these reasons, in balancing the cost and benefits of the use of collaborating witnesses, DEA agents ought to have adopted a restrictive view, which presumes its disproportionality unless there are strong reasons to the contrary. They did not, with tragic consequences. In sum, the DEA's use of collaborating witnesses in this case was justified neither on the basis of the principle of necessity, since less harmful alternatives were available to the agency, nor on the basis of the principle of proportionality, since it is unclear that the harm to the collaborators and civilians (through recklessness) outweighed the gains in terms of violence-reducing intelligence.

Conclusion

This article provided a framework to ethically evaluate the use of witness collaborators in Mexico. The framework can be extrapolated to other cases where law enforcement authorities must assess the ethical stakes of the practice in settings of weak institutions, systematic violence, and significant socioeconomic disparities. Incidentally, the framework should also draw attention to some of the objectionable features of the criminal justice system to which it is applied, most notably the lack of concern of its operators for the well-being of those who stand before it as the accused party. This is the case in Mexico, but also in other developing nations (Colombia or more recently the Philippines come to mind) whose criminal justice systems have a legacy of mistreating suspected offenders, who oftentimes are believed to “deserve” whatever violence befalls them.

At a more abstract level, the article raised two points. First, the case of Mexico furnishes a critique of the just intelligence theory while at the same time utilizing it as a standard to morally assess the state's use of collaborators. The article reveals a limitation of the paradigm in contexts of massive violence and institutional deficits. Second, the use of collaborators raises a moral predicament that has surfaced in related scholarly debates, for instance in relation to the use of torture (Shue, 1978; Walzer, 1973), the proper balance between liberty and security (Waldron, 2003), and the so-called criminal law of the enemy (Jakobs, 2014): whether it is acceptable to violate the rights of a (culpable) few in order to make strides in criminal investigations, thus solving specific crimes and most importantly serving collective goals such as abating crime rates and curbing violence. The study of the use of penal rewards for cooperation in contexts of drug-related extraordinary violence raises some of these same concerns and should therefore help to illuminate them.