From 2012 to 2015, over forty thousand people in Mexico armed themselves to defend their communities from cartels. The movement consisted of mostly of rural men, but some women, and even minors, joined armed defense groups which are, essentially, vigilantes. By 2014, one or more vigilante group existed in at least nine of Mexico’s thirty-two states. Vigilantes sometimes detained alleged criminals and, in some cases, police or soldiers they believed worked for or enabled cartels. Although most of their actions were not lethal, vigilantes sometimes engaged criminals, and even local police suspected of working for a criminal organization, in full-scale firefights. Over one hundred people died in vigilante-related violence during this period.
While a government might be expected to try to stop armed movements that occupy towns with the goal of “cleansing” them of criminals, Mexican officials initially tolerated them. They even allowed federal security forces to offer some vigilante groups operational support. This initial tolerance created the political opportunities for the movements’ rapid expansion in multiple states. Armed citizens in the neighboring states of Guerrero and Michoacán, however, were the epicenter of the vigilante movement. Despite many contextual similarities, the movements’ organizational forms, trajectories, and death tolls varied. I explain why.
Given that armed groups in both states participated in the same general vigilante mobilization cycle between 2012 and 2015, the central question motivating this research is, why was the armed conflict more lethal in Michoacán than Guerrero? This difference, moreover, is counterintuitive given Guerrero’s enduring and more recent history with armed guerrilla movements and politically motivated assassinations. I show not only that armed groups de-escalated more quickly in Guerrero than in Michoacán, but also that Guerrero’s movement leaders and their affiliated “community police” remained strategically defensive over time, retreating in a disciplined way when the federal government deployed the military to stop their protests. My study, thus, contributes to scholarship on why some armed movements de-escalate while others produce more violence by escalating.
My findings are twofold: First, I demonstrate that movement fragmentation led to greater violence both within and between states, and these results are consistent with theoretical expectations. Second, I show that the unequal death rates associated with the two movements also reflects strategic differences, themselves influenced by each movement’s organizational form. With few exceptions, the finding concerning the importance of organizational form has not been fully appreciated by scholars of violent political contention. As such, I contribute to a growing understanding of the role of organizational form in escalation and de-escalation dynamics in violent contention. I submit that intraorganizational arrangements structure social relations, including with base constituents, and these prove as important as intergroup dynamics in producing violent spirals.
In what follows, I define vigilante movements and explain relevant theories of violent political contention. Before describing the data for my comparative analysis, I provide contextual information about the cases. Then I compare the two movements, describing their tactics and targets over time. I contrast my findings to possible alternative explanations before concluding with inferences that speak to studies of the lethal outcomes of violent political contention.
The scholarship
Armed groups that mobilize to defend their communities are vigilantes, although in Mexico such groups take a number of forms, some with complex legal statuses, as described later. Unlike actors who participate in spontaneous and episodic acts of vigilantism, armed community defense groups that sustain organized collective action over time are vigilante movements. (Fuentes Díaz, Gamallo, and Quiroz Rojas Reference Fuentes Díaz, Gamallo, Quiroz Rojas and Gammalo2022,16). As such, they develop strategy, establish a division of labor, and authorize leaders to make decisions. But unlike typical movements that develop elaborate political ideologies for change, vigilantes tend to react to local security threats that, from their viewpoints, states fail to address (Goldstein Reference Goldstein2003; Trevizo Reference Trevizo2022; Le Cour Grandmaison Reference Le Cour Grandmaison2016).
Thus, while vigilante movements arm themselves, they are not guerrillas who seek to overthrow a state. Neither are they militias in the sense of being organized “on behalf of a political actor” (Schuberth Reference Schuberth2015, 306) or paramilitaries mobilized and systematically trained by some faction of the state (Mazzei Reference Mazzei2009, 4). As noted, vigilante movements merely seek to provide security for their local communities in a territorially delimited way. Given their efforts to fill a real or perceived security gap, many see themselves as the state’s interlocuters (Le Cour Grandmaison Reference Le Cour Grandmaison2021, 146).
Despite efforts to coproduce security at the community level, vigilante movements explicitly or implicitly censure the state’s security failures. And notwithstanding their cooperative stance vis-à-vis the authorities, their “relation to the state is not [one of] absolute subordination to its rationale, … but contentious with the state in various ways and at various times,” as Fuentes Díaz and Fini (Reference Fuentes Díaz and Fini2021, 87) argue (see also Cruz and Kloppe-Santamaría Reference Cruz and Kloppe-Santamaría2019; Curry and de Vries Reference Curry and Ansems de Vries2020). This cooperative but also contentious relation with the state has been documented in a number of places in Latin America and Africa, but also in the United States, with its anti-immigrant border vigilantes. Given their sustained protests over time, the groups analyzed in this study—including those funded by the local state of Guerrero to organize indigenous forms of policing—illustrate such contradictory relations with the state. But even vigilantes who do not sustain disruptive political contention unwittingly undermine the state’s claim to a monopoly on coercive force by arming themselves (Goldstein Reference Goldstein2003; Sierra Reference Sierra2017, 194–195; Fuentes Díaz and Fini Reference Fuentes Díaz and Fini2021, 88). Moreover, wherever they exist, vigilantes contribute to public insecurity by taking the law into their own hands, and especially so when their armed movements injure and kill.
Precisely as armed movements, they have implications for research on political violence. In the past twenty years, this scholarship has focused on the escalation dynamics between opposed groups (Alimi, Bosi, and Demetriou Reference Alimi, Bosi and Demetriou2012; Della Porta Reference Della Porta2018), between activists and the state’s security forces, and between splinter groups resulting from movement fragmentation. Della Porta (Reference Della Porta2018, 463) explains that violent protest tactics result from gradual intergroup escalation spirals, which are frequently, but not always, initiated by police violence. In keeping with this emphasis on intergroup dynamics in violent spirals, research also shows that some groups adopt violent tactics to distinguish themselves from their competition. This competition increases when movements fragment, usually toward the end of a protest cycle (Alimi, Bosi, and Demetriou Reference Alimi, Bosi and Demetriou2012; Krause Reference Krause2013; Vogt, Skrede Gleditsch, and Cederman Reference Vogt, Skrede Gleditsch and Cederman2021). Movement fragmentation can also lead to “chain-ganging,” for example, when nonviolent groups get sucked into violence started by others (Krause Reference Krause2013; Vogt, Skrede Gleditsch, and Cederman Reference Vogt, Skrede Gleditsch and Cederman2021). The first hypothesis follows from this latter work:
H1: The greater the movement fragmentation, the greater the violence.
Organizational resources and even their forms, or modes of organizing, also matter for dynamics of escalation and de-escalation. First, armed groups need financial resources to sustain violent conflict (Bosi, Della Porta, and Malthaner Reference Bosi, Della Porta, Malthaner, Chenoweth, English, Gofas and Kalyvas2019, 3). Second, leaders themselves are human resources (Della Porta Reference Della Porta2018, 466; Weinstein Reference Weinstein2006; Bosi, Della Porta, and Malthaner Reference Bosi, Della Porta, Malthaner, Chenoweth, English, Gofas and Kalyvas2019), and some are more experienced than others at reassessing threats in evolving and volatile conflict situations (Dudouet Reference Dudouet2012). Because threat assessment has many dimensions, experienced movement leaders might de-escalate more quickly than inexperienced leaders. Finally, organizational form also affects strategy—that is, decisions about tactics, targets, as well as to escalate or de-escalate. Ganz (Reference Ganz2000, 1016–1017) maintains that organizational characteristics determine the regularity with which leaders meet their constituents, how deliberation occurs, whether it is binding, as well as how groups mobilize their human and financial resources. Organizations whose leaders regularly consult constituents in binding deliberation are more accountable to their base than are organizations whose leaders do not regularly consult their base (Dudouet Reference Dudouet2012). Regular consultation also results in greater collective ownership of decisions, trust, and discipline (Dudouet Reference Dudouet2012), dimensions of solidarity that reduce the odds of organizational splintering, violent competitive outbidding, and “spoiler violence” (efforts to sabotage de-escalation).
Yet only a few studies have documented the relationship between organizational form and escalation or de-escalation dynamics in armed conflict. One study shows that patron-sponsored armed movements are more likely to engage in indiscriminate violence than financially constrained groups (Weinstein Reference Weinstein2006). Well-resourced “opportunistic” groups are theorized to be more violent than resource-poor groups because they recruit by offering selective incentives. Because this is not a careful vetting process, selective incentives undermine both discipline and accountability (Weinstein Reference Weinstein2006). In contrast, resource-poor armed groups typically mobilize the “social endowments” of their local communities by appealing to solidarity when requesting food, information, and labor. Shared cultural, religious or secular ideologies, or simply established “norms of cooperation and reciprocity” contribute to organizational accountability and discipline and, thus, reduce violence (Weinstein Reference Weinstein2006, 7–8). Al-Hashimi and Goerzig (Reference Al-Hashimi, Goerzig, Heitmeyer, Haupt, Malthaner and Kirschner2011, 468) agree, explaining that armed groups that depend on the local population are more likely to de-escalate vis-à-vis a well-matched opponent because leaders develop “sympathy” for their constituents. They concur with Weinstein that financially independent groups without bases of local support are less likely to de-escalate, explaining that such recruits are treated as mere instruments in a struggle in which victory is more important than protecting community. Similarly, Gutiérrez Sanín (Reference Gutiérrez Sanín2008, 4) argues that well-resourced groups “wield violence more indiscriminately.” He concludes that “accounting for the ‘organizational ecological variety’ of non-state armies appears to be a requisite to understand properly the dynamics and meaning of civil war and lethal political violence” (Gutiérrez Sanín Reference Gutiérrez Sanín2008, 29). In short, some research on armed movements shows that organizational form affects lethal violence via the following mechanisms: vetting, accountability to and solidarity (“sympathy”) with the base, and discipline. The hypothesis follows:
H2: Patron-sponsored armed groups will engage in greater violence than community-sponsored armed groups.
Finally, political process theory maintains that beyond their internal characteristics, and beyond their role in escalation spirals, a movement’s backdrop conditions also matter for their trajectories and outcomes. That is, contextual variation—for example, differences in the degree of political support by elite allies, or in the risk of violence—can explain whether and how contention occurs and, thus, movement outcomes (McAdam and Tarrow Reference McAdam, Tarrow, Snow, Soule and Kriesi2019). Therefore, before concluding, I assess two plausible alternative explanations. First, I examine whether partisan politics affected President Peña Nieto’s responses to the movements in ways that might explain their unequal death rates. Second, I assess whether the differences in the criminal contexts—the fact that there were eleven cartels in Guerrero, but only one in Michoacán—drove the variation in each movement’s degree of lethality.
Method of analysis and sources
I conduct a comparative analysis focused on explaining the differences in lethal violence between the armed movements in Guerrero versus Michoacán between 2012 and 2015. Because my two cases are subnational variants of the broader phenomena of armed vigilante movements (in which, as noted, ordinary citizens organize collectively to redress security gaps), the logic of my comparison follows the “most similar case design” and controls for the following four conditions: First, both movements articulated grievances about criminal insecurity caused by a weak state. Second, the role of Mexico’s central state, given that one president had to respond to both movements. Third, since the armed mobilizations occurred in the same period, 2012–2015, the comparison also controls for temporality. Fourth, there is a degree of regional control because I compare movements in neighboring southern pacific coastal states. Indeed, many mobilizations occurred in the Tierra Caliente region traversing Michoacán and Guerrero and the municipalities of that region in both states share many geographic, economic, political, and social-cultural characteristics (Trejo and Ley Reference Trejo and Ley2021).
Although the vigilante movements under study are comparable, they varied in some important ways, such as organizational form. I leverage this difference by asking whether the variation in lethal violence resulted from the organizational characteristics and corresponding strategies of the two vigilante movements. Aside from this independent variable, there were two other contextual differences that could, per political process theory, explain the variation in movement fatalities. Therefore, before concluding, I consider the potential alternative explanations that state-level differences in partisan alignment or criminal context drove the intensity of vigilante-related fatalities in each state.
I used a variety of sources for this investigation, but especially my own small data set on vigilante collective actions from 2012 through 2015. Intense media coverage made it possible to track the escalation and de-escalation dynamics of concurrent vigilante movements over time by engaging keyword searches for “autodefensas” or “vigilantes” in Google and LexisNexis. While there is much self-censorship, especially among local Mexican news sources, national newspapers have more autonomy, and their reporting was enhanced by the fact that vigilante events were public, sometimes violent, spectacles. Most of the news reports that I coded were from Proceso (57 percent of all articles), El Universal (14 percent), and Milenio (at 13 percent), respectively. A student and I coded the news reports of 134 events according to where and when it occurred, the vigilantes’ tactics, their targets, as well as whether violence occurred. If so, we coded the number of people injured or killed, as well as who initiated the violence, inferring escalation and de-escalation from the number of such events and tactics over time. I ensured that no event was counted twice.
I also analyzed two investigative reports by Mexico’s politically independent National Human Rights Commission (CNDH). In 2013, a team of thirty-five CNDH researchers began to visit Guerrero where they interviewed 295 people, including those who organized themselves to defend their communities. An even larger team of seventy-two CNDH researchers traveled to Michoacán to interview 3,027 victims from 183 communities; 316 interviews were with vigilantes. These interviews shed light on how Michoacanos met vigilante leaders, why they may or may not have joined the movement, and whether they were paid to participate. If the CNDH’s report was more extensive for Michoacán than for Guerrero, the secondary sources on the latter’s more established community defense groups supplemented information. Both the primary and secondary sources allowed me to triangulate information.
The three-year span under study illuminates both inter- and intragroup dynamics over time and, thus, some of the causal complexity motivating escalation or de-escalation, including whether and how organizational form affected the extent of lethal violence in two states. Before presenting my findings, I first describe the criminal-political contexts of each state to explain why Michoacán and Guerrero were the epicenter of Mexico’s vigilante movement from 2012 to 2015.
The vigilante movements’ criminal-political contexts
The CNDH described the vigilante movements as emerging in contexts where the rule of law was nearly absent. Some towns lacked police, or had forces that were either intimidated by, or actually worked for, criminal organizations. The cartels’ economic and political power had increased with Mexico’s democratization, enabling criminals to put police on their payroll, to influence elections and even to establish criminal governance in both states (Trejo and Ley 2015; Reference Trejo and Ley2016).Footnote 1 Intercartel warfare increased when President Felipe Calderón declared war against them in late 2006 because jailing leaders multiplied criminal organizations, which in turn only intensified their competition for territory and/or for larger shares of the markets in drug and human trafficking.
Because no fewer than eleven cartels fought for dominance in Guerrero between 2012 and 2015, the homicide rate there was second only to Chihuahua. Further, many of Guerrero’s local state institutions were, as noted, corrupted by cartels. To illustrate, the state’s governor Ángel Aguirre Rivero resigned from power in 2014 after the enforced disappearance and murder of Ayotzinapa students spotlighted the alliance of drug traffickers, Iguala’s mayor, and local police forces. Many other mayors and ex-mayors were also linked to organized crime in 2014 (de Buck Reference De Buck2019, 113), and the police forces of at least fourteen municipalities were found to work for cartels (Sánchez Valdés Reference Sánchez Valdés2015, 15).
While Michoacán faced similar problems, one hegemonic cartel, the Caballeros Templarios (henceforth, Templarios), became the de facto authority in much of the state. And where the Templarios backed successful candidates on or after the November 2011 gubernatorial and mayoral elections, they directly influenced the legal authorities. To illustrate, the state’s former interim governor, Jesús Reyna-García, of the PRI (Institutional Revolutionary Party), was arrested in 2014 and ultimately convicted for his links to the cartel. Meanwhile, Fausto Vallejo, the elected governor, resigned from power in mid-2014, shortly after returning from medical leave amid rumors of back-channel connections to the Templarios (his son was arrested for cartel ties in 2014). The Templarios also influenced judicial institutions and many mayors in that state (CNDH 2016, 185–196). Thus, whether through their own members, their allies in municipal and state government, or the ranks of local police forces (Curry and de Vries Reference Curry and Ansems de Vries2020; CNDH 2016), the Templarios held sway in much of the state, but especially along coastal and Tierra Caliente areas. Their criminal governance was so secure that a cartel leader nicknamed “La Tuta” demanded that Michoacán’s mayors turn over 30 percent of their municipal government budgets to the Templarios (Trejo and Ley Reference Trejo and Ley2016).
Ordinary citizens in both states responded to the violence that cartels unleashed by organizing armed community defense movements that neither governor could stop. At least 56 percent of the municipios in both states had either autodefensas or contentious community policing groups during the period under study. Their prototype was the community policing model of various indigenous communities, but as I show, their actual organizational forms differed in ways that affected their strategies—tactics and targets—and, ultimately, movement fatalities.
Findings
Roughly twenty thousand people mobilized in each state, and Figure 1 shows that their movements organized demonstrations, rallies or formed new groups. While the sit-in tactic shown in the middle panel can result in violence, sit-ins are less likely to be lethal as compared to events that collectively display hunting rifles or military-grade weapons. Such weapons were especially on display in Michoacán, as in the barricades vigilantes built to block entrances to towns or at checkpoints (retenes) at or near defended towns. Figure 1 clearly shows that these potentially violent events happened more frequently in Michoacán than in Guerrero.
Figure 1. Number of Events by Vigilantes’ Tactics, Guerrero vs. Michoacán.
Although the vast majority of vigilante actions were not lethal, over one hundred people, including civilians, died between 2013 and 2015. The precise death toll is disputed, but my conservative estimates from news reports are that at least eighty-four people died in Michoacán, and at least twenty-six people died in Guerrero, as shown in Table 1. Returning to Figure 1, the right-hand panel strongly suggests that people were more likely to die in the former state because vigilantes in Guerrero de-escalated more quickly than in Michoacán. Figure 1 specifically shows that Guerrero’s movement began to de-escalate between 2013 and 2014, a period when vigilantes in Michoacán continued to block roads at gunpoint, “take” towns and/or engage in shoot-outs. The more defensive strategy taken by Guerrero’s groups is also evident in Table 1, which shows that they were less likely to violently engage the cartels or the federal security forces than were Michoacán’s vigilantes. Consequently, more people died in Michoacán, and the difference in fatalities in Table 1 is statistically significant.
Table 1. The Percentage Distribution of Vigilante Related Deaths between 2013 and 2015, by Movement Targets
Notes: Author’s data. No deaths reported for 2012. Rounding error explains Guerrero’s 101 percent.
† Sixteen people died when the Templarios ambushed a demonstration.
***p < .01.
Thus, while both movements navigated treacherous criminal contexts, they followed different strategies with regard to tactics, targets, and the timing of de-escalation. In what follows I describe how these strategic differences that led to such divergent lethal outcomes between the movements could have resulted from their distinct organizational forms, as well as degree of fragmentation.
Organizational form, movement fragmentation and violence: Guerrero vs. Michoacán
Roughly twenty thousand people in Guerrero mobilized to defend their communities from cartels between 2012 and 2015. By 2014, fully fifty-three of eighty-one municipios had a group, whether organized as community police, an autodefensa, or—as with the majority—with features of both (Sánchez Valdés Reference Sánchez Valdés2015, 13). Whereas community police organize through indigenous institutions, many with legal recognition, some autodefensas were eventually authorized to function as “citizen police” forces (CNDH 2013; Sierra Reference Sierra2017; Fuentes Díaz and Fini Reference Fuentes Díaz and Fini2021, 92). In practice, many defense groups organized in the gray area between the two types and were not, therefore, legally sanctioned (Fini Reference Fini2022, 250). Moreover, if they carried high-caliber weapons, they acted illegally. Variation and complexity aside, most of Guerrero’s community defense groups were affiliated either with the CRAC-PC (Coordinadora Regional de Autoridades Comunitarias de la Costa Chica–Montana de Guerrero, Policía Comunitaria) or its main offshoot organization, the UPOEG (Unión de Pueblos y Organizaciones del Estado de Guerrero). Still, roughly one-third of all armed defense groups were independent of the two organizations (CNDH 2013, 14–15).
The more established and respected organization, the CRAC, began organizing community police according to indigenous traditions in the mid-1990s in Guerrero’s Costa Chica–Montaña region (Vasconcelos Reference Vasconcelos2020). By 1998, officials authorized them to detain alleged criminals, as well as to prosecute, judge (in local justice tribunals), punish, acquit, forgive, or reeducate “delinquents” (CNDH 2013, 54–55; Sierra Reference Sierra2015; Ley et al. Reference Ley, Mattiace and Trejo2019, 192). The CRAC thus created a parallel criminal judicial system, one partly dependent on the state’s budget (Sierra Reference Sierra2015, 136). But although the CRAC is legally recognized and receives state funds to administer criminal justice according to indigenous traditions, most of its leaders are experienced leftists (Fini Reference Fini2019, 54–59; Hernández Navarro Reference Hernández Navarro2020, 8–10; Sierra Reference Sierra2015, 134) committed to political autonomy. As such, the CRAC regularly organizes protests, including against mining concessions (Vasconcelos Reference Vasconcelos2020; Fuentes Díaz and Fini Reference Fuentes Díaz and Fini2021) and, in some instances, has even temporarily arrested state officials (Sierra Reference Sierra2015, Reference Sierra2017). Thus, despite cooperating with the authorities, the CRAC has a history of being repressed by state officials (Vasconcelos Reference Vasconcelos2020; Gaussens Reference Gaussens2021; Sierra Reference Sierra2017). Most importantly, CRAC leaders have rejected their local state government’s offer to deputize them on multiple occasions (Fini Reference Fini2019, 67–68; Sierra Reference Sierra2017, 208) because official incorporation would violate the principle of community autonomy, community policing, indigenous rights, customary law, and, for many, their very ethnic identities.
Despite being an interethnic federation, the CRAC represents constituents in a locally embedded and deeply consultative way. Briefly, the community police (PC), their regional commanders, and the commissioners (or prosecutors) are elected via village assembly according to indigenous “cargo” traditions prescribing unpaid community service (Gaussens Reference Gaussens2019). Village police serve one or two years on a rotating basis without financial compensation (Fini Reference Fini2019; Benítez Reference Benítez Rivera2015, 48; Sierra Reference Sierra2015) and have strong ties to many of the people they serve. Because community members help their police execute their duties (Sierra Reference Sierra2017; Gaussens Reference Gaussens2019), service obligations reinforce trust, collective solidarity, and identity. Similarly of the community, PC commanders coordinate policing work; while the CRAC’s Executive Committee members coordinate joint policing operations in different pueblos (and register the PC’s weapons with the army) (Sierra Reference Sierra2015, 137). Such coordination requires regular deliberation, but the Executive Committee is not excessively centralized (Fini Reference Fini2018, 52). Further, because the community police live in the villages where people elect them, the latter maintain an oversight role. Thus, villagers not only carefully vet police recruits (Ley et al. Reference Ley, Mattiace and Trejo2019) but also discipline them and their commanders by potentially shaming—even expelling—those who fall short of expectations. To illustrate, a PC coordinator, Eliseo Villar, was expelled by village assembly in 2014 for not being transparent about his use of government funds (Elizondo and Torres González Reference Elizondo and Paola Torres González2018, 122). In fact, the CRAC’s history of organizing extralocally since 1995 developed “multiple sources of oversight” that protect CRAC affiliates not only from corruption (Ley et al. Reference Ley, Mattiace and Trejo2019, 194) but also from cartel penetration in an entire region (Sierra Reference Sierra2015). The CRAC’s organizational form thus created a disciplinary support system that both holds people accountable and reinforces trust, including across linguistically distinct indigenous communities at the regional level.
But outside of the CRAC’s strongholds, increasing cartel harassment led to differences about who, how, why, and where to organize, and these disagreements split the CRAC in 2011 (Fini Reference Fini2019, 76). A flurry of collective actions followed. Although mestizos (nonindigenous people) live in some of the municipalities represented by the CRAC (Sierra Reference Sierra2015), UPOEG leaders sought to create “citizen,” rather than indigenous, police forces by organizing urban centers, as well as among nonindigenous groups, including Afro-Mexicans. UPOEG leaders successfully negotiated with Guerrero’s state government to classify them as “auxiliary” citizen police (CNDH 2013, 60), even though their agenda exceeded community defense (Fini Reference Fini2019, 76; de Buck Reference De Buck2019, 109–116). By 2013, the competition between the two federations led to the formation of many new self-defense groups. In the first six months of 2013, the number of CRAC affiliates grew from 107 communities (in thirteen municipalities) to more than 190 communities (in twenty-two municipalities) (Sierra Reference Sierra2015, 152). Where newly organized communities had weak or absent indigenous cargo and assembly traditions, the CRAC instituted popular assemblies (Fini 2022, 253; Reference Fini2018, 48–49), including in urban areas where it organized neighborhood assemblies (Elizondo and Torres González Reference Elizondo and Paola Torres González2018, 117). The CRAC withdrew support where internal conflict doomed collective decision-making (Fini Reference Fini2022). The CRAC, in short, prioritized regular consultations with community members whose assemblies also elect the police forces on which they depend for free labor.
The UPOEG similarly made rapid and significant gains, initially organizing autodefensas until Guerrero’s government authorized them as “citizen” police forces. Between 2013 and 2015, the UPOEG organized in twenty-one municipalities that comprised both indigenous and mestizo communities (De Buck Reference De Buck2019, 109). Given their history with the CRAC, many UPOEG leaders similarly emphasized the participation and oversight of community members by insisting on the village or, in urban centers, popular assemblies (Fini Reference Fini2022, 265; de Buck Reference De Buck2019, 118–26). Their rapid expansion, however, led to poor training, which, in turn, led some UPOEG affiliates to use unjustifiable deadly force at roadblocks in two separate incidents (CNDH 2013, 65–66). The rapidity of their expansion as well as the fact that they organized many people without strong indigenous communal institutions, several in urban places, also made it possible for members of organized crime to infiltrate, and further fragment, their movement (Ferri Reference Ferri2023; de Buck Reference De Buck2019, 117). The Frente Unido por la Seguridad y el Desarrollo del Estado de Guerrero (FUSDEG) split from the UPOEG in late 2014, and the defectors carried the high-caliber weapons typically associated with organized crime. In contrast to both the CRAC and UPOEG, the FUSDEG was loyal neither to the popular assembly model nor to electing community police members. Rather, it offered community protection in exchange for “donations,” a practice many called extortion. Because of their arbitrary and violent behavior, the people of Petaquillas expelled the FUSDEG (Serrano Reference Serrano2019); the splinter group was, ultimately, short-lived (Ferri Reference Ferri2023).
Regarding lethal violence, my data show that the overwhelming majority of vigilante-related deaths in Guerrero (twenty of twenty-six cases) happened in 2015 during shoot-outs involving FUSDEG members. Seven people died in March 2015 when FUSDEG and UPOEG affiliates clashed, and thirteen more died in June of that year in a firefight between two FUSDEG factions (Flores Contreras Reference Flores Contreras2015). My finding that intravigilante conflict drove lethal violence in Guerrero is consistent with Fini’s (Reference Fini2022, 253) fieldwork, and our independent conclusions are consistent with theoretical expectations that movement fragmentation leads to violence, as predicted by the first hypothesis.
But since none of the deaths in my data (or CNDH report) implicates the CRAC, the evidence illuminates a defensive strategy, one plausibly related to its organizational form—specifically its dependence on and embeddedness within the communities it represents. This defensive strategy is also evident in the CRAC’s response to military aggression in August 2013, when the army, navy and state police deployed to detain dozens of community police. The CRAC leader Gonzalo Molina González explained that CRAC affiliates remained “prudent” in the face of this military sweep precisely to “prevent bloodshed” (Reyes Reference Reyes2013). A Guardian correspondent corroborates his analysis, documenting the CRAC’s disciplined retreat on August 21 as follows: “Hoping to avoid a full-blown crackdown, most of the Guerrero self-defence groups have since retreated into their strongholds and hidden away their higher caliber guns. But there is little evidence that they are about to fade away meekly.” She then quotes a vigilante who explains: “The army wanted to detain and disarm us so we have to be careful now, but this is still our territory…. If the army ever tries to come in and shoot at us, then we will shoot back” (Tuckman Reference Tuckman2013). Days later, armed CRAC affiliates from Tixtla marched to the capital to demand the release of those detained, but they fled in the face of overwhelming military and state police force (Gatica Polco Reference Gatica Polco2018, 143). With tensions high, some armed CRAC affiliates then occupied Tixtla’s City Hall on August 25 but left without further violence after negotiating with state government officials (Gatica Polco Reference Gatica Polco2018,143). Because the three tactical retreats are consistent with the CRAC’s well-documented nonconfrontational stance toward cartels (Hernández Navarro Reference Hernández Navarro2020, 112–114; Ferri Reference Ferri2023), and because no fatalities were documented in any of these episodes, it is fair to conclude that the CRAC was strategically defensive. Its strategy is consistent with theoretical expectations that armed groups that depend on the community’s participation are less violent. Certainly for the leader Molina González, the CRAC’s “prudence” was honorable.
Although the UPOEG’s splinter groups proved less disciplined than the CRAC, they were, nevertheless, far less violent than the FUSDEG. My data indicate that when UPOEG affiliates mobilized, they were far more likely to detain people, including local policemen whom they suspected of working for a cartel, without lethal force. UPOEG activists even managed to evade hundreds of soldiers, state, and federal police without incident during a demonstration of roughly one thousand people demanding that the military leave El Ocotito (Flores Contreras Reference Flores Contreras2014). The UPOEG’s restraint in the face of a military deployment is similarly consistent with the hypothesis that armed groups dependent on community are less violent.
In summary, most of Guerrero’s community defense forces were led by experienced activists distrustful of the government, military, and the cartels even if they negotiated with local officials because they depended on public finances for some of their work. But the CRAC and UPOEG also depended on communities for free labor. The village and popular assemblies they organized not only vetted their defense forces, but villagers also supported and disciplined them in the towns where they lived. This disciplinary support system undoubtedly reinforced solidarity, or “sympathy” for the base, motivating de-escalation to “avoid bloodshed” in the face of military sweeps. As shown in Table 1, most vigilante groups in Guerrero demobilized a full year before their counterparts did in Michoacán when facing comparable military deployments. The FUSDEG, in contrast, neither routinely elected its forces through general assembly nor regularly consulted the communities it policed (Serrano Reference Serrano2019). The group was expelled from at least one community, withered elsewhere, but not before involving itself in two massacres. This variation in violent outcomes within Guerrero is consistent with theoretical expectations that armed groups that do not depend on community are more likely to engage in violence, as per the second hypothesis. Beyond the organizational differences that predisposed the FUSDEG to violence as compared to the CRAC or even the UPOEG, lethal force also happened in the context of the movement’s fragmentation. The evidence from Guerrero, in short, is consistent with both hypotheses.
Michoacán’s vigilante movement took some inspiration from an indigenous community’s success at expelling Templarios from Cherán (Le Cour Grandmaison Reference Le Cour Grandmaison2016; Fini Reference Fini2022, 247), as well as from the broader legitimacy of the community police “label” (Le Cour Grandmaison Reference Le Cour Grandmaison2016). Most of Michoacán’s vigilantes, however, were rural mestizos, although a few indigenous pueblos created new community police during the period under study (CNDH 2016, 145; Fuentes Díaz and Paleta Reference Fuentes Díaz and Pérez2015). Inspiration notwithstanding, Michoacán’s vigilante leaders created an organizational structure that differed from the community police model. And given their goal of “cleansing” their communities of cartel operatives, they pursued what they called war against the Templarios.
Their bellicose strategy reflects the organizational attributes shaped by their movement’s business patrons. Newspapers widely reported that, along with small and large shopkeepers, large landowners, as well as owners of orchard, mining, logging, shipping, and packing companies, paid for logistics (e.g., radios, gas, food), high-caliber weapons, and some vigilante labor. Some were paid $200 Mexican pesos per day; others who lived near Apatzingán, a cartel stronghold, reported being paid $1,700 pesos per week (CNDH 2016,161; Manzo Reference Manzo2022, 340). Business leaders financed and led the movement to regain operational control of their businesses or property (e.g., cattle) stolen by the Templarios, which also extorted them (Fuentes Díaz and Paleta Reference Fuentes Díaz and Pérez2015).
To “cleanse” Michoacán of Templarios, vigilante leaders not only hunted down well-known cartel members but also organized “liberation” campaigns by traveling to towns in which they did not live in convoys of anywhere between eighty to two hundred (or more) armed men, frequently with the support of federal police or army units (CNDH 2016). At the town square, vigilantes invited locals to join the movement, and once a group of volunteers formed, the leaders returned to their hometowns. This improvisational approach was not a serious way of consulting all townspeople, nor did it lend itself to careful vetting, accountability, or disciplinary procedures for the autodefensas. The failure to regularly consult civilians was the movement’s modus operandi from the start, as illustrated by the fact that Dr. Mireles (an MD and rancher), Hipólito Mora, and Simón “El Americano” Torres strategized secretly to plan a surprise uprising for February 2013. Meanwhile, the self-defense groups in Tancítaro, led by avocado producers, were described as having an “authoritarian vision” because fifteen armed men believed “that security matters should be dealt with without consultation or prior discussion” (Álvarez-Rodríguez, Román-Burgos, and Jesperson Reference Álvarez-Rodríguez, Román-Burgos, Jesperson, Álvarez-Rodríguez, Román-Burgos and Jesperson2020, 89). As importantly, neither the business leaders nor rank-and-file autodefensas were elected by most townspeople. The CNDH (2016, 281) even documented a couple of cases in which men with high-caliber weapons in barricades were not even community members but hired armed labor. As noted, the failure to regularly consult communities led not only to poor vetting but also to poor accountability, because community members could not report or otherwise discipline bad behavior after leaders left the “liberated” towns with ad hoc autodefensas. Because Michoacán’s leaders failed to train or supervise volunteers, random punishments of alleged cartel members and their families resulted. The CNDH (2016, 279) documented cases in which vigilantes ransacked homes, sometimes pointing their weapons at household members. Behaviors like these led “most” autodefensas to develop a “negative image” among locals, according to Le Cour Grandmaison (Reference Le Cour Grandmaison2021, 149).
Of the thirty-four to thirty-six communities with vigilante groups, fully twenty-four were organized by outsiders (CNDH 2016, 156–158). As importantly, all autodefensas were only loosely coordinated by the “Consejo de Autodefensas de Michoacán” (henceforth, Consejo) (Hernández Navarro Reference Hernández Navarro2020,178; Le Cour Grandmaison Reference Le Cour Grandmaison2021,145). Although local leaders sometimes coordinated town occupations (takeovers), they generally acted autonomously (CNDH 2016,161). The Consejo was not only less centralized than either Guerrero’s CRAC or UPOEG, but the alliances between the groups were “fragile” (Le Cour Grandmaison Reference Le Cour Grandmaison2021, 149). Therefore, just as they failed to regularly consult their base, Michoacán’s vigilantes did not generally deliberate with one another. As Le Cour Grandmaison (Reference Le Cour Grandmaison2021,152–153) explains, “Leaders individually sought to maximize their strategic advantages.” In many ways, this lack of regular deliberation led to unilateral and risky decisions that together led to the movement’s eventual loss of legitimacy and to its fragmentation.
The collateral damage resulting from this loose federation of rather undisciplined armed groups could not have been anticipated by those who, initially, supported the movement. Autodefensas spread rapidly in 2013 as a multiclass movement because at least twenty thousand activists (Hernández Navarro Reference Hernández Navarro2020,178) sympathized with the goal of “cleansing” their community of Templarios (CNDH 2016). Despite initial apprehension, the movement proved appealing because it produced quick security results (Le Cour Grandmaison Reference Le Cour Grandmaison2021, 146) in a context where the Templarios routinely taxed even rural people’s wages and food; they also systematically kidnapped, extorted, and raped (Fuentes Díaz and Paleta Reference Fuentes Díaz and Pérez2015). Given their daily insecurity, many people hedged their bets by donating food or by participating in demonstrations during the movement’s first year. With such strong initial support, over half of Michoacán’s municipios developed a vigilante group (CNDH 2016, 163). But the lack of vetting procedures and decentralized decision-making made it possible for criminals interested in defeating the Templarios to infiltrate their movement (CNDH 2016, 139; Le Cour Grandmaison Reference Le Cour Grandmaison2016, 105). As criminal infiltration became evident in 2014, many Michoacanos withdrew their support for the movement (CNDH 2016, 165–166). Others feared that the autodefensas could become another cartel or that their leaders aimed to become local strongmen (CNDH 2016; Le Cour Grandmaison Reference Le Cour Grandmaison2021, 149).
Regarding lethal violence, the vigilantes’ “cleansing” strategy was, unsurprisingly, met with extremely violent counteroffensives (Hernández Navarro Reference Hernández Navarro2020, 178), such as the massacre sixteen people (some children) when the Templarios ambushed a demonstration in 2013 (Table 1, column 4). But the fragile alliances between key vigilante leaders who did not regularly deliberate also cemented strategic and tactical differences that further fragmented Michoacán’s movement in ways that increased lethal violence. For example, criminals infiltrated the vigilante movement not only because of hasty group formation but also because some leaders opted to integrate remorseful cartel members, the so-called arrepentidos, into their ranks. Absent routine procedures for disciplining poor vigilante behavior or for removing obvious criminals from their ranks, disagreements over the arrepentidos led to a deadly conflict between two factions of the movement that accused one another of having cartel members. This conflict led to at least one firefight between vigilantes in which eleven people died in 2014 (Milenio, December 18, 2014; CNDH 2016, 165).
Yet there were many more casualties from firefights between vigilantes and the federal security forces after negotiations with the federal government broke down, as shown in Table 1. Briefly, military units and/or federal police forces initially backed Michoacán’s vigilantes by supporting or including them in high-risk operations (CNDH 2016). Their alliances weakened after military troops attempted to disarm vigilantes in Michoacán, twice in late 2013, then again in January 2014 (Proceso 2014c). Then, in May 2014 officials asked the vigilantes to join the newly created Rural Defense Forces (RDF) in exchange for putting down their arms (CNDH 2016). Several groupings refused on the grounds that disarming would make them vulnerable to cartel retaliation. The few thousand men who joined the RDF, cartel affiliates among them, were a fraction of the twenty thousand who had participated in the movement (CNDH 2016, 205–212). Clearly the federal government’s legalization plan further fragmented the movement (Manzo Reference Manzo2015; Le Cour Grandmaison Reference Le Cour Grandmaison2021, 148). Those who joined received uniforms, registered their firearms, and agreed to refrain from carrying military-grade weapons publicly. But shortly thereafter, many of those certified exited the RDF to resume disruptive movement tactics from late 2014 through 2015 (CNDH 2016, 212).
After their negotiations with officials broke down, some leaders accused the central government of betrayal and persecution, and Dr. Mireles vowed in social media to go to war against the government: “Michoacán is governed by Templarios. Daily life has changed for the worse [as] today we are persecuted by the Templarios and the government, the Army, the Marines and all police forces…. If the government wants war, it will have a war, it betrayed us. They sat at our tables and said we were their allies … [then] they jailed Hipólito [a vigilante] and intend to jail the rest of the [vigilante] leaders” (Proceso 2014a; author’s translation). The broken negotiations with the central government thus added another well-armed opponent to the conflict. After fifteen vigilantes were detained by the government, the autodefensas threatened to block the roads leading to all towns, warning, “The government wants war, and we’re going to give it to them” (Proceso 2014b). As such, some vigilantes in Michoacán were far more likely than their counterparts in Guerrero to engage in firefights with a federal security force (see Table 1).
Importantly, both the cartels and the security forces were better armed and better trained in violence than were the vigilantes. The gross asymmetry in these violent confrontations that extended for an additional year in Michoacán helps to explain the higher death counts there as compared to Guerrero, where vigilantes remained defensive.
Summarizing the differences between the two movements, Table 2 shows that both fragmentation (H1) and organizational form (H2) correlate with movement-related fatalities. While fragmentation in both states led to lethal violence, Table 2 shows that Michoacán’s movement was far more fragmented than Guerrero’s and that the violence there was greater as a result. While Guerrero’s movement fragmented into three main groupings, dozens of armed groups in Michoacán’s were mostly autonomous. Their unilateral decision-making further fragmented their movement in ways that led not only to intravigilante violence but also to firefights between some autodefensas and the security forces, as predicted by the first hypothesis.
Table 2. Movement Fragmentation and Organizational Form, by State, 2012–2015
Table 2 also suggests how organizational attributes contributed to the fatalities. It highlights that community policing was more deeply embedded in Guerrero, where the movement depended on community, as compared to the autodefensas in Michoacán, which were patron sponsored. The vigilante business leaders from Michoacán were not elected, and with few mechanisms for regular consultation with their base, they were not beholden or otherwise accountable to the people they claimed to defend. Rather, because their livelihoods depended on recovering their businesses, entrepreneurs declared war on the Templarios, a strategy that led to lethal counteroffensives. Further, their impromptu town-square gatherings hastily recruited volunteers, or the entrepreneurs paid for some labor, but neither approach is the careful vetting procedure observed in Guerrero (Table 2, row 3). And because Michoacán’s leaders were not embedded in most of the communities they “liberated,” they could not regularly consult their base nor train or hold their rank-and-file to account for undisciplined behavior, such as the ransacking of homes belonging to alleged criminals (Table 2, row 4). These organizational attributes—specifically the failure to vet, train and discipline recruits—led to behaviors that cost the movement legitimacy among many Michoacanos. Their failure to vet recruits specifically led to significant criminal infiltration and to intravigilante violence. Given their loosely federated organizational structure, they also failed to deliberate regularly, and that failure cemented strategic differences. The latter resulted in further fragmentation, intravigilante violence, and an armed confrontation between Mireles’s supporters and the state’s security forces. In summary, the organizational features of Michoacán’s vigilante movement led to decentralized and bellicose strategies among rather undisciplined armed groups. This finding is consistent with the second hypothesis, which predicts that patron-sponsored armed groups engage in greater violence than community-sponsored armed groups.
Alternative explanations
As noted, political process theory holds that contextual differences, whether political or otherwise violent, could explain variation in movement outcomes (McAdam and Tarrow Reference McAdam, Tarrow, Snow, Soule and Kriesi2019). In terms of partisan politics, by 2012, the incumbent and interim governors (between 2012 and 2015) of Michoacán belonged to the PRI, the same party as Mexico’s then president Peña Nieto, whereas Guerrero’s governor was a member of the left-of-center PRD (Party of the Democratic Revolution). If this vertical partisan alignment illuminates why Michoacán’s vigilantes initially received operational support from the military and federal police, it fails to explain why the federal government “alternated between tolerance, vigilance, and repression,” detaining multiple autodefensas members as early as 2013 (Le Cour Grandmaison Reference Le Cour Grandmaison2021, 145; Curry and de Vries Reference Curry and Ansems de Vries2020). Nor does the federal government’s initial, if inconsistent, support of Michoacán’s movement explain why some vigilantes militarily confronted these same security forces when the government escalated repression. As importantly, the federal government’s sudden switch from tolerance to a more repressive strategy happened in both states, but the movement in Guerrero responded by retreating, whereas many vigilantes in Michoacán remained bellicose. Partisan politics, in short, did not drive movement tactics in the face of military sweeps.
If, as noted, local state capacity was weak in both places, there were important differences in the criminal contexts that could, theoretically, help to explain variation in movement lethality. Given the violent competition for territory among eleven cartels in Guerrero, the homicide rate there between 2012 and 2015 was more than three times greater (average of 64.5 homicides per 100,000 people) than in Michoacán (19.25 homicides per 100,000), according to INEGI (National Institute of Statistics and Geography). Could this stark difference explain the community defense forces’ restraint in Guerrero, as suggested by some research indicating that extremely violent contexts lower the odds of violent movement tactics? The evidence suggests not because people were not significantly more afraid in Guerrero than they were in Michoacán during the period under study. INEGI’s survey data show, for example, that the average percentage of people reporting insecurity was comparable for both states between 2011 and 2015: 79.9 percent for Guerrero’s residents and 78.8 percent for Michoacán’s. The higher homicide rate did not create significantly greater fear in Guerrero than in Michoacán because insecurity is driven by all violent crimes—stolen businesses, extortion, rapes—not just homicides (Fuentes Díaz and Fini Reference Fuentes Díaz and Fini2021). In this regard, Michoacán’s was also a treacherous environment. Finally, the fact that ordinary citizens protested the collusion between cartels and government officials in the aftermath of the Ayotzinapa tragedy by rioting through 2015 is further evidence that Guerrero’s criminal context did not deter violent protest tactics. Rioters not only clashed with soldiers and various police forces but also set fire to a government building to demand that the governor resign (because a cartel-affiliated mayor had local police murder unarmed students). If anything, the Ayotzinapa riots show that Guerrero’s extremely violent context could, in fact, provoke violent protest tactics but failed to do so among that state’s community police.
The evidence shows that the levels of vigilante violence resulted more from each movement’s strategy, as influenced by their organizational form as well as degree of fragmentation, than from their variable political or violent contexts.
Conclusion
Extreme insecurity in Guerrero and Michoacán provoked concurrent vigilante movements whose organizations, trajectories, and death tolls varied. They did so because the loosely federated organizational model of Michoacán’s movement proved more vulnerable to fragmentation than did Guerrero’s movement. Therefore, while fragmentation in both states led to deaths, the violence was much worse in Michoacán, conforming to theoretical expectations (H1) that fragmentation and lethal violence correlate.
But the organizational forms of the groups also led to differences in movement-related fatalities. Guerrero’s community defense groups cohered through a centralized organization whose experienced leaders regularly consulted one another and the communities on which they depended. Both they and the community police were vetted via general assembly and lived in the communities that elected them. This embeddedness created a system of accountability that disciplined Guerrero’s armed actors while also nourishing solidarity. This disciplinary solidarity system kept the community defense mission in clear focus in Guerrero. I have shown that, unlike their counterparts in Michoacán, Guerrero’s defense forces did not hunt down cartel members, and both the CRAC and even the UPOEG demonstrated restraint in the face of military sweeps. In fact, most of Guerrero’s vigilante groups de-escalated when the military deployed, and they did so a full year before their counterparts did in Michoacán when facing comparable military offensives. The decentralized, rather autonomous, patron-sponsored groups led by Michoacán’s businessmen, in contrast, were clearly more confrontational than defensive. They declared war against the Templarios, and one vigilante faction declared war on the state. Because both targets are better-armed experts in violence, the death toll was high in Michoacán. The trajectories and outcomes of both movements thus conform to theoretical expectations that armed groups that mobilize the social endowments of the communities on which they depend de-escalate more quickly than well-resourced, independent, groups and, specifically, that patron-sponsored armed groups will engage in greater violence than community-sponsored armed groups (H2). As such, my Mexico findings support Weinstein (Reference Weinstein2006), Al-Hashimi and Goerzig (Reference Al-Hashimi, Goerzig, Heitmeyer, Haupt, Malthaner and Kirschner2011), and Gutiérrez Sanín (Reference Gutiérrez Sanín2008), who have argued that organizational form contributes to escalation and de-escalation dynamics in violent contention.
These findings, moreover, are net of the political and criminal contexts in which vigilantes mobilized, insofar as neither partisan politics nor number of cartels per state drove the outcomes. Instead, as demonstrated, each movement’s organizational form, as well as degree of fragmentation, proved decisive to the degree of vigilante-related fatalities. My findings thus support Ganz’s argument that organizational characteristics influence social movement outcomes by structuring both the frequency and form of deliberation with the base. Along with Dudouet, I extend Ganz’s theory by showing that infrequent and/or nonbinding consultation with other leaders, rank-and-file activists, or community constituents can make movement organizations vulnerable to fragmentation and, thus, to greater violence. Although scholars have long recognized that fragmentation leads to violence, I show that organizational characteristics can make some more vulnerable to splintering. Simply put, organizational forms and procedures structure intraorganizational dynamics that predispose some armed groups to escalate, and others to de-escalate. Specifically, organizations that routinely consult their base are more accountable to them, as Dudouet argued, and the trust and solidarity that accrue guards against violence because leaders strategize to avoid bloodshed.
This latter point has been underappreciated by scholars of violent contention who have, in the past twenty years, emphasized intergroup dynamics. Latin Americanists similarly take a relational approach, but one focused more on the cooperative relations that develop in the region’s clandestine gray zones—that is, those in which ordinary citizens, police, government officials, and local power brokers collude to coproduce collective violence (Auyero Reference Auyero2007; Roldán Reference Roldán2002; Romero Reference Romero, Davis and Pereira2003; Pereira Reference Pereira, Davis and Pereira2003; Arias Reference Arias2019; Trejo and Ley Reference Trejo and Ley2020). While my analysis is compatible with both approaches, I illuminate different dynamics whereby intraorganizational characteristics shape internal relations and processes that also contribute both to escalation spirals, as well as to the propensity to de-escalate and, ultimately, to the intensity of violence.
Understanding the paths to and intensity of violent contention is important because the numbers of civilians who join “irregular” armed forces is increasing globally (Davis and Pereira Reference Davis and Pereira2003). In Latin America, moreover, this radicalization is to be expected given that the region’s weak state capacity and impunity contribute to the proliferation of organized criminal groups (Romero Reference Romero, Davis and Pereira2003; Roldán Reference Roldán2002). Worse, the increasing economic and political power of organized crime expands the gray zones of corruption in which other violent nonstate actors may collude with mafias, local elected officials, and some security forces. Further, the unexpected and unstable alliances that armed citizens may forge—including with organized crime—presage further polarization (Romero Reference Romero, Davis and Pereira2003) and, ultimately, more violence (Roldán Reference Roldán2002; Pereira Reference Pereira, Davis and Pereira2003).
Acknowledgments
I thank Jeff Haydu, Zep Kalb, Warren Montag, Karen O’Neill, Bill Roy, Joel Stillerman, and LARR’s anonymous reviewers for their constructive feedback.
Open access
