Violence and Urban Order in Nairobi, Kenya and Lagos, Nigeria

Abstract

This article examines the interaction between politics and informal institutions of order in two of Africa’s most violent and crime-ridden cities, Nairobi, Kenya and Lagos, Nigeria. In both cities, governments have failed to provide basic public services and security to citizens, especially to those who reside in informal settlements or slums. A variety of informal institutions, including ethnic militia and block-level vigilante groups, fulfill security and enforcement roles in these relatively ungoverned urban spaces. This article examines the differences in the character and organization of these “specialists in violence,” and it argues that these differences are often integrally linked to the political strategies and aims of elites. The article makes two primary contributions to existing understandings of informal order in violent cities in the developing world. First, I find that organizations seemingly organically linked to local communities, such as ethnic militia, are strongly influenced by national-level political struggles. Violent organizations can gain a foothold and degree of legitimacy by appealing to traditional loyalties, including ethnicity, but organizations with these advantages are also attractive targets for cooptation by political actors. Secondly, both direct state repression and electoral use of militia lead to more predatory forms of interaction between these groups and local communities.

A short story by Uwem Akpan, published in the New Yorker, paints a vivid portrait of Lagos, Nigeria (Akpan 2010). It is a portrait the reader is expected to recognize: interminable traffic provides the backdrop and structure for the story, and ethnic militia, corpses, bribe-demanding policemen, and public defecation all make appearances. The story’s protagonist is a young priest from outside of Lagos, who runs into car trouble as he is attempting to leave the city. A Samaritan, known throughout the story only as “the Lagosian,” comes to his assistance, though the protagonist soon notes a gun in his pocket. As the two set out together to find a mechanic for the car, the priest is consumed with thoughts of how to escape from his armed kidnapper. Initially, our protagonist hopes for rescue at a roadblock manned by a police unit tasked with crime prevention, but the officers merely demand bribes and wave the car along. At another point, the priest slips away, abandoning his car to his abductor, only to run into members of the Oodua People’s Congress (OPC), one of Lagos’s most notorious militia. In the end, the two find a mechanic, and the Lagosian’s “gun” is revealed to be an oversized handkerchief. In some ways, the tale operates as a simple morality tale (i.e., trust others; charity exists even in the harshest of environments), but it also makes more subtle points about violence and order in urban environments.

The Lagosian grudgingly navigates a world that seems, to him, both orderly and routine. The rules of interaction in this world are not readily apparent to outsiders. Indeed, the outsider is misled if he relies on the most visible or the most intuitively reliable signposts. Throughout the short story, the protagonist attempts to use ethnicity to assess the intentions of others, but ethnicity is both difficult to discern and does not seem to predict behavior in the expected directions. During his encounter with the OPC, for instance, the protagonist’s appeal to ethnic solidarity is laughingly dismissed, while the Lagosian, a member of a rival ethnic group, is quickly able to negotiate the priest’s release. In this article, I will attempt to show that ethnicity—too often seen as a master key to understanding African politics—has only weak effects on the character of urban-based militia. Though these militia respond to and reflect ethnic politics at the national level, the ethnic character of militia does not necessarily yield more rooted or less predatory relations with local communities. Instead, this article stresses that local armed actors in urban Africa are more strongly shaped by political mobilization and state policy than they are by other factors.

This article examines two major questions about the sources and functioning of informal order in the fast-changing cities of the developing world. First, do vigilantes and other local armed actors simply emerge to fill the vacuum left by an absent state, as Akpan’s story seems to suggest? Or do these groups have a stronger connection to national politics and state strategy than is typically presumed? Secondly, where local armed actors draw upon communal, ethnic, or other “pre-urban” identities, are they more likely to play a benign or “shadow state” role vis-à-vis local communities? To put this differently, can co-ethnicity—or other forms of traditional legitimacy—be counted upon to restrain those actors that emerge in contexts of low state capacity and weak rule of law?

Akpan’s story is set in 1999. In the 10 years since 2000, Lagos has undergone remarkable change, which has been driven by large-scale shifts in the policies of the Lagos State Government, in public attitudes toward the state, and in broader economic conditions in the city. The city is less crime-ridden than it was in the 1990s, and state presence is far more visible, even in the most marginal informal settlements. Yet despite some expansion of state authority in Lagos, urban governance remains uncertain and volatile. An estimated 10,000 migrants arrive in Lagos each week. Their first homes tend to be in the city’s burgeoning informal settlements or slums, where the state has a limited ability to deliver security or other state services. Both new migrants and established residents of these informal settlements are embedded in economies and reliant on institutions that exist apart from the formal state realm. Nor can they safely or reliably depend on social ties that are derived from rural home areas. Both migrants and established urban residents may occasionally make use of “economies of affection,” but they navigate a world largely peopled by strangers and institutions over which traditional loyalties hold limited sway. Lagos resembles other cities in the Global South in so far as it is characterized by a complex interplay between local communities, state and local armed actors, and the global economy. In his introduction to this volume, Moncada (2013) suggests that both violence and order result from the shifting relations between these “multiple territorial and institutional scales.” This article takes a similar tack, arguing that changes in interactions between communities, militia, and political leaders can have significant effects on levels of violence and the character of violent organizations. The character and activities of meso-level organizations, such as vigilante groups and ethnic militia, provide an ideal lens through which to examine the interaction between local communities and national or state politics.

In terms of argument, the paper first suggests that constraints on local armed actors’ behavior, such as shared ethnic identity, popular legitimacy, or other forms of embedding in local culture, are easily weakened by engagement with political elites. Politicians often make instrumental use of specialists in violence, to advance short-run electoral interests, but they do so at the cost of undermining formal state capacity. Political manipulation increases the freedom of operation for these groups, and the resulting organizational autonomy weakens ties between armed actors and local communities. As is empirically borne out in both of the cases to differing extents, the combination of physical freedom of operation and weak local embedding produces more predatory organizations.

Secondly, the paper demonstrates that shared ethnicity only weakly structures the behavior of organizations that ostensibly exist to provide ethnic security. Militia may be initially motivated by notions of ethnic defense, but they evolve over time and may be repurposed to serve either criminal or other political ends. This should not suggest that ethnicity and other forms of community identity are epiphenomenal; indeed, both vigilantes and ethnic militia are strengthened by their ability to capitalize on these identities and recruit from their identity communities. But identity rarely has the strongly determinative and stable effects that the political science literature on ethnic security dilemmas often presumes (e.g., Posen 1993). In other words, ethnic militia and vigilante groups are not reliably the creatures of the populations they ostensibly protect. Similar to the criminal actors that Arias (2013) analyzes in his contribution to this volume, ethnic militia and other vigilante groups have corporate interests that trump stated community loyalties. Their interactions with politicians and police complicate the relationships these organizations forge with local communities. An initial reliance on ethnic mobilization—such as is often entailed in the political use of militia to advance or defend group interests—does not necessarily imply protection of or accountability to members of that ethnic group.

Thirdly, apart from electoral politics, organizations are also shaped by collaborative or repressive interaction with formal government institutions. Previously, I stressed how individual politicians’ instrumental use of vigilantes and militia during electoral periods creates greater space and scope for action on the part of specialists in violence. But, in both cases, political mobilization of violent organizations was followed by state efforts to rein in the very armed actors that politicians had helped to forge. Both the Nigerian and the Kenyan governments officially banned ethnic militia and other vigilante groups in 2000 and 2002, respectively, but subsequent state interactions with these groups have been pursued differently in the two cities under examination. In Nairobi, the state has undertaken violent repression of Mungiki, one of the country’s most notorious gangs, and this policy has increased slum-dwellers’ distrust of state institutions. In Lagos, on the other hand, the state government has entered into more accommodating, more fluid relationships with both organized ethnic militia and less organized vigilante groups. Based on the comparison of these two cases, the paper concludes that violent organizations—once constituted—are difficult to displace. Direct repression tends to result in changes in organizational structure that still further attenuate the organizations’ ties to local communities, but more collaborative relations with government institutions do little to challenge these organizations’ autonomy or presence.

Urban order is, therefore, shaped by three factors: political mobilization, which generates stronger and more autonomous organizations; ethnicity or other community identities, which can conceivably serve as a weak constraint on predatory behavior; and the strategies of control chosen by formal government institutions. But what do we mean by urban order, and who are its main providers in Nairobi and Lagos? By order, we might mean a system of navigable rules and structures, which may be less visible to outsiders (Akpan’s priest, for instance) than to those who reside inside the system. This definition is broad enough to encompass a variety of social rules and cultural norms, but the paper focuses its attention on two kinds of local armed actors, both of which specialize in the provision of order by way of coercion. Put simply, these groups specialize in the use of violence. First of all, vigilantes refer to neighborhood-level groups that provide policing and, occasionally, other public goods to communities. This is an umbrella term that represents a wide range of actors, including community-organized policing forces and the block-level gangs known in Nigeria as “area boys.” As others have pointed out, vigilante justice is a global urban phenomenon, and vigilante groups can be seen as local enterprises that aim to create order in a context of state failure and increasing crime (e.g., Pratten and Sen 2008). In Nigeria, area boys serve this local vigilante security role, but they are also occasionally implicated in political action and coercion for political purposes (Momoh 2003; Pratten 2009). On a day-to-day basis, area boys are active in petty larceny and criminal rackets. The second type of violent organization addressed in this article is the ethnic militia, which plays an important role in both local and national politics in both Kenya and Nigeria. Group insecurity is often used as a justification for the existence of these militia and they are often much more organized and politically consequential than their lower-level security competitors.¹

In terms of the paper’s organization, the next section provides an introduction to the national political dynamics in each of the two case studies. The paper then turns to an analysis of the local armed actors that serve as both the primary agents of order and the primary perpetrators of violence in slum areas in the two cities. A larger literature on criminal and violent organizations has pointed out that the operations of these groups are Janus-faced (Arias 2006; Gambetta 1996; Varese 2001; Volkov 2002). On one hand, these organizations often provide public goods to communities, including contract enforcement, protection, and dispute resolution. In ethnicized settings, gangs may provide security against rival groups; in other places, effective territorial gangs may have a dampening effect on petty crime. On the other hand, gangs and violent organizations are also predatory and extortionary business interests, and they often create the demand for security that they then supply. Electoral politics—and other governmental action—often shapes the balance between these roles. I first address the creation of conflictive relations between the government and militia in Nairobi, and then I turn to how dynamics of order and violence differ in Lagos. Following these two empirical sections, I present some evidence on how popular attitudes in slum areas in both cities differ from the attitudes of those who live elsewhere. This data cannot establish a direct link between militia or vigilante presence and attitudinal change, but it is suggestive of a distinctive orientation toward safety and violence in areas where informal institutions are more powerful in shaping urban order. In these areas, perceptions of safety are higher than elsewhere in the same city, but there is also a much greater tolerance expressed toward the political use of violence. This seems compatible with the Janus-faced approach mentioned above. Finally, the conclusions point out ways in which this analysis should shape our understanding of the legitimacy claims made by these actors and the policies of control pursued by politicians and governments.

An Introduction to the Two Cases

There are notable similarities in the national politics of both Kenya and Nigeria. Since independence, politics in both countries has been organized around ethnic or ethno-regional cleavages. In Kenya, throughout much of the post-independence period, power and authority over state spending have rested with ethnic groups that had control over the executive, namely, the Kikuyu and the Kalenjin. The ethnicization of national politics led to differential regional development, as well as resentment in western Kenya, where ethnic groups had little access to power or patronage resources. Since the transition to multiparty politics in 1992, ethnic competition has been largely grafted onto electoral politics, and regional voting tends to follow ethnic blocs, particularly for the presidential vote. In Nigeria, national politics has similarly been organized around large regional blocks, even after the return to multiparty electoral politics in the late 1990s. Despite complicated constitutional provisions that create incentives for coalition-building, elections remain characterized by sharp divisions between the mainly Hausa–Fulani north, the predominantly Yoruba southwest, and the more fragmented ethnic groups of the Niger Delta and southeast.

These ethnic and ethno-regional cleavages have not merely been means of mobilizing constituencies or organizing vote blocks. Since the return to multiparty electoral politics, the interaction of grievances and exclusionary political mobilization has resulted in several episodes of violence in both countries. In Nigeria, the most significant of these have been the ostensibly religious riots that have plagued several towns in the country’s “Middle Belt” or the intermediary geographic zone between the predominantly Muslim north and the predominantly Christian south. Some of these riots were directly related to the election cycle, as in the case of riots in the north following the April 2011 presidential elections that left over 800 dead (International Crisis Group 2011). Other riots are less clearly related to electoral politics, but they often result in an intensification of ethnic and religious polarization nationwide. For instance, in 2002, riots surrounding the Miss World contest, scheduled to take place in Abuja, degenerated into Christian–Muslim conflict in Kaduna that left over 200 dead. In cities in southern Nigeria, notably in Aba and Port Harcourt, retaliatory riots resulted in deaths and mass displacement of Hausa-speaking northerners. In Kenya, ethnic violence has had a much more transparent connection to electoral politics, and clashes over land and resources have been strategically used by politicians to redraw electoral geography and win elections. Ethnic clashes from 1991 to 1997 took over 1,500 lives and left 300,000 internally displaced persons, many of whom have yet to be resettled (Republic of Kenya 1999). In December 2007, contested elections led to large-scale ethnic violence across the country, leaving up to 1,500 dead in just over 2 months.

Nairobi and Lagos have also been affected by both election-related violence and ethnic conflict. In Nairobi, several slum areas, particularly those in the eastern half of the city, were the sites of violence, forced eviction, and alleged “cleansing” during the 2007 post-election violence. In Kibera, Nairobi’s largest slum, there have been occasional incidents of ethnic violence outside election periods, and Luo and Luhya residents of the slum often center their grievances on the demands of allegedly Kikuyu “landlords.” In Lagos, ethnic violence occurs and has escalated following the first transition to multiparty rule in 1999. Many have pointed to the formation of the OPC, mentioned previously, as a cause of increased violence in Lagos against both Hausas and other ethnic minorities from the Niger Delta (Ikelegbe 2001; Ukiwo 2003; Guichaoua 2009), and the OPC has been implicated in many of the worst cases of targeted attacks against ethnic minority traders in Lagos.² In addition to ethnic conflict, Lagos has also been a site of election-related violence and intimidation, though this rarely reaches the level of severity experienced elsewhere in the country.

In addition to politically motivated violence and ethnic clashes, residents of Nairobi and Lagos face everyday insecurity due to very high rates of violent crime in the two cities. Violence has long been endemic in both Nairobi and Lagos, but crime escalated in the 1990s and 2000s. UN-HABITAT’s exhaustive survey of crime in Nairobi, published in 2002, found that 37 % of Nairobi residents had been victims of robbery in the preceding year and a further 17 % had been victims of physical assault (UN-HABITAT 2002). In the years since the publication of this report, there has not been further research comparable to the report’s survey of over 10,000 Nairobi residents. Recent reports suggest that Nairobi’s crime rate has dropped, though the victimization figures do not seem substantially lower than those reported by UN-HABITAT in 2002 (e.g., World Bank 2011, p. 60). Over the past decade, there has been a marked decrease in violent crime in the central business district (CBD) in Nairobi, but this has not been matched by comparable improvements in security in low-income neighborhoods. Instead, many Kenyans believe that the crackdown on criminal activity in the CBD has simply pushed criminal activity into low-income and slum neighborhoods in Nairobi. In a similar survey-based study of criminal victimization in Lagos, Alemika and Chukwuma (2005) found that 25 % of respondents had been victims of robbery in the past 5 years, while 12 % had been victims of physical assault. Recent Afrobarometer data suggest slightly higher rates of victimization in Lagos, as 40 % reported an instance of robbery or physical assault in the previous year.³

Empirically, this paper drills down from both the national and city levels to focus on organizations that tend to be more present and more powerful within low-income neighborhoods. In both Lagos and Nairobi, as elsewhere in the developing world, slums are the terrain in which violent non-state organizations operate with the greatest autonomy and, in some cases, the greatest levels of local support.⁴ As we shall see, the public attitudes of slum-dwellers are distinct in important ways from those who live elsewhere. In both cities, the rapid expansion of the urban population means that many residential areas are unplanned and lack access to municipal services. The quality of these residential developments varies widely, and some informal settlements contain a wider range of income levels or better living and environmental conditions than one might commonly expect to find in areas that are characterized as slums. Slum communities in both Nairobi and Lagos differ substantially from one another in terms of density, levels of insecurity, and ethnic composition. In Nairobi, these communities are fairly well demarcated, though they may contain sections that have been extended access to infrastructure or have regularized property rights. In Lagos, the boundaries between slum and non-slum areas are less sharply defined, and there is greater diversity of conditions within local government areas that are commonly viewed as slum areas.⁵ In the following section, I will focus on the shifting landscape of informal order and attempts at state control in Nairobi, while the subsequent section then shifts to the case of Lagos.

Political Violence, State Repression, and Crime in Nairobi

Since the 1990s, Kenya has experienced substantial growth in the number of non-state actors involved in the provision of local security and, in some cases, in the provision of political violence. The primary reason for the growth and increasing scope of activity for violence specialists has been electoral politics. In 1992, at the point when multiparty electoral competition was reinstituted in Kenya, the ruling Kenya African National Union (KANU) had a thin base of support, and opposition was split on ethnic lines (Throup and Hornsby 1998; also, LeBas 2011). KANU officials played key roles in the incitement and organization of ethnic land clashes in politically divided areas of the country.⁶ These land clashes were transparent attempts to redraw electoral geography, especially in the portions of Rift Valley province where Kikuyu residents were perceived both as nonindigenous and as likely opposition supporters. By forcing these populations to become internally displaced persons, KANU transformed swing constituencies into “safe” electoral zones. Urban areas were initially less implicated in these formal campaigns of electoral violence, but, by 1997, the ruling party youth brigade was deployed in Nairobi, where it violently disrupted rallies for constitutional reform and attacked opposition leaders (Kenya Human Rights Commission 1998, pp. 35–36). Nor did urban areas escape from the broader consequences of the ethnic polarization that accompanied KANU’s violence campaigns. Over the course of the 1990s, ethnic militia proliferated in Nairobi and other urban centers (Katumanga 2005; Mwagiru et al. 2002). Both formal ethnic militia and less organized gangs were intended to serve as vigilantes and provide protection for increasingly insecure ethnic communities. Once built, these organizations became attractive resources for aspirants to both local and national office. Residents of Nairobi and other urban slums reported that both ruling party and opposition candidates hired “security” for both intra-party primaries and general elections by the late 1990s.

The instrumental use of state-sponsored violence allowed KANU to hold onto power with only small proportions of the popular vote, but it also had a corrosive effect on the rule of law. As political elites invested in their own private networks of violence specialists, they systematically disinvested in public institutions of rule, a process Katumanga (2005) refers to as the simultaneous “privatization of public violence” and “political appropriation of private violence.” There were also more complicated mechanisms linking the decline of public institutions and the rise of local armed actors. In his study of Rift Valley, where violent cross-border cattle raids became more common in the 1990s, Mkutu (2008, p. 38) suggests that community protection was delegated to paramilitary vigilante groups in part due to the public’s association of local governments with corruption and inefficacy. The incorporation of traditional warriors into the region’s police reserves may have initially boosted the security of some groups in the province, but it also rendered subsequent government attempts to disarm raiding communities more difficult. In urban areas, public institutions were being hollowed out at the same time that demand for their services increased. The Nairobi City Council and City Commission become more deeply enmeshed in corruption and economic mismanagement in the 1980s and 1990s, and there was little enforcement of city ordinances governing urban development (Mueller 2008, p. 192). A rapid expansion in the size and number of informal settlements within the city limits followed. These communities had minimal access to state services and police protection, creating demand for the private security services that gangs and militia ostensibly provided. In cities, new urban migrants—many of whom had been displaced during ethnic clashes—provided both a client base and a recruitment pool for emerging ethnic militia and less organized urban gangs.

In Nairobi and, to a lesser extent, other Kenyan cities, gangs began to proliferate in the late 1990s (e.g., Jeshi la Mzee, the Taliban, the Bagdad Boys, 42 Brothers, etc.).⁷ Many of these groupings had loose associations with particular parties or politicians, but it was increasingly difficult for politicians to retain control over their networks of violence specialists, who during non-electoral periods had found work within organized crime (Anderson 2002; Kagwanja 2003; Mueller 2008). By the 2002 elections, violence and intimidation were clearly no longer the preserves of the ruling party. Individual politicians—belonging to both KANU and the opposition—hired or organized their own vigilante groups to attack rival candidates, intimidate prospective voters, and forcibly disenfranchise ethnic groups that were seen as likely opponents (e.g., Mutahi 2005, pp. 78–86). An increasingly fragmented party system generated ever-larger numbers of political candidates who hired security for election campaigns (LeBas 2011, pp. 239–243). The relationships between violence specialists and politicians also became marked by a new opportunism and transparency of aims. By the late 2000s, when I was conducting fieldwork, informants told of violent gangs working for two rival politicians and of brokers in Nairobi who practiced price discrimination when arranging election “security” for politicians.⁸

The sophisticated economy of political violence no longer produced clear-cut electoral dividends, but it did produce increasingly well-organized and well-resourced criminal organizations. Mungiki, now Kenya’s most notorious criminal gang, is a case in point. The organization has its roots in Central Province, where it initially gathered adherents by offering a return to traditional Kikuyu religious practices. It retained this character for some populations, particularly in predominantly Kikuyu Central Province, but it also evolved into an armed ethnic militia and a significant independent political actor. During the land clashes of the 1990s, Mungiki served as a Kikuyu self-defense force in rural areas in and outside of Central Province. By the early 2000s, the movement had an established footing in Nairobi, was involved in both extortion rackets and community policing, and claimed a membership of anywhere from 3 to 4.5 million members (Katumanga 2005, p. 513). Changes in the organization’s base of activities eroded Mungiki’s commitment to ethnic security and its early pro-opposition political loyalties.

In a surprise move, Mungiki leadership announced that the movement would back the ruling party in the 2002 elections and even field electoral candidates on the KANU ticket (Kagwanja 2003, p. 48). David Anderson argues that Mungiki was rendered vulnerable to cooptation by its migration from its ethnic heartland to the cities, where it became involved in organized crime and gradually evolved into yet another urban vigilante group for hire (Anderson 2002). Following the 2002 elections, political control over Mungiki collapsed entirely: the organization became immersed in bloody battles over matatu (commuter bus) routes, arms caching, and homicidal attacks on “defectors” (Kagwanja 2006). Contact between the militia and politicians became still more complicated and commercialized (Kanneworff 2008). For urban residents, the continuing battles between various Mungiki factions had disastrous effects on security in slum and low-income areas. Mungiki may have been rooted in traditional loyalties to a much greater extent than other violent organizations in Nairobi, but this has not made it more accountable to local communities over time. There is little evidence that it served as an effective security guarantor for Kikuyus or the residents of the slums where it is most present.⁹

State action against the organization has reinforced Mungiki’s disconnection from local communities and its predatory character. In an attempt to regain lost control over urban militia, the Kenyan state outlawed all vigilante groups and ethnic militia in March 2002, but this had little effect on their operations. In 2007, the Kenyan police formed a special unit to take stronger action against Mungiki. Between 2007 and 2008, the unit essentially operated as a death squad, targeting Kikuyu youths resident in slum neighborhoods, regardless of whether there was evidence of their involvement in Mungiki or not. The unit is alleged to be responsible for well over 1,000 extrajudicial killings of suspected Mungiki and other urban “undesirables.”¹⁰ Because of this violent state repression of Mungiki, very little is known about the gang’s organizational structure. Scholars have typically focused on the rural religious aspects of Mungiki, observed changes in patterns of violence and state repression, and the actions and political allegiances of leaders such as Maina Njenga, who long served as the public face of Mungiki. The lack of attention to the criminal structures of Mungiki in Nairobi—which are the elements of the movement that come closest to “non-state governance structures”—is driven in large part by the inaccessibility of sources within these structures. Participants in focus groups and other slum residents consistently reported that crime-linked members of Mungiki are not free to leave the organization, and several suggested that Mungiki has killed those who leave the organization or refuse work.

What little information I have suggests that Mungiki is a far more hierarchical and disciplined organization than the militia and vigilante groups found in Lagos. At the national level, as suggested previously, there has been splintering of the organization and increased factional maneuvering. To a large extent, academic and media commentary has focused on this disorganization at the top. At the level of the community, operations seem to have remained hierarchical and characterized by an impressive array of firewalls.¹¹ State repression in 2007–2008 required the rotation of foot soldiers between Kayole, Dandora, and other Mathare Valley slums, and many of those who were moved out of an area were not given stable employment elsewhere.¹² In particular areas or stations, Mungiki foot soldiers are charged with collecting fees from matatu (bus) operators and other “protected” businesses; they work in small units (at a bus stage, for instance) and report to a lieutenant or munene. The munene serves as the foot soldiers’ only contact point into the organization, and he jealously preserves the isolation of each unit. He lives elsewhere, provides no contact beyond a cellphone, and does not make separate small local units aware of each other’s existence. When the need for discipline arises (e.g., a matatu driver refuses to pay protection), the munene arranges for a special disciplinary unit to come from outside the slum.¹³ My informant suggested that these were distinct from foot soldiers: they were typically from “up-country,” and their lack of education was prized, as Mungiki viewed them as “a man among men.”

In terms of relations with the local community, this organizational structure has implications for residents of the slums where Mungiki is most operational. Unlike vigilante groups and area boys in Lagos, Mungiki attempt to limit their visibility to residents of the areas where they operate. My informant reported that he actively concealed his association with Mungiki from neighbors, friends, and others with whom he did not have direct Mungiki business. Rasmussen reports similar attempts at concealment among Mungiki in Dagoretti, another Nairobi neighborhood. During ethnographic fieldwork in 2009, he found that Mungiki in the area had cut off their distinctive dreadlocks and refrained from taking snuff in public, another distinctive Mungiki practice (Rasmussen 2010, pp. 309–310). Mungiki’s communal and religious origins may still hold sway over some members within the group, but concealment and the actor’s increasingly cellular structure suggests an effective severing of the organization from a broader Kikuyu constituency. In informal conversations and focus groups, slum residents spoke of the “shadowy” nature of Mungiki and other urban gangs. Informants repeatedly suggested that “you can’t know when you are talking to Mungiki” or “who is who.” As we will see, the anonymity of Mungiki members sharply contrasts with the visibility of members of ethnic militia and other vigilantes in Lagos. This ambiguity doese not seem a recipe for the stable urban order which some suggest that territorial gangs can provide. In an account of the large-scale violence that followed the contested 2007 elections, Michelle Osborn (2008) details how this uncertainty gave rise to rumors of Mungiki’s arrival in Kibera, rumors that may have sparked greater interethnic conflict within the slum.¹⁴

In earlier periods, Mungiki and other urban gangs were described as playing “shadow state” roles, either by providing protection from rival ethnic groups, enforcing a zero-tolerance regime toward petty criminals, or even by providing funds for public goods (e.g., Kagwanja 2003). In the 2007–2010 period during which I was conducting fieldwork in Nairobi slums, I saw little evidence of these roles. Rasmussen (2012) stresses the security and protection work that Mungiki purportedly provides in areas that have little access to state services or effective policing, but this seems an overly sanguine view of Mungiki’s commercial operations, which remain marked by both violence and unpredictability. During the post-election violence of 2007–2008, Mungiki was active in Nairobi as in other areas of the country, and it seems to have been implicated in targeted attacks against other ethnic groups in slum neighborhoods in eastern Nairobi (Waki Report 2008). The Waki Report on post-election violence also, however, included reports that Mungiki used the post-election violence period to opportunistically evict squatters from rental properties and also took pay from local politicians for intimidation and other political tasks (pp. 198–199). This evidence complicates a narrative of Mungiki as an ethnic self-defense force. As for the utility of the state’s chosen policy toward Mungiki and other ethnic militia, it seems that brutal repression has been largely ineffectual in reducing the organization’s size or its involvement in key commercial sectors, such as urban minibus transport. Even though Mungiki is geographically concentrated in a small number of Nairobi neighborhoods, the state’s repressive operations of 2007–2009 do not seem to have reduced Mungiki’s organizational presence in these areas.¹⁵

Crime and Violence in Lagos

As in Nairobi, the creation of ethnic militia in Lagos is interlinked with the return to multiparty politics. In 1993, M.K. Abiola was elected president in the first multiparty presidential elections since a coup had overthrown Nigeria’s Second Republic in 1984. Unhappy with the results, the northern-dominated military government quickly nullified the results, leading to political crisis and the seizure of power by General Sani Abacha, who ushered in a brief but extraordinarily predatory period of military rule. Others have argued that the brutal character of Abacha’s regime led ordinary Nigerians to look for security and political representation via the formation of ethnic militia. Shortly after the annulment of Abiola’s victory, the OPC was formed to represent Yoruba political interests. This was intended to be an explicitly political organization, but it also had a substantial grassroots presence. By 1999, the OPC had nearly 3,000 local branches across southeastern Nigeria, and it claimed up to 3 million members (Akinyele 2001, p. 626). The growth of the OPC coincided with a period of rising crime in Lagos, and the OPC soon embraced a dual role as political action group and security organization. Large numbers of less formal vigilante groups and community organizations were formed in the 1990s to deal with the problem of rising urban crime. In 1999, soon after the return to multiparty rule, the OPC formally took on vigilante or security functions as well. In some cases, the OPC supplemented state institutions, chasing criminals into slum areas where the police dared not venture and even delivering them to police for trial (Guichaoua 2009, p. 528). Even more intriguing, Guichaoua (2009, pp. 528–529) suggests that the organization became a full-fledged “extralegal governance agency” by the early 2000s: for a fee, OPC officials would arbitrate legal disputes, collect debts, represent workers threatened with dismissal, and provide other services that the Nigerian state had long ceased providing.

Despite a formal legal ban in 2000, the OPC remains the largest and most visible of the ethnic militias in operation in Lagos. Other ethnic militias have emerged in Lagos, partly to protect non-Yoruba communities from the extraction of protection rents by the OPC. In other parts of Nigeria, militia like the Bakassi Boys and Egbesu Boys operate along similar lines (e.g., Harnischfeger 2003; Meagher 2007; Smith 2004). Militia groups often see themselves as effective state substitutes, and it seems that some members may even support the formal integration of vigilante organizations into the structures of the state. For instance, Adebanwi quotes an OPC member:

It is possible for the government to integrate us into the law-enforcement system. All they need to do is to number the guns given to us and hand them over collectively to a sectional commander. Once we return from vigilante activities, we hand them back to our leader (Adebanwi 2005, p. 355).

This sentiment seems to have been echoed by the actions of several politicians. In 2000, the Imo State House of Assembly passed a resolution requesting the assistance of the Bakassi Boys militia to apprehend armed criminals, and governors and other political figures in Lagos have periodically made supportive statements regarding the crime-fighting actions of the OPC and other militia. In Kano state in northern Nigeria, the state government directly funds the substantial operating costs of hisbah, the non-state vigilante groups tasked with the implementation of Sharia (Olaniyi 2011).

Collectively, these militia are precariously perched between local communities and the state: they seek collaboration with or even funding from the state, but they must also prevent a perception of cooptation that would erode their standing among local communities. This is a substantial difference between ethnic militia in Lagos and the quasi-ethnic criminal gangs that operate in Nairobi. Guichaoua has argued that the OPC has sustained itself to the present, despite a formal legal ban, because it is supported by an underlying “moral economy” that connects the militia and the Yoruba ethnic constituency they ostensibly serve (Guichaoua 2010). In contrast to Mungiki and other gangs in Nairobi, militia members in Lagos are known in local communities, and they are connected to preexisting social networks. In a survey of 167 OPC members, 70 % reported that their membership in the organization was known to their neighbors and nearly 45 % believed that joining the OPC had improved the social standing and respect they were accorded in their communities (Guichaoua 2010, p. 1662). Other scholars note a similar dynamic at work with ethnic militia in other areas of Nigeria. As with the OPC, which recruited members through preexisting social networks, the origins of the Bakassi Boys of southeastern Nigeria lie in dense urban networks. Faced with rising crime in Aba and Onitsha, the major towns of the southeast, merchants’ associations sponsored the formation and initially paid salaries of the Bakassi Boys. Harnischfeger (2003, p. 31) argues that the militia was more restrained, fair, and accountable than state security forces “because they are more under the control of the local population.” Kate Meagher (2007) argues that the Bakassi Boys were effective and earned the support of their communities because the organization was perceived as incorruptible and walled off from the strategies and campaigns of political elites. Once the group began to accept the patronage of powerful state governors, the group’s narrow commitment to community security was undermined. Unjustified killings became more common; the group became implicated in ethnic violence and the anti-Hausa riots of 2000; and “supernatural claims to legitimacy were used to cover for an increasingly compromised commitment to the public good” (Meagher 2007, p. 99; also, Harnischfeger 2003, pp. 41–44; Smith 2007, chapter 6).

Relations between the state, ethnic militia, and individual politicians are complicated. For many politicians, particularly for powerful regional governors, militias represent an attractive political resource, and there have been consistent reports since 1999 of the use of the OPC and other militia to intimidate voters, organize poll fraud, and threaten rival candidates. In Lagos, the OPC was involved in clashes between Yoruba and Ijaw (Delta) dockworkers at Apapa in 1999, and it also engineered an attempted takeover of the Mile 12 market from Hausa traders in 1999 (Akinyele 2001, pp. 627–628; Human Rights Watch 2003). These incidents resulted in several deaths, possibly as high as 115 in the Mile 12 market dispute. The OPC’s turn to attacks on other communities provoked the formation and arming of rival ethnic militia. Militias from other areas of the country, notably the Niger Delta region, were present in Lagos by the early 2000s. Clashes between ethnic militia led to increasing insecurity in slum communities during this period, as did the state’s pursuit of aggressive policies of slum clearance, which began during this period. Though militias have greater levels of popular support than do their counterparts in Nairobi, they have also played a part in fostering ethnic antagonism and segregation in Lagos and other Nigerian cities. On occasion, they have played central roles in riots and large-scale ethnic confrontations.

Less organized vigilante groups have a comparatively more mixed effect on urban order. Despite the extraordinarily negative perceptions of outsiders toward the violent and predatory youths known as “area boys” in Lagos, demographically similar youths also compose vigilante groups that serve as community police in many areas (Ismail 2009). In a 2005 survey, 81 % of Lagos residents reported that a vigilante group provided security in their area and 77 % reported that residents paid these groups for their services (Alemika and Chukwuma 2005, p. 34). Area youths seem to provide communities with other goods as well. During fieldwork in 2010, I was struck by different markers and house numbers painted on residences in many slum areas, especially in Agege and Mushin. When asked about who had numbered these houses, residents noted that the project was undertaken by youths or by “our area youth association.” These house number markers—which can be representations of peacocks and bulls, as well as more plain markers—can be used to map the effective territorial bounds of various groups of vigilantes or area youths. Momoh (2000) points out that this community embedding of area boys was still stronger prior to the implementation of economic structural adjustment in the 1980s. In this earlier period, area boys were initially associated with neighborhood Yoruba festivals, and they provided security, prevented incursions from potentially criminal youths from other areas, and organized sports competitions (Momoh 2000, pp. 187–189).

Though area boys still play some shadow state roles in their local communities, they also have predatory qualities, and they are vulnerable to cooptation by politicians. Momoh (2003), relying on numerous interviews with area boys in Isale Eko in Lagos, found that politicians regularly gave these unorganized groups money and relied on them to mount attacks on particular economic or political targets. Vigilante groups seem to have little direct role in ethnic conflict, but they are implicated in criminal economies in slum areas and in Lagos more broadly. Several individuals reported that area boys were nearly always present in commercial areas and bus ranks, where they extracted “small money” or around 50 naira in “tax” from residents. As in Nairobi, these groups are also involved in larger bus route extortion and business protection rackets.

Though state capacity is expanding, the Lagos State Government has recognized that it currently has limited capacity to extend direct control over slum communities or provide the public services that residents of these areas desire. In contrast to both police forces in Nairobi and earlier efforts at control in Lagos, the administration of current Governor Babatunde Fashola has not attempted to directly repress or liquidate the informal networks that characterize urban Lagos. Many of the signposts of localized orders in the slums of the 1990s—vigilante groups (“our area youth association”), even organized militia such as the OPC—remain intact. Ethnically targeted roundups of suspected OPC members occurred in the early 2000s, but they were never as extensive as state repression of Mungiki has been in Nairobi. State attempts to take action against criminal actions by “area boys” tended toward the creation of vocational training and educational courses for these youths.

More recently, the state has attempted to utilize existing informal networks to expand the state’s ability to levy taxes in low-income areas. There is broad agreement that local government has relied on informal institutions, including criminal networks, to raise revenue for some time. Some slum residents also suggest that the Lagos State Government’s expansion of its state tax collection has come about through the cooptation of area boys and agberos, the quasi-criminal figures who maintain order and charge protection rents at bus stands across Lagos. By giving these individuals uniforms and an official state role, local government councils have experimented with a different means of constructing localized order in violent cities. The resulting form of tax collection resembles the tax farming characteristic of Rome and early modern Europe, in that individuals submit only a set amount to the state and retain monies collected above that amount. But the state is also able to rely on the local knowledge and resources possessed by these agents, resulting, in all likelihood, in higher tax receipts than they would otherwise be able to collect. At an October 2012 workshop, tax officials from state governments outside of Nairobi also mentioned the use of community associations and other organizations to provide information and assist in tax collection.¹⁶ Some participants emphasized the cooperative or facilitative nature of these linkages; others suggested that relations with informal organizations, including vigilante groups, were closer to those reported in Lagos. Overall, this is a remarkably different pattern of state–society interaction than the one we see in Kenya, where the Kenyan state has more typically resorted to direct repression or, at best, a benign neglect of local armed actors.

Popular Perceptions and Informal Order

A key question that emerges from the comparative analysis developed previously is whether the presence of local armed actors actually provides benefits for the communities they inhabit. As we know, there are often real or claimed social ties between violent organizations and ordinary people, of which shared ethnicity is only the most visible. And even if these organizations do have predatory attributes, they may still retain the Janus-faced qualities mentioned previously. Put simply, do violent organizations produce order? Does the presence of vigilantes or other non-state actors make individuals feel safer? There is no direct way to measure the presence or character of vigilantes and militia using survey data, as individuals are unlikely to share information on sensitive topics. But comparisons of my own surveys of slum-dwellers with more representative national and city samples yield interesting patterns that will contextualize the preceding discussion of armed actors.

Data for this section are drawn from two surveys that I conducted in slum communities in Nairobi and Lagos, which I then compare with representative national and city samples collected by the Afrobarometer, the major public opinion instrument on the continent. The Nairobi survey that I conducted sampled 600 respondents in six slums in July 2009.¹⁷ In July 2010, as part of a larger project, I conducted a pilot survey of 654 respondents in 7 neighborhoods in Lagos, 5 of which are urban slum communities. For the purposes of this paper, I drop respondents in two neighborhoods in order to have a set of neighborhoods that are comparable to those sampled in Nairobi.¹⁸ And so as to have national-level and city-level means with which to compare the attitudes of slum-dwellers, I rely on Afrobarometer data collected during roughly the same time period. In Kenya, I use the October 2008 Afrobarometer (round 4) as a frame of comparison: like my own survey, this was carried out well after the wave of post-election violence in January and February 2008. In Nigeria, I compare my Lagos slum data with round 4 of the Afrobarometer, which was conducted in May of 2008.¹⁹ There were no federal elections that intervened between these two surveys, but Lagos State did hold local council elections in October 2008. These were characterized by irregularities.²⁰ There are much weaker differences between my sample and the Afrobarometer Lagos sample than is the case in Nairobi. I suspect this is due to differences in urban geography: because neighborhoods are more economically diverse in Lagos and because slums are rarely segregated to the degree they are in Nairobi, random or nonpurposive sampling will yield a more economically diverse sample.²¹

Surprisingly, analysis of the data shows slum communities emerge as safer and as characterized by higher levels of perceived safety than the larger urban communities that surround them. In Nairobi, only 17.7 % of slum residents (July 2009 survey) reported that they or a member of their households had suffered a crime of any kind in the previous year. In contrast, 34 % of the Nairobi Afrobarometer sample had experienced robbery or violent assault. In Nairobi, there is no significant difference in crime rates across slum neighborhoods: neighborhoods such as Kayole and Dandora, for instance, with which both Mungiki and violent state repression are most associated, do not have significantly higher or lower crime rates than other neighborhoods that are less characterized by militia presence. In terms of perceptions of safety, there are differences across neighborhoods, though these do not seem to be associated with the degree of militia activity in different neighborhoods. Residents of the Dandora and Kayole slums in Nairobi do express significantly lower levels of safety when they go to markets, but these perceptions of insecurity do not map onto other activities (e.g., unsafe when sleeping, attending political rallies, etc.). Slum populations in Nairobi also expressed high levels of security: 52 % reported that they felt “very safe” going to the nearest market, and 73 % reported that they felt very or somewhat safe sleeping in their homes at night.²²

The patterns were similar in Lagos. Only 15.6 % of respondents in the Lagos slum survey (July 2010 survey) reported that they had experienced any kind of crime in the past year, versus 39.8 % of Lagos residents surveyed by the Afrobarometer. Though there are higher overall levels of insecurity in Lagos than in Nairobi, a majority of slum residents in Lagos still report that they feel very or somewhat safe meeting strangers, and 67 % report feeling safe when attending meeting or community events. This tracks well with my limited qualitative work in Lagos slums, which has suggested vibrant associational and party branch activity.²³ Residents of the surveyed slums with the greatest reputations for militia activity and criminal violence (e.g., Ajegunle, Mushin) express neither higher nor lower levels of distrust and insecurity than residents of neighborhoods less characterized by violence. This data does not offer firm support for the crime-dampening and insecurity-dampening effects of violent organizations, but the distinct response patterns of slum-dwellers regarding crime and insecurity are suggestive. In Lagos, lower rates of crime may be expected since militia do provide security and are based in slum areas. But these findings are surprising for Nairobi, since these neighborhoods rarely possess organizations that claim to provide security and services as Lagosian militia and vigilante groups do.

A second key question is whether weaker state presence in slums—and the attendant growth of organizational competitors to the state—results in different attitudes toward the rule of law? In other words, might slum-dwellers’ greater acquaintance with local armed actors and other “extralegal governance structures” make them less opposed to the use of violence to achieve certain ends? Several rounds of the Afrobarometer have asked a question that serves as a rough measure of the acceptability of political violence. Respondents are asked to agree with one of two statements: (1) “In [country name], the use of violence in never justified” and (2) “In this country, it is sometimes necessary to use violence in support of a just cause.” They are then probed for the intensity of their agreement. Using cross-national data from several rounds of the Afrobarometer, Kirwin and Cho (2009) have found that weaker state institutions and higher levels of crime at the national level correlate with higher popular toleration of violence. In an earlier paper, Bhavnani and Backer (2007) found that some forms of social capital—membership in religious organizations, for instance, but not in business or professional associations—were correlated with lower acceptance of violence. Both of these studies, however, have relied on pooled data, and they do not examine differences in attitudes toward the acceptability of violence within countries.

Figures 1 and 2 provide survey data on levels of acceptability of violence in Nairobi and Lagos at the level of slums and cities as a whole, as well as for Kenya and Nigeria. In both Lagos and Nairobi, slum residents express attitudes toward violence that are distinct from those that live elsewhere in each city. Among the general population in both Kenya and Nigeria, tolerant attitudes toward violence are not typical. In Kenya, large majorities—71.5 % of all Afrobarometer respondents and 83 % of Nairobi residents—find violence unacceptable. In Nigeria, rejection of violence is equally overwhelming.²⁴ At the national level and in Lagos, over 78 % of Nigerians express agreement with the statement that “violence is never justified.” Residents of slums, however, do not seem equally unified in their support for this societal norm. In stark contrast to the 83 % of Nairobi residents who reject violence, only 46 % of Nairobi slum residents agree that violence is never justified, 27.3 % agree that violence is sometimes necessary, and a further 22.2 % strongly agree with this statement.

Fig. 1

Acceptability of violence in Kenya, across samples

Fig. 2

Acceptability of violence in Nigeria, across samples

Nairobi slum residents strongly agree with “violence is sometimes necessary” at a rate roughly twice that of those who live elsewhere in Kenya. Rejection of violence in Lagos was higher than in Nairobi slums, as 58 % expressed some degree of agreement with the first statement and a further 19 % agreed and 18 % strongly agreed with the statement “violence is sometimes necessary.” In Lagos, only 15 % of residents agree or strongly agree with this statement. These differences in attitudes toward the acceptability of violence seem to suggest that slum-dwellers in both Lagos and Nairobi have a distinct conceptualization of order, one that allows greater latitude for local armed actors to play security or enforcement roles. Without more fine-grained data, it is difficult to investigate what exactly this means in terms of societal trust or recent state attempts to penetrate and police slum neighborhoods. These patterns of public opinion do seem to suggest, however, one potential obstacle to the establishment of a state monopoly on the use of force in both cities’ slum areas. In this attitudinal environment, it seems likely that the cooptation-oriented strategies of Lagosian authorities would have greater traction than the Kenyan state’s repression-focused approaches.

Conclusions

This paper began with the stated presumption that urban communities are governed by forms of predictable order or knowable sets of rules that manage interactions. In both of the cities examined here, the state is incapable of providing this order or even the most rudimentary security to residents, which opens up space for the establishment of informal institutions of order. This paper has argued that local armed actors only erratically play this role; instead, I stress that both political manipulation and state repression can erode whatever links and ties of accountability might have existed between armed actors and the communities upon which they ostensibly rely. Predation is as likely as public goods provision, since these organizations, at their root, directly rely on coercion to sustain their material base. In some contexts, violent organizations reinforce cultures of violence and intergroup hostility (e.g., Adebanwi 2005; Pratten 2009; Vigil 2003). Therefore, rather than improving group security, ethnic militia may directly undermine ethnic security by provoking countermobilization, as evidenced by the proliferation of violent organizations in both Lagos and Nairobi over the course of the 1990s. Further, as Arias (2013) argues, small differences in the relations between violent organizations and politicians can have substantial effects on the degree of protection and order provided by these violent organizations. Where political manipulation or direct repression upset existing patterns of business and behavior, violent organizations can take on more predatory—or less protective—roles vis-à-vis local communities.

I have argued that ties between local communities and violent organizations are much weaker than is typically presumed, and the general tendency of organizations is to develop stronger autonomy and operational freedom over time, particularly where strategies of political control are often contradictory. The rhetoric used to mobilize recruits—or subsequently used by politicians to link violence specialists to their own political aims—often conflicts with the actions of these organizations. Thus, violent organizations often stress their connection to local communities and assert a form of legitimacy based on the failure of governments to protect those communities. In the cases of the two ethnic militias discussed in greater detail previously, feelings of ethnic marginalization featured prominently in each group’s early self-descriptions. Both groups periodically used a rhetoric that stressed group victimization and the “hijacking” of power by rival groups. In Kenya, the ethnic militia Mungiki was formed in order to revive traditional Kikuyu religious practice, though it also had overtly political elements (Wamue 2001). In Nigeria, the OPC was even more explicitly dedicated to the “rightful” restoration of power to the Yoruba, an ethnic group that composes 20 % of the national population and is the dominant ethnic group in Lagos. The previous analysis does not clearly suggest that militia, especially those in Nairobi, consistently play this role—or necessarily possess any form of stable societal legitimacy.

This article also shows that states respond to established armed actors in very different ways. Though further research is necessary, the comparison here suggests that violent organizations are difficult to dismantle once constituted. Further, popular attitudes toward violence in slum areas may also make the state’s attempts to enforce a monopoly on violence difficult. The poor legibility of low-income and informal settlements suggests that brute force and repression are unlikely to have substantive effects on the activities of militia or the ability of states to expand their reach into ungoverned urban spaces. In both Nairobi and Lagos, state security forces face very high levels of distrust from local communities, which will make any direct assault on actors that reside in these communities less likely to succeed. In contrast to Weinstein’s (2013) article in this volume, which suggests the general efficacy of state attempts to exert authority over slum communities, this article underlines the potential of local actors to evade and adapt their organizations in the face of large-scale interventions aimed at state control. The Lagos State Government seems to have recognized this fact, and it has pursued a different approach to expanding state reach and extractive capacity in urban slums. As suggested previously, rather than attempting to repress or dismantle the informal networks that have been built in Lagos slums over the past several decades, the state government has attempted to coopt these networks to serve state purposes.

Of course, collaborative relations between the state and local armed actors are problematic. Informants in Lagos noted that coercion and intimidation remained the key tactics used by the state’s new urban enforcers, and the form of tax farming practiced by these agents likely results in some degree of extortion and overtaxation. There is also a danger that this collaboration reinforces the violent and undemocratic forms of linkage that connect local communities to vigilante groups, militia, and indeed, the Nigerian state. Further, as noted previously, the political mobilization of ethnic militia for electoral purposes can undermine whatever stores of legitimacy these organizations possess among local communities. At the same time, slum residents have seen substantial improvements in state services and in security, and the governor enjoys staggering levels of approval in slum communities. In the Lagos slum survey, a surprising proportion of slum residents expressed willingness to pay taxes in exchange for improved state services and approval of the right of the state to collect taxes. It will be important to monitor and assess whether this pattern is replicated in other Nigerian cities, which have made substantially fewer investments in tax collection capacity and in services improvement than has Lagos State. There remain questions about how current bargains might be transformed into more bureaucratic and regular forms of state control in slum communities in Lagos. But the contrast between the two cities suggests that indirect subversion of informal networks may be a more fruitful means of confronting non-state actors than the more directly repressive policies pursued in Nairobi and in many other cities in the developing world.