Violence in the city that belongs to no one: urban distinctiveness and interconnected insecurities in Nairobi (Kenya)

ABSTRACT Rapid urbanisation in the global South has prompted attention to the causes and dynamics of urban violence. Yet, much research tends to either analyse urban violence without attention to the broader conflict complexes of which it forms a part, neglecting linkages between different forms of urban violence and between urban and rural dynamics, or conversely study violence in cities without acknowledging the particularities of the urban context. In this article, we conceptualise urban violence, theorise how it is shaped by urban dynamics and explore its manifestations in Nairobi, Kenya. We find that while Nairobi is not uniquely violent inside Kenya, violence takes on distinct urban forms given city-level processes, and also that urban violence has led to policies that increase securitisation and militarisation of the city. Our analysis thus improves knowledge of how criminal and political violence is shaped by and shapes the stability of developing cities.


Introduction
More than half of the world's citizens today live in cities, and countries in the global South are home to the most rapid urbanisation processes. 1 Sub-Saharan Africa, in particular, will see dramatic rates of urbanisation as still relatively rural populations become more and more urban. For instance, Dar es Salaam (Tanzania), Kinshasa (DRC) and Luanda (Angola) are all projected to grow by about 5 million inhabitants from 2016 to 2030, and Lagos (Nigeria) by over 10 million (from 13.7 million in 2016 to 24.2 by 2030). 2 In a region heavily affected by political turmoil as well as challenges related to climate change and food insecurity, analysts and policy-makers have increasingly sought to understand how these factors relate to an elevated risk of violence in these fast-growing cities.
This article investigates the distinct features of urban violence in Nairobi. Nairobi is the largest city in Kenya, its capital, and an international hub in East Africa. It has been subject to rapid urbanisation for decades and while modern and cosmopolitan in its outlook, Nairobi displays significant inequality and separation between the wealthy and poor, hosting large informal settlements, where living conditions are dire. Nairobi is sometimes referred to as the city of no one and everyone, being a melting pot of cultures rather than historically belonging to a specific ethnic group. 3 Yet, violent contestation in Nairobi is fierce and ranges from criminal violence to electoral violence and even large-scale terrorist attacks. Whereas many forms of violence in Nairobi can only be understood in relation to the broader dynamics of conflict in Kenya, we argue that violence in the city takes on specific characteristics and unique manifestations. Several scholars have highlighted the connection between local and national processes to explain the prevalence of urban violence. We contribute to the literature on urban violence by making explicit both how urban violence is distinct from (in terms of the dynamics it takes), and intertwined with (in terms of its drivers and causes), broader conflict complexes in Kenya. To this end, we focus on four categories of urban violence in present-day Nairobi, all connected to broader political dynamics: (1) urban land conflict; (2) election-related violence; (3) state repression and extrajudicial violence; and (4) terrorism and radicalisation.
Establishing a theoretical and empirical foundation for understanding the manifestations of violence in rapidly urbanising cities, such as Nairobi, is important for several reasons. By placing urban violence in the broader context of rural-urban dynamics and national politics we avoid the trap of localism and consider factors important for understanding urban insecurity, while simultaneously accounting for the impact of the unique features of the city.
Grasping these dynamics is critical for understanding the causes and consequences of violence and, in turn, to design measures required to stem and prevent violence. Reducing urban insecurity is also of great relevance for achieving political and social development, given that urban elites and constituencies tend to be responsible for spearheading and promoting change. In addition, cities serve as vehicles for national economic growth and are therefore constituent of the larger economic development of any country.

Urban violence in previous research
Urban violence has been approached from different disciplines, including political science, political geography, sociology and criminology, and consequently with different conceptual understanding of its key features. 4 Across fields, research has identified trends towards forms of conflict that are 'fundamentally urban in character', 5 underscored by the violent protests of the Arab Spring and subsequent armed conflicts in the Middle East which have seen a high proportion of urban violence. 6 Broadly, urban violence has been characterised as belonging to four different categories which often overlap and reinforce each other: political, institutional, economic and social. 7 A large share of research on urban violence focuses on criminal violence, and the actors associated with it, such a gangs, armed political groups and criminal cartels. 8 Several scholars demonstrate how criminal and political violence are closely intertwined. 9 For instance, Barnes has proposed 'criminal politics' as a concept that captures the 'interaction between states and violent organizations that are motivated more by the accumulation of wealth and informal power and which seldom have formal political ambitions pertaining to the state itself'. 10 Empirical studies have tended to analyse urban violence in isolation, as a distinct phenomenon, overlooking how broader political processes and conflict complexes affect how it arises. 11 However, some studies point to how urban violence is influenced by rural and national-level conflict dynamics. Büscher notes how 'the nature of socio-economic or political networks within urban neighbourhoods is often strongly connected to conflict dynamics on the national or regional level', 12 while Beall et al. spell out how organised violence between armed actors and other forms of violence often overlap and intersect in the city. 13 Specifically focusing on how politics affects urban violence, LeBas analyses the role of 'ethnic militias' in urban violence and informal governance in Nairobi and Lagos, and shows that their role is strongly influenced by political dynamics in other sites and scales. For instance, during Moi's rule in Kenya, modes of electoral repression outsourced to ethnically mobilised armed groups were initially employed in rural Rift Valley, but later adopted in Nairobi as well; and urban migrants displaced by violence in rural areas became a source of recruitment for violent groups operating in the city. 14 A general observation is that in many African contexts there is a close connection between urban and rural communities which relies on the 'commitment of many urbanites to "the village"'. 15 In Kenya, this partly refers to the reproduction of ethnic cleavages, where Nairobilike much of Kenyaremains politically and spatially divided along ethnic lines. 16 Some scholars also note how violence, as it moves into the city, takes on new dynamics. For instance, Raleigh finds that urban violence comprises a broader range of actors and strategies compared to conventional (largely rural) civil war violence. 17 Others show how countries which historically have experienced rural-based rebellionsuch as Uganda and Burundihave seen a shift in the locus and mode of violence, with a current predominance of political violence being manifested as urban riots. 18 Similarly, several scholars have an interest in how migration to urban centres influences urban unrest. With migration to urban centres, settlement patterns change and generate dynamics of competition for space and land in cities. 19 While this may increase the general risk of violence, the effect is not direct: For instance, Østby finds no direct effect of urban in-migration on urban violence, but concludes that the effect is conditional on the marginalisation of certain groups. 20 We build on these insights to explore how rural conflicts and national-level politics influence violence in the city, while simultaneously considering how violence, because it takes place in the city, manifests. We demonstrate how violence in Nairobi is linked to broader conflict complexes, but at the same time is distinct in the sense that urban density, the political and economic importance of the city, and the presence of a high number of groups and political elite actors shape the intensity and forms of violence.

Urban violence and its distinct features
We recognise that urban violence has close connections to broader rural and national processes that drive conflict and insecurity. However, we propose that violence in cities takes on distinct features and theorise how violence manifests because it takes place in an urban environment. By violence, we mean physical forcekilling, injuring or coercionthat is intentionally employed against another person, group or community. 21 We do not restrict the definition to explicitly political violence; while we are interested in the political origins and impact of urban violence, previous research has shown that criminal and political violence are closely intertwined, particularly in states characterised by a history of armed conflict, and that the distinction between them is often analytically problematic. 22 Urban violence is commonly understood and operationalised as violence that takes place in the city. 23 However, the city also affects the form that violence takes. 24 Cities are understood as geographically delimited political entities with a high concentration of population and infrastructure. There is no universally accepted definition of what constitutes a city; rather, states tend to have their own criteria for exactly how and where to delimit cities geographically and in terms of inhabitants. 25 Within the field of urban studies, the city and the urban are theoretically distinct but mutually constitutive; the city is understood as the arena, while the urban is the processes and ways of life that emerge as a consequence of the city's nature. 26 In turn, like the concept 'urban' more broadly, urban violence is contingent on the nature of the city. Specifically, we expect violence geographically located in citiesurban violenceto take on distinct manifestations because of the heterogeneity, density, openness, and centrality of cities in the wider political context. 27 First, the heterogeneous nature of cities implies that a broad range of communities, groups and interests are represented in and inhabit the city. While this heterogeneity may foster cosmopolitanism and tolerance, 28 it may also create conflict, depending on the wider context. The frequent intersection in cities between heterogeneity and large socioeconomic inequalities increases the risk of conflict and violence. 29 Second, cities are densely populated. Together with heterogeneity, this means that different groups and interests cannot stay isolated: the city functions as a mediation point where people and ideas come into frequent contact. 30 In addition, density can result in a shortage of resources such as housing, or employment opportunities, frequently connected to access to space and land. 31 In the urban context, with different identity groups living closely together, competition for scarce resources is often induced to take on an identity dimension, which can transform individual conflicts into collective causes.
Third, the openness or permeability of cities implies that people can move freely into the city and within it, and there are public spaces which enable contact and public expressions of opinion. This means that cities are the main sites for civil mobilisation and protest, phenomena that are usually peaceful but that in some contexts become violent. For instance, the origins and development of the Arab Spring uprisings were closely connected to the urban context and to a significant extent played out in public spaces such as the Tahrir Square. 32 Similarly, electoral violence often takes the form of protest and clashes in urban public spaces. 33 The openness of cities facilitates intense networking and connectivity both among urban groups and with actors on the global stage. 34 But this also means that violent individuals and groups can relatively easily enter cities, or organise within them, and remain undetected until they perpetrate an attack. 35 Finally, citiesespecially capitalshave high symbolic and political importance and carry a particular weight within their broader context. 36 Political power tends to be concentrated in the cities, as well as economic wealth and cultural resources. This means that cities become magnets for those seeking better opportunities; in many countries, this has led to a fast growing urban poor population, often residing in informal settlements. 37 It also means that cities become valuable as an object of conquest, or as a target for highimpact attacks. In state-based conflict, a key objective is often to take over the national or regional capital and hence secure control over the state apparatus. Coups, a particular form of armed conflict, usually emerge from within the capital itself. 38 Furthermore, the symbolic importance of cities together with their density makes them prime targets for terrorist attacks, as illustrated by recent attacks in Mumbai (2008), Mogadishu (2011), Paris (2015), Stockholm (2017) and Colombo (2019). Such attacks are often framed and/or perceived as attacks on the urban way of life itself. 39 In turn, the threat of terrorist attacks and other attacks on the state within the urban context affect levels of militarisation and coercion on behalf of governmental security forces. 40 Finally, the high concentration of important economic resources and infrastructure contribute to high levels of crime, but also to intergroup conflicts over valuable urban land.

Violence in Nairobi
Nairobi is the capital and largest city in Kenya. It is a modern and cosmopolitan city, but also characterised by significant inequality and separation between wealthy and poor segments of the city's inhabitants. Nairobi also serves as an international hub in East Africa; the stability and peacefulness that characterised the country from independence to the mid-1980s contributed to the influx of investment and tourism as well as establishment of regional headquarters by major international organisations, including the UN. 41 Politically, Nairobi, along with Mombasa, has a unique status within Kenya in that it is both a city and a County. 42 As the country's major political and economic centre, Nairobi attracts migrants from both rural Kenya and from other urban centres in the country, and has been the site of rapid urbanisation for decades. As a result, 60-70 per cent of its inhabitants live in poor and densely populated areas, and much of the urban violence takes place in these settlements. 43 Security constitutes a main concern for many of Nairobi's citizens, in particular in the informal settlements. 44 Nairobi has also witnessed several other forms of violence, including terrorist attacks. Rampant insecurity has stimulated the emergence of a broad range of private security providers in both public and private spaces, creating a 'pluralized security landscape'. 45 In a global comparison, Nairobi is a young city: it was established in 1899 by the British colonial authorities, and at Kenya's independence in 1963 only had around 300,000 inhabitants. 46 At the time of the latest census, the population figure had risen above 3 million. 47 The rapid growth of the urban population, much of which has been informal and unregulated and beyond the control of city planners, 48 mirrors major cities elsewhere in the developing world. 49 The relatively recent urbanisation also means that power is not yet entrenched in the city; although Nairobi is 'the key to Kenyan political life', much of the political power base remains centered on rural areas and issues. 50 Also, the city has grown in a manner that has promoted segregation, including along ethnic lines; partly this is because waves of urban growth have been connected to conflict and violence around the country. These dynamics are similar to many other developing states, and make Nairobi a suitable context to study the dynamics of urban violence and to generate insights with broader relevance to contemporary urbanising societies.
To explore the different manifestations of urban violence in Nairobi, Kenya, we conduct a qualitative analysis drawing on insights from a range of different sources, including a thorough review of secondary sources (including reports by human rights organisations and NGOs, official sources, and academic articles), as well as interviews conducted in Nairobi in 2014 and 2018 with local experts and academics, journalists, NGOs working on issues related to urban security and conflict resolution, and government officials. Interviews with in total 59 individuals were carried out in a semistructured format. The questions centered on the main forms of violence prevalent in Nairobi, the main actors involved, as well as strategies to manage and prevent violence. Two local research assistants, with experience from academic and practical work on conflict and peace-building, assisted in identifying relevant interviewees and served as interpreters when needed. Interviewees were strategically selected based on their expected insights into phenomena of relevance for the study. We used multiple entry points to identify and access interviewees in order to avoid potential biases. The material from the interviews, thus, represents the views of a broad range of different stakeholders within Nairobi, and is complemented and triangulated using news reports and other secondary sources.
The empirical analysis is structured along four main categories of urban violence in Nairobi identified in previous research: urban land conflict, electoral violence, terrorism and radicalisation, and state repression. While recognising that these are interlinked, for each category we analyse how urban violence relates to broader conflict complexes in Kenya, and how the urban context affects the manifestations of violence in Nairobi.

Blood and soil: urban land conflict
Land has long been at the heart of politics and violent conflicts in Kenya. Political developments during colonial rule and following independence have meant that land is closely associated with ethnic identity, which is often activated and mobilised in conflict over resources and political power. 51 Importantly, land tenure has remained closely connected to communal identity, and political parties have to a significant degree formed and mobilised along ethnic lines. 52 While largely spared from large-scale civil unrest, rural Kenya has experienced numerous violent conflicts at more localised level over the right to land. For example, violent conflicts over land in Kenya's Rift valley have historical roots and are linked to the rural property regime, where the state, in the post-colonial period reallocated land and thereby 'structured the geographic pattern of land-related grievances, defined rival constituencies of land claimants, and created opportunities and incentives for ruling elites to manipulate existing land grievances and land-tenure relationships for electoral gain'. 53 In particular, government resettlement schemes contributed to pit the indigenous Kalenjin and Masaii against groups settled in the area later and perceived to be newcomers, in particular the Kikuyu.
In Nairobi, land conflict takes on a different dynamic than in rural areas due to the heterogeneity and density of the city, and the value of urban land. Land conflicts are common in the informal settlements, where much of the land is government-owned but is informally used and developed for private purposes and profit. One prominent example is the conflict over the right to land in Kibera, which is labelled as the city's largest slum and characterised by overcrowding and a predominance of poor neighbourhoods. 54 Urban land is generally highly valued and because of its proximity to central Nairobi, land in Kibera is a valued asset and rental property a major source of income. 55 While some areas of Kibera are privately owned and some parts are occupied by middle class estates, much of Kibera has been informally developed and occupied by housing structures on land owned by the government. Without any formal authority, local government officials 'approve' housing structures on state-owned land. In this way, 'housing in Kibera is far from temporary and those receiving permission [. . .] to build are acting as landlords'. 56 Under such circumstances of informal land use, the question of which community arrived first becomes important, since 'they can make a claim to own the land'. 57 Kibera originated as a settlement for Sudanese soldiers who retired from the British colonial forces. Although not a homogeneous ethnic community, the Sudanese exsoldiers and their descendants adopted the Nubian identity in independent Kenya in order to promote their rights in a political system which places strong emphasis on ethnic affinity, and where identity, citizenship and land rights are closely intertwined. 58 A community leader interviewed in 2014 argued: 'Nobody can be stateless, and with citizenship you shall have all rights and privileges. The constitution mentions community land. Many think of rural land, but Nubians are urban based, their community land is urban. Urban land is not necessarily for everyone. If this is not recognised, there will be cause for conflict'. 59 The Nubians' claims to the land for long went unrecognised, and Kibera's unregulated status over the years attracted a large influx of migrants from all over the country. With Kibera today being ethnically divided, largely poor and densely populated, tensions have persisted between the Nubian community on the one hand and the government or other ethnic communities in Kibera on the other, as well as between landlords (from the Nubian and other communities) and tenants. 60 The struggle for land in Nairobi's informal settlements is closely linked to the political economy of their large-scale informal rental markets. 61 In Kibera, the Nubian community has retained a major share of the housing structures, but the Kikuyu landlords have surpassed them in numbers; the majority of the tenants have come from the Luo and other communities dominant in western Kenya. 62 Tensions between landlords and tenants, which are largely interpreted along identity lines, have on numerous occasions escalated into intercommunal violence, often connected to broader political dynamics. In 1995 and 2001 for instance, there were violent clashes between Luo 'tenants' and Nubian 'landlords'. In 2001 the clashes were triggered by a political rally held in Kibera by Presidential aspirant Raila Odinga, who stated that 'the government is the true landlord' (given that the land is formally government land), and that landlords in Kibera should be forced to reduce their rents. 63 More broadly, land issues in Nairobi's informal settlements are linked to politics since many national and local politicians own structures in these areas and the residents make up important electoral constituencies. 64 Kibera has become a key electoral mobilisation ground especially since Raila Odinga began to establish himself as a national politician in the early 1990s. Odinga, a Luo, has relied on patronage networks to consolidate Kibera as one of his major strongholds. While the conflict over the Nubians' claims to land is somewhat unique, dynamics related to patronage, insecure tenure and contests over 'slum upgrading' are common focal points for urban violence also in Nairobi's other informal settlements, such as Mathare and Kawangware. Like in Kibera, these areas are ethnically segregated, politically powerful individuals are prominent stakeholders, and disputes over land or tenancy claims have often escalated into intergroup violence. 65 For example, in connection to the 2017 elections, some candidates allegedly made promises to tenants in Kawangware that rents would be reduced if those candidates were successful, and this led to conflict between landlords and tenants along ethnic lines; according to one interviewee, the conflict parties enlisted organised gangs such as Mungiki in the ensuing violence. 66 Another interviewee further emphasised the political dimension of this violence, noting that 'Kawangware is a multiethnic settlement, where people generally live peacefully, except during political campaigns'. 67 'Slum clearances'where people have forcefully been displaced from their homes in informal settlements, ostensibly with the purpose to provide better and safer housinghave also often been violent, and at times interpreted as targeting certain communities. 68 One interviewee referred to controversies during different phases of upgrading in Kibera, and noted: 'the problem is that politicians infiltrate the programs and they become politicised and ethnicised'. 69 'Ethnic politics' and urban armed groups As mentioned, Kenyan politics is closely connected to ethnic networks which determine access to power and resources, and violence in the urban informal settlements as well as in the rural areas can often be traced to the national political arena and the divisions between parties and politicians who mobilise along ethnic lines. 70 Kenya has been heavily affected by electoral violence, which similar to land conflicts has both a national scope and an urban dimension. Electoral violence constitutes violence that is intended to influence the process surrounding an election and its outcome. 71 Scholars have brought attention to urban-rural divides in voting patterns, and a rural bias where many electoral systems disproportionally favour the rural populations' vote. 72 But from a violence perspective, the urban is also of high significance. In Nairobi and elsewhere, cities have become a focal point for violent mobilisation around and after elections. While electoral violence takes place in both rural and urban areas, it has also been spurred on by urban processes, making cities vulnerable to such violence. Several factors make urban dynamics particularly relevant for electoral conflict.
A first issue concerns connectivity and high population density which have implications for electoral outcomes as well as violent mobilisation. 73 Firstly, relatively small geographical units become electorally important: for instance, during the Moi era, gerrymandering, i.e. the redrawing of constituency boundaries, was commonplace in urban areas as a way of reducing the influence of populous, opposition-leaning areas 74 ; evictions and state-sponsored violence are other strategies available to incumbents to affect the vote. 75 Secondly, since rural areas in many parts of the world remain isolated and difficult to penetrate, political parties tend to have their greatest organisational capacity in urban areas. 76 Thus, the opposition's ability to organise electoral protest that may turn violent tends to be high in urban areas, as well as the incumbent's capability to suppress or influence the vote in opposition strongholds. Combined with the processes of urbanisation which create a high proportion of underemployed young men, many political parties will be connected to networks of individuals and groups that can be utilised for violent electoral purposes. This holds true for Nairobi, where politicians across the political spectrum have been able to mobilise 'private security' providers and violent protestors with ease. 77 In the latest national elections, 2017, electoral violence was predominantly urban and concentrated in informal settlements such as Kawangware where it intertwined with existing conflicts over land and rent. 78 In Kenya, electoral violence has also entailed a transportation of previously rural dimensions of conflict along identity lines into urban areas. Appeals to land grievanceswhich tend to be used to mobilise support among particular ethnic constituencies -have been important in violent electoral competition and have played a role ever since the reintroduction of multiparty elections in 1992. 79 More recently, massive electoral violence erupted in Kenya in connection with the 2007 election. The violence, which began in 2007 and continued in early 2008, took place both in rural areas (such as the Rift Valley) and in urban contexts such as Kisumu and Kibera settlement in Nairobi. 80 The violence was primarily a response to what was seen as an unjust election result, but rooted in a longer history of perceived political and economic injustice, including uneven distribution of and access to land. Among groups who felt excluded from power and economic opportunity, resentment had been brewing due to a growing centralisation of power around the President, and these groups held hope for the opposition candidate Raila Odinga. 81 When it was announced that the incumbent Kibaki had won the election, violence escalated and spread quickly, including in Nairobi, and only ceased after a power-sharing agreement and the leaders' call for the violence to come to a halt.
Nathan has observed that contrary to electoral dynamics in middle-and upper-class urban areas, where ethnicity has become a less salient mobilising force in electoral mobilisation and vote choice, poor areas in African cities such as Accra remain susceptible to ethnic competition and political clientelism structured along ethnic lines to a larger extent than in rural areas. 82 In Nairobi, the violence took a distinct ethnic dynamic, where specific communities were targeted and in turn mobilised counter-attacks. According to Jacobs, 'traditional myths about the existence of "ancestral homelands"considered to be bound by specific ethnic communities by bloodwere transferred to Nairobi's suburbs and violently enforced'. 83 Along the same lines, one of our interviewees remarked that 'Kenya is unique in that everyone wants to have a rural homeland. This means that conflicts will have an immediate spillover effect'. 84 'Ethnic zones' were established in the informal settlements, and when looting and arson became wide-spread people were forced to flee to their respective ethnic enclaves, which they regarded as safe since tribally-based gangs took over control. Especially affected was Kibera, where armed groups of local Luo attacked and looted Kikuyu houses and businesses. The observation that urban poor youths can be easily mobilised for political interests held true during the post-election violence, when poor Luo youth looted Kikuyu property while inciting, if not forcing, others to join them. At least partly this violence was politically orchestrated, and analysts concluded that the lower party level, such as local politicians and youth leaders, was ambitious to climb up the ranks by any means available. 85 In the poor, densely populated and marginalised areas of Nairobi, socio-economic frustrations could be easily manipulated for political interests.
The 2007-2008 violence was particularly severe, but electoral violence and the use of militant youth gangs has a long history in Kenya, and also in Nairobi. 86 While 'urban areas were initially less implicated in [. . .] formal campaigns of electoral violence, [. . .] by 1997, the ruling party youth brigade was deployed in Nairobi, where it violently disrupted rallies for constitutional reform and attacked opposition leaders'. 87 Armed groups mobilised along ethnic lines were formed and were 'hired' by both the incumbent and opposition parties for 'security'. The recruitment of members into such groups, most prominently the Mungiki, has been connected to the growing population of unemployed young men in Nairobi's informal settlements. 88 As mentioned above, the elections in 2002 were also marked by violent conflict stemming from a politicisation of the informal rental market which was drawn into the electoral campaigning by the MP of Kibera, Raila Odinga from the Luo community. Simultaneously, election-related conflicts have become intertwined with pre-existing conflicts over local control between urban armed groups which transcend the criminal-political divide. Schuberth notes that Nairobi hosts a range of 'communitybased armed groups' -CBAGswhich, depending on the time and circumstances, engage in activities ranging from crime and political violence to local service provision and governance. 89 For instance, the Kariobangi North massacrea massacre by the Mungiki in a Luo residential areahas been described as a 'milestone in the 2002 electoral campaign', 90 but it is representative of a larger pattern of violent conflict between rival armed groups that fight over control over certain areas. 91 During electoral campaigns, such groups negotiate their relationship with political patrons and exchange their violent agency for monetary rewards as well as promises for monopoly over local business sectors if their patron wins. 92 The armed groups often take on a life of their own after the elections are over; one of our interviewees observed that 'when they are abandoned by the politicians they engage in other sectors such as providing security, and make a point of being seen'. 93 Several of the armed groups, including Mungiki, have also carved out major roles in organised crime. Like many other rapidly growing cities, Nairobi from the mid-1980s experienced rising criminal violence. 94 In the city's densely populated informal settlements, criminal violence is claimed to be endemic, and 'widespread poverty and lack of livelihood opportunities for a large proportion of the urban poor exacerbate their vulnerability to being victimised by non-state violent actors and sectors of the state implicated in inciting and perpetrating violence and crime'. 95 At the same time, levels of crime vary a lot across informal settlements (as well as within them), 96 but the lack of consistent and disaggregated statistics makes patterns difficult to ascertain. There are few comprehensive and transparent sources of statistics on violent crime; while the Kenyan police reports on overall trends in crime, the reliability of the data is strongly questioned. 97 Policing the city: from collaboration to state repression As has been observed in several violence-affected cities, 98 violent crime in Nairobi is in many cases linked to politics, or at the very least facilitated by political dynamics and the presence of ethnically mobilised armed groups which also engage in organised and violent crime. 99 During the electoral violence in 2007-2008 for instance, much of the violence in the informal settlements was simply criminal in nature, demonstrated by the prevalence of looting when formal legal structures were overridden by chaos and lawlessness; yet it was clearly linked to the political violence (and had political effects). In terms of linkages to conflicts in rural areas, the influx into the city of internally displaced persons in the aftermath of different waves of political violence has also contributed to rising urban crime.
More broadly, over the past few decades, Nairobi has experienced 'an increase in physical violence that is popularly labelled as "criminal"since it is seemingly carried out for "personal gain" rather than [political purposes]', but these '"everyday" forms of violence in multi-party Kenya are intrinsically intertwined with the country's politics'. 100 Katumanga goes even further, arguing that political elites seeking to retain power in a context of political and economic liberalisation have facilitated 'the criminalization of urban existence'. 101 For instance, impunity for criminal enterprises is used as a form of patronage. Limited law enforcement has opened up violent competition in a number of domains, all of which have seen a high presence of vigilante groups: for instance, in the informal housing market and 'slum governance', public transport and street markets. 102 More broadly, applying the 'criminal politics' framework developed by Barnes,103 the relationship between state authorities and armed urban groups has ranged (over time, and across different groups) from close alliance to violent confrontation. At one end of the spectrum, politicians and other elites have entered into deals and alliances with armed gangs, for instance employing them as private militias for voter intimidation. 104 At the other end, the state has at times cracked down extremely hard on certain groups. For instance, around 2005 the Kenyan police formed a special unit to take action against 'ethnic militias' and vigilante groups, and the responsible Minister, John Michuki, ordered a 'shoot to kill' policy. 105 This was a response to the uncontrolled growth of armed groups, including the notorious Mungiki, which had developed close ties to national politics and even backed the ruling party in the 2002 election. 106 By one account, the state cracked down on these groups because they became too powerful and began to challenge those in power. 107 It is estimated that during the late 2000s, the special unit was responsible for over 1,000 extrajudicial killings. 108 Similar excessive force has remained common in Nairobi, particularly in the informal settlements, where reports suggest that young men are often shot on the mere suspicion of involvement in crime. 109 Such extrajudicial killings are perceived to be a particularly urban phenomenon, targeting urban poor neighbourhoods; 110 van Stapele notes that the stereotyping of poor, young men in Nairobi as criminals has resonated among the middle class in the city, rendering relatively high popular support for harsh measures. 111 As a result of heavy-handed policing measures, many of those killed in urban violence in Nairobi have died at the hands of the security forces. There have been repeated reports of excessive use of force by the police against protesters, demonstrators, crime suspects, and alleged gang and terrorist members. 112 To a large extent this violence can be related to a militarisation of urban policing in response to different security threats, but there are also important political undertones. This can be seen against a broader pattern where the Kenyan state has, under different regimes, engaged in politically-motivated violence and repression against its citizens. 113 In 2009, in the context of broader reform processes following the 2007-2008 violence and subsequent national political accords, Kenya embarked on an ambitious police reform project to increase the capacity and professionalisation of the force; however, available evidence suggests that such reforms have had limited effects, 114 and that attempts to improve police conduct have been hampered both by the existence of parallel structures and chains of command, and by political interference. 115 Such challenges are particularly present in Nairobi, which features a pluralised security landscape, 116 where national, county and paramilitary structures co-exist along with vigilante groups and private security operators. 117 In addition, urban planning affects security provision; in the words of one of our interviewees, 'the way the city was set up makes it difficult to police'. 118 There have been strong claims that those in power have used the security forces to repress and intimidate the political opposition, and police in Nairobi are broadly considered to be corrupt as well as biased toward certain groups. 119 In the aftermath of the August 2017 elections, there were numerous confrontations between opposition supporters and police in Nairobi, and widespread reports that the police used excessive force. A report by Amnesty International and Human Rights Watch documents 33 deaths in Nairobi, mainly in different informal settlements, that can be attributed to police actions; the report notes that '[r]esidents [. . .] believed they were being punished for the way that they had voted. Indeed, police statements to witnesses suggested the same'. 120 The terrorist threat, counter-measures and radicalisation The conduct of the security forces is also related to an increasing securitisation of the city in response to a terrorist threat which has intensified in recent years. Nairobi has been the site of several high-profile, transnational terrorist attacks; most notably, the 1998 Al-Qaeda bomb attack against the US embassy which caused over 250 deaths, the 2013 attack on the Westgate shopping mall by Al-Shabaab which resulted in 80 casualties, and the January 2019 attack on the 14 Riverside hotel and business complex, referred to as the DusitD2 complex attack, in the suburb of Westlands. While terrorist groupsin particular Al-Shabaabhave been engaged in violence across the country, attacks in Nairobi have entailed a certain dynamic and impact, as the urban density and symbolic importance of targets in the city serve to amplify the destructiveness and psychological impact of attacks. 121 The Westgate and DusitD2 complex attacks were directed against civilian and economic targets that may be seen as epitomes of contemporary urban, cosmopolitan life. 122 Al-Shabaab attacks in Kenya are linked to broader national and transnational dynamics, where the Kenya Defence Forces have participated in joint military operations in southern Somalia with the Somali Armed Forces to combat the Islamist insurgency. 123 While Al-Shabaab had already been stepping up its attacks on Kenyan civilians and security personnel near the Kenya-Somali border, the Westgate attack caught the authorities by surprisereportedly despite the availability of intelligence that should have led to preventive measures. 124 The attack transformed into a drawn-out siege that ended only after four days when the army, backed up by Israeli, British and US special troops, were able to defeat the attackers. The attack had the effect of transforming the security landscape in Nairobi, with increased securitisation and surveillance in public spaces. 125 A number of factors have been claimed to facilitate terrorist activities on Kenyan territory: low capacity of the security forces, spillover from the crisis in neighbouring Somalia, the establishment of Saudi-sponsored radical Islamic schools in Kenya, and not least the strong grievances among Kenya's Muslim population which has been marginalised and discriminated for decades. 126 The responses of the Kenyan security forces to terrorist attacks and threats have served to increase these grievances further by indiscriminately punishing Kenyan Muslims. Notably, in early 2014 'Operation Usalama [Peace] Watch' was launched and thousands of troops were deployed in Eastleigh, a large and strongly Somali neighbourhood in Nairobi (colloquially known as 'Little Mogadishu'), to identify and detain illegal immigrants argued to be linked to terrorism. 127 In the context of such antiterror operations, the security forces have been under critique for 'a discourse and understanding that "sees" outsiders, namely Somalis, as a threat, both internally and with regard to conflict spillovers from Somalia'. 128 While the Kenyan government has sought to portray al-Shabaab as an external threat which can be contained by stabilising Somalia and through better border and immigration control, terrorist networks have been able to exploit local grievances and recruit inside Kenya. Nairobi (and other urban centres) reportedly provide fertile grounds for radicalisation, including among non-Muslims. 129 For instance, a number of mosques in Nairobi were identified as recruitment hubs several years before Westgate. 130 Recruiters have also increasingly targeted unemployed youth in informal settlements, who can be enticed by financial rewards. 131

Concluding discussion
The investigation of urban violence in Nairobi gives important insights into both the distinct features of urban violence and how it intertwines with national and rural dynamics. Table 1 illustrates how concrete examples of violence connects to broader conflict complexes, while retaining distinguishing characteristics shaped by the urban environment in which they arise and unfold. Table 1 showcases the usefulness of unpacking urban violence across different dimensions. First, it highlights how the distinct features of cities and the urban processes contribute to explain how urban violence arises and unfolds. For instance, electoral violence affects both urban and rural areas, but in the urban context the eruption of electoral violence is facilitated by high connectivity, population density and the symbolic value of cities from a political point of view. Like other major cities, Nairobi is an opposition stronghold and at the same time it is the seat of national power. This heightens its political importance, generally, and during election periods, specifically. In 2013, devolution was implemented and Nairobi became a county, increasing the stakes in the governance of the city. 132 Second, our analysis underlines the close connections between different forms of violence, in line with previous research. For instance, LeBas highlights how vigilante groups and ethnic 'militias' in urban centres are closely linked to local as well as national politics. 133 But these groups also have taken on a life of their own, have become violence specialists, engage in violent and organised crime, and claim or fill security gaps where the state security apparatus is unwilling or incapable of providing security for its citizens. As suggested by Barnes' 'criminal politics' framework,134 the relationship between such groups and the state ranges from open confrontation, via passive coexistence, to alliance between political actors and armed gangs for mutual gain.
Third, we show how urban violence taps into regional and national conflict dynamics, and in some instances, it is even relevant to talk about a transfer of conflicts from the rural areas into the city, with high-impact terror attacks by al-Shabaab perhaps the most striking example. Notably, terrorism is linked to both state-based conflict (Kenya in Somalia), but also to a history of state repression and marginalisation of Muslim communities inside Kenya. Similarly, urban manifestations of land conflicts, linked to a narrative whereby ethnic identity, ancestral land and citizenship are closely intertwined. Reversely, urbanisation itself is a result of violent conflict in other parts of the country. In particular, the growth of Nairobi is in part connected to the influx of IDPs at various stages of Kenya's historyduring the Mau Mau conflict, during ethnic clashes under Moi, and as a result of communal conflicts in rural areas. 135 A final conclusion is that urban violencewhile often affecting specific areas in the city or aimed at specific targetsoften has broader implications for security and perceptions of insecurity in the city. As emphasised by Stephen Graham, a rise in urban violence has created an urbanised perception of the enemy, which in turn has Terrorism and radicalisation Al-Qaeda attack on US embassy in 1998 Transnational links and connected both to global and regional dynamics High-impact attacks, symbolic importance Al-Shabaab attacks, e.g. Westgate shopping complex in 2013 Shifting conflict patterns linked to regional dynamics, e.g.
before Westgate, Al-Shabaab mainly in the border area with Somalia/north coast Openness/permeability, combined with urban poverty and inequality, enables radicalisation created an urbanisation of the security architecture, including increased public surveillance and new security methods. 136 For example, violent criminal gangs and terrorist attacks in Nairobi have given the police a pretext for using excessive force, and created further ethnic segregation. 137 In addition, the threat of terrorism has securitised public space in Nairobi, with security checks and monitoring pervading everyday city lifeat shopping malls, hotels, and the university campuses. While such measures are aimed at forestalling security threats, it inadvertently also infringes on the city dwellers' freedom and creates a sense of insecurity.