Is There Anybody There? Police, Communities and Communications Technology in Hargeisa

The full-text may be used and/or reproduced, and given to third parties in any format or medium, without prior permission or charge, for personal research or study, educational, or not-for-pro t purposes provided that: • a full bibliographic reference is made to the original source • a link is made to the metadata record in DRO • the full-text is not changed in any way The full-text must not be sold in any format or medium without the formal permission of the copyright holders. Please consult the full DRO policy for further details.

On 19 August 2015, the Somaliland Ministry of Interior launched a text alert community police engagement programme at a small police station in the Macalin Haruun district of Hargeisa. Promoted enthusiastically by the minister and developed by advisers from EUCAP Nestor, a civilian mission forming part of the EC's external action programme, the project's objective was straightforward: members of the public would use their mobile phones to alert the police to security issues requiring attention. In becoming 'the eyes and ears' of the police, residents would help to improve policecommunity engagement and local security. But the project did not work out as EUCAP hoped. In the days following the launch, the minister returned to his office, the publicity campaign evaporated, the mobiles donated to the police stayed in their boxes, the station's commander went on leave for six weeks, the minister returned to his office, and the station's radio room was locked. By December, few if any, calls had been received.
Why should a modest and unsuccessful experiment in a dusty residential district (Wikimapia 2014) in the capital of a selfproclaimed republic deserve attention? The project's unexceptional record is typical as far as the region's crime reporting lines are concerned. There are, for instance, two tollfree lines in Mogadishu, one, 888, a crime reporting number and one, 5555, a rape reporting hotline run by an NGO. But while the 5555 line receives -and responds torequests for support on a daily basis, 1 the 888 line has yet to receive calls from the general public. And with the possible exception of Kenya, the record of reporting lines elsewhere in Africa is similar, with repeated awarenessraising campaigns failing to increase call rates. 2 Indeed, the record of such lines suggests that the failure of text-based systems is only to be expected. Nevertheless, Macalin Haruun's text alert system is noteworthy because it offers contextualised insights into both the specifics of police-community engagement and the use of mobiles as a two-way technology capable of reaching low-income or marginalized populations in relatively safe environments. It also raises general questions about the connection between communications technology and police-community relations in fragile states: Can information and communications technology (ICT) help to facilitate trust and communication between police and residents in societies with low literacy rates but high access to mobiles? What aspects of ICT help residents manage their everyday security? Is one-to-one communication between police and residents more important than international models of ' community policing'? How important is the police station as a site for engagement? What is the role of initiative in police-community interactions? What shapes low-level policing in safe urban environments?
Macalin Haruun's project may have done little to increase crime reporting rates, yet the experiment is more significant than it first appears. One reason for this is that it introduces a note of caution into overly ambitious discussions of developmentoriented policing. Although donors such as the UK's Department for International Development (DFID) have long assumed that ICT 'plays a key and integrated role in accelerating progress' towards achieving politically desirable goals such as poverty reduction (DFID 2007), and the World Bank regularly presents ICT as a tool for improving service delivery and accountability (World Bank 2016), ICT's potential application to the communal security underpinning development needs to be assessed realistically.
A second reason is that Macalin Haruun warns against assuming that ICT-related practices travel easily between societies. Prompted by the success in Kenya of ICTbased banking systems such as M-Pesa, commercial companies such as M-Kopa, open-source tracking projects such as Ushahidi, which allows users to send crisis information via mobiles, and the ease with which social media can be integrated into community policing (Omanga 2015), and reinforced by Somaliland's cheap call tariffs and high rates of access to mobiles, donors identify ICT as a tool for improving police-community engagement. Mobiles are increasingly seen as a tool for sharing information and lessening the distrust and poor response rates characterising everyday policing while facilitating desirable goals such as partnership and security for all. Macalin Haruun's experience suggests that this is wishful thinking: what works in Nairobi does not work in Hargeisa. Indeed, Macalin Haruun emphasises that local norms and preferences can counteract the availability of globalised technology.
This article explores these issues using Macalin Haruun as an instance of local responses to a donor-designed policing project, rather than as a case study of textbased ICT and how it might change the current scenario. It contributes to current understanding of police-community engagement in fragile environments by demonstrating that the key to sustainable and locally acceptable forms of policecommunity engagement is to be found in the knowledge and technical skills police need to fulfil societal expectations and preferences regarding the management of low-level insecurity. This takes us beyond value-based Hills: Is There Anybody There? Police, Communities and Communications Technology in Hargeisa Art. 2, page 3 of 16 assessments of formal and informal policing provisions, and debates about the significance of trust and procedural justice in determining police-community relations (Sunshine and Tyler 2003;Tankebe 2009), while emphasising that the relationship between police and community and the role of ICT within it depends on a range of context-specific variables.
The discussion that follows develops in four parts. The first provides background on the dominant perspectives influencing donor expectations regarding ICT, policing strategies and police-community engagement. The second focuses on the rationale behind the introduction of Macalin Haruun's text alert system and the Somaliland police's experience of it. The third shifts to why the project failed and what local people thought of it. Based on the views of focus groups exploring residents' expectations of the police, experience of visiting Macalin Haruun station, and knowledge of the text alert system, it offers an explanation for what actually happened. The fourth part concludes that Macalin Haruun's experience reflects local expectations of police provision and how they are, rather than should be, fulfilled.

Connecting Technology and Police-Community Engagement
At first glance, the connection between technology and police-community engagement in Africa is tenuous. The continent's police forces are usually described as corrupt, ineffective, resistant to change, and lacking in the technical skills and resources needed for humane engagement practices; most provide regime policing rather than community service, and are tolerated rather than trusted. General duties constables in countries from Ghana and Nigeria to Kenya and Somalia may have access to personal mobiles but many are also badly paid, untrained, unfit or illiterate. With the exception of the occurrence books (i.e. ledgers for logging incidents and enquiries) found on front desks, most police stations lack access to stationery and type-writers, let alone computers and connectivity. 3 Despite this, the export of democratic policing models has become a major industry -and scholarly sub-field -over the last two decades, with millions of dollars poured into ambitious projects intended to transfer 'professional' policing strategies, procedures and tactics to Africa's police (OECD-DAC 2007). The results are, at best, uneven. Perhaps because of this, reform advocates now look to exploit the opportunities for change and innovation associated with ICT even though ICT's record as a tool for improving people's quality of life, let alone their policing, is patchy; social realities are such that ICT cannot affect the overall incidence of insecurity, poverty and ill-health (Avergerou 2010: 3-6;World Bank 2011;Banks 2013). Furthermore, much of the donor-supplied technology addresses the internal technicalities of police work (e.g., forensic laboratories), rather than the everyday concerns of the communities populating the policing environment. Donors prioritize sophisticated forms of ICT, like computers, over the basic and inclusive forms of communication, such as radio programmes, songs, murals, and painted advertisements that most people rely on. In other words, donors' aims and objectives are disconnected from local realities, so ICT fails to relate to the socio-economic situation in which it is applied.
Our knowledge base on the connections between technology and police-community engagement in Africa's societies remains heavily dependent on developments in Kenya, which is widely regarded as a beacon for the application of ICT to policerelated issues. There is some truth in this. Kenya's National Police Service Strategic Plan 2013-2017 identifies the application of ICT in policing work as one of its eight strategic priority areas on the basis that ICT is a tool for modernising police-community engagement, with social media a means to increase accountability and trust in the police (NPS 2016: 9, 10, 12). The inspector general

Dominant perspectives on policing provision
The literature available on ICT and development adds little depth to this picture. While it seems reasonable to expect the debate on ICT for everyday safety to be influenced by empirically-based analyses from inter-governmental or nongovernmental organisations, in practice, decision-makers discuss policing in the light of liberal values such as accountability, diversity and ' community' engagement, rather than local preferences. Meanwhile the academy's approach to police and policing provision ranges from focusing on the informal or community-based groups providing the bulk of Africa's everyday security and justice (Albrecht and Kyed 2011) to theoretically informed analyses drawing on international relations or incorporating insights and approaches developed in critical security and post-colonial studies and the anthropology of the state (Göpfert 2013; Beek 2016). found in Somaliland, Cooper-Knock and Owen's insight prompts questions about the ways in which Hargeisa's inhabitants use -and help to reproduce -the police they often criticise. Macalin Haruun's experience reinforces this insight by showing how police and populace share an understanding of their respective roles in managing low-level insecurity.

Community policing
At the same time as international researchers rediscover the importance of police for local people, donors and researchers laud ' communities' with both evidently finding comfort in the co-operation and harmony associated with community symbolism (Findlay and Zvekić 1993: 32). Engagement is framed in terms of democratic ideals and values such as co-operation and partnership (Lindberg 2011), rather than, as is more realistic, the political economy of policing (de Waal 2015), and this introduces ambiguity into discussions of police-community relations. Even so, the term ' community' is difficult to avoid. It is therefore is used here descriptively, rather than analytically, referring merely to the inhabitants of neighbourhoods or districts falling within the responsibility of specific police stations; in this case, Macalin Haruun.
One reason why it is difficult to find an alternative to ' community' is that donor projects on security and justice are typically framed in terms of ' community policing', a controversial notion which can be defined as a philosophy or ideal that promotes policing as a shared endeavour in which police and communities work together to address crime and disorder. This results in conceptual confusion about the operational direction of policing, with understanding ranging from 'policing the community' to ' communities policing the police' (Brogden and Nijhar 2005;Denney 2015). But in Africa, where police forces reproduce the political and social order that those authorising or permitting their activities promote (Marenin 1995), community policing is always directed at managing the community, and this is bound to influence people's assessment of ICT for engagement. Donor definitions may emphasise joint problem solving, service, diversity and accountability, but the politically and functionally successful forms of community policing found in countries such as Nigeria and Ethiopia require communities to take responsibility for their own security and feed information to the police (Hills 2014a; Denney 2013). There is also policy-relevant confusion because community policing reflects donors' belief that police can be agents for social change; as Findlay and Zvekić note, donors are pre-occupied with making policing more relevant in its social context, rather than merely improving its crime control capacity (Findlay and Zvekić 1993: 33). Even when this is not the case -and EUCAP's advisers in Hargeisa focused on improving crime control capacity in order to improve policing and local security -community policing means whatever the speaker concerned wants it to mean.
Findlay and Zvekić define community policing as a 'selective process of communication and accountability' (Findlay and Zvekić 1993: 33), and this is the understanding adopted here. Their insight that the 'interactions of interest, power and authority' distinguishing 'the structures and functions of police work' should be viewed as ' constructed around expectations for policing within a given cultural, political and situational context' is similarly helpful (Findlay and Zvekić 1993: 6). In other words, police and communities ideally develop a pragmatic working relationship that builds on a locally acceptable understanding of their respective roles regarding the management of low-level forms of disorder. Macalin Haruun is one such case. Its officers and residents have seemingly developed an understanding in which neither introduces disruptive activities or ideas. Low-ranking officers may spend their days in the station, rather than in the surrounding streets, but residents also fail to engage proactively, and the attitudes of both are shaped by security levels, legacy factors, political sensitivities, and social norms as well as the resources available. The part played by discretion and initiative in such relationships has yet to be investigated but could prove informative about attitudes to ICT, especially among youths. For now, officers' use of discretion is more evident than their use of initiative. Indeed, discretion, or the tempering of strict rules for policy for operational reasons, is not only widely regarded as an essential and legitimate element in policing (Findlay and Zvekić 1993: 21), but also is a feature of Somaliland's legally plural society. Most crime is managed by customary law (xeer), rather than state law, and officers routinely enlist the support of traditional authorities or settle minor incidents informally. Furthermore, and despite anecdotal evidence of the unwillingness of many officers to delegate or accept responsibility, Somali notions of 'policeness', of what it means to be police, require tactical flexibility (Hills 2014b). Entrepreneurial ingenuity drives many aspects of Somali life and there is no obvious reason why policing provision should be exempt from this. For such reasons, focusing on the use of ICT in a small discrete project such as Macalin Haruun's text alert system is helpful. Its granular detail helps to avoid sweeping and unrealistic generalisations about the utility of ICT and liberal ideals of policeness and policecommunity engagement.

Rationale for a Text Alert System
Launched by the Ministry of Interior (MoI) though originally identified and developed by two Hargeisa-based EUCAP police advisers, the text alert project was seen as a way to prevent crime, target resources and improve police-community engagement in the area of the city known as New Hargeisa. 4 Drawing on their personal experience of a text alert system used in rural Ireland, the advisers developed a plan for blending crime reporting and community engagement in which text alerts would act as a tool for diffusing knowledge. They wanted to enhance local security while helping the police direct scarce resources to the areas where they were most needed (European Union External Action 2015).
The police station of Macalin Haruun was chosen by the MoI on the basis that it was small and easily monitored and its English-speaking commander would facilitate communication with EUCAP. It was also a showpiece, having been opened by the United Nations Development Programme (UNDP) as a model station in 2012 (UNDP 2012); that is, it was a place where police and community were already expected to interact (Lum and Fyfe 2015). In the event, the project quickly lost momentum. At the time of a visit on 9 December 2015, signs in Somali and English indicated the offices allocated to, for instance, the commander and the women and children's desk, but the compound was cluttered, the radio room from which the system is administered was locked, the most IT-proficient officer (a woman) had been replaced by a less competent man, and the front desk's occurrence book was not up-to-date.
With the benefit of hindsight it is evident that the project's chances of success were slim: the imported text-based system was introduced into an oral culture in which a high percentage of the population is illiterate; there are no examples of successful callbased crime prevention lines operating in the region, let alone text-based ones; and neither the MoI nor EUCAP was fully committed to the project which was, moreover, run on a shoestring budget of €8,500. It is difficult to avoid the conclusion that the project says more about donor dynamics and well-intentioned advisers than policecommunity engagement in Somaliland. Yet the initial assumptions of EUCAP's advisers were not unrealistic. Somaliland has high rates of access to mobiles, tariffs in its unregulated industry are amongst the lowest in Africa (Budde.com.au 2015), its government promotes ICT-based solutions to the management of criminal records, and anecdotal accounts of people's willingness to 'tell the government' (i.e. ring the police) about crime-related issues implied that police-community communications were relatively good.
Although there are no open-source analyses of police-community relations in Hargeisa available, a combination of NGO surveys and informal responses from Macalin Haruun residents suggest additional reasons as to why EUCAP considered the project plausible. Admittedly, there were reports of officers demanding payment for responding to crime and stations being used as detention centres (Human Rights Centre 2015), but these were offset by anecdotal evidence of police treating Macalin Haruun's residents politely, and by the generally positive attitudes towards police found elsewhere in Somaliland by the Hargeisa-based Observatory on Conflict and Violence Prevention (OCVP). OCVP has yet to address security perspectives in Hargeisa but its work in towns such as Buroa, 178 km to the east of Hargeisa, is suggestive of what might be found. 5 Admittedly, OCVP's reports offer surveys, rather than analysis. Also, respondents probably tell its UN-funded researchers what they think the UN wants to hear, and allowance must be made for this. Nevertheless, the resultant picture is probably not misleading: Buroa's respondents say that the police is, ideally, their main security provider even if its role is actually supplemented by the activities of informal groups such as security committees and night guards (OCVP 2015a: xii, 26). Almost all are aware of the location of the town's police stations and the time it takes to walk to them (this is used as an indicator of people's awareness of state provision), and almost all prefer to report civil disputes and petty and serious crime to the police rather than to elders. Further, respondents stress that communities should support their under-resourced police by taking responsibility for their own security and giving relevant information to the police (OCVP 2015a: 23). These sentiments, combined with the widespread access to mobile phones and, critically, EUCAP's need to be seen to act, encouraged its advisers to believe that a text alert system could be introduced successfully, especially when based at a model station such as Macalin Haruun.

Model police stations
One of the strategies donors use to introduce change into police-community relations involves building or renovating a police station according to international design principles and operating procedures. Such stations emphasise service provision, with designated places for public access, weapons storage, a women and children's desk, and separate male and female cells and lavatories. They do not incorporate ICT, but in practice this does not matter because local residents rarely share the objectives promoted in the name of such stations (gender equality, empowerment and protection for the vulnerable are cases in point) and this, combined with budgetary, organisational and political constraints means that the expense of such stations is too high to be sustainable, let alone replicable across the country concerned (Independent Commission for Aid Impact 2015: 26).
This was the fate of the model station opened by the UNDP in Macalin Haruun in 2012. Designed to strengthen community participation in policing and 'fostering partnership' via community contact groups involving elders, women, youths, NGOs and businesses, its 70 officers were deployed to 'provide security services for 30,000 people in local communities' (UNDP 2012). Officers were to collaborate with communities in order to identify security issues while communities were to support officers in responding to insecurity. In the event, the project was never rolled out across Hargeisa and there is no evidence that it achieved its goals.

Mobile phone usage
Although there were no significant developments in police-community relations in the three years that followed the UNDP initiative, access to mobiles increased dramatically and by 2015 ICT had a marked impact on, for example, money transfers and communications between Hargeisa and the diaspora in the US, Scandinavia and EU member states. Its failure to affect police provision is consequently striking, especially when data from the World Bank and Gallup cite cell phone ownership in Somaliland at 70%; that is, on a par with Kenya and well above the regional median (Gallup 2016; World Bank 2016). In practice, access is even higher because mobiles are often shared and it is possible to subscribe to mobile services without buying a phone; many people buy a pre-paid SIM card, which they use in other people's mobiles (James and Versteeg 2007; Adam 2010). More significantly, the money transfer and telecommunications industries have used ICT to bridge the country's governance gap, exploiting mobiles to leap-frog the limited number of landlines, banks and roads, and there is no obvious technological reason why ICT could not help to mitigate Hargeisa's poor-quality police response. There are, however, functional and cultural reasons. Whether police or residents bear the primary responsibility for shaping policecommunity engagement is debatable though officers' role as state representatives suggests that their attitudes are key.

Somaliland Police Force and the status quo
Regardless of its technological resources, the key organisation involved in operationalising the project, the Somaliland Police Force (SLPF), has a chequered history. On the one hand, the combination of a predominantly Isaaq population (the Isaaq are one of the main Somali clans), resilient customary law, active civil society, a relatively peaceful capital city and orderly presidential elections has helped to ensure that Somaliland has the most developed police system in the former Somalia. The number of officers based in Hargeisa is unknown though there are approximately 6,800 established officers throughout the country (Hills 2014c: 97).
Although a significant number are unfit 6 or, like a high percentage of the population, illiterate, most have received a basic introduction to rights-based policing, the Cairo Declaration on Islam and Human Rights, and the Somaliland police charter and constitution. Further, the more educated are aware of international practices and procedures even as they filter them through local interests and dispositions. Indeed, balancing the demands and resources of modern and traditional institutions and technologies against the pressures affecting Somali society is a key factor in developing legitimate forms of police-community engagement (Harris and Foresti 2011). Increasingly, the SLPF must also accommodate people's familiarity with ICT. Many officers in Hargeisa have little or no interest in ICT but it is clear from personal conversations with senior officers and recent recruits in December 2015 (most notably in the Immigration Police) that some embrace ICT as a tool for addressing issues ranging from street crime and uncontrolled migration to combatting groups like al-Shabaab and ISIS (Interview 2015a). Nonetheless, there is no evidence to suggest that ICT can address the SLPF's more immediate challenges of inadequate resources and personnel shortages, both of which impact on the low-ranking general duties officers working in districts such as Macalin Haruun.
Whether police-community engagement is unsatisfactory as far as officers are concerned is arguable. On the one hand, internal displacement from droughtaffected areas has exacerbated the fragmentation of clan cohesiveness formerly found in many districts in Hargeisa so police work is less predictable and the notion of clan-based community is less meaningful. Yet the absence of significant change in officers' approaches to people in the vicinity of Macalin Haruun -and vice versa -over the last two or three years suggests that the situation is in some way acceptable or tolerable to both sides even if it remains unclear as to whether this due to Hills: Is There Anybody There? Police, Communities and Communications Technology in Hargeisa Art. 2, page 9 of 16 apathy or a lack of engagement rather than satisfaction with the police. Donors may argue that officers' jobs are safer and easier when they work in partnership with local people, identifying and solving problems collaboratively and responding to incidents quickly and efficiently, but in practice most officers spend their day in station compounds because there is no occupational culture of response, let alone of partnership or service. Admittedly, culture is a blanket concept but it is clear that access to ICT will not change such attitudes. The SLPF has had no history of engagement with the general population since the 1960s and is not yet fully civilianised; there is no culture of recording or reporting, the reformist police bill of 2012 is still on the statute book as awaiting formal confirmation (Somaliland Law 2014), and there is no evidence of police -or politicians or inhabitants -looking for fundamental change. Overall, low ranking officers seem relatively content with the status quo; their job may be of low status but they have uniforms (i.e., free clothes) and while they may not get paid much -or regularly -neither do they need to work hard or protect their job against political interference in the way that senior and chief officers do. The picture emerging suggests that police behaviour conforms to local expectations and requirements, and ICT plays no part in this. Nevertheless, the reasons why the text alert system failed add depth to the picture. It helps to throw light on the nature of police-community relations and people's everyday security strategies in a relatively safe urban environment.

Why the Text Alert System Failed
Based on the premise that closer policecommunity engagement is desirable, Macalin Haruun's text alert system depends on two elements: the willingness of the public to contact officers via mobile messages and officers' willingness to answer the call, log it, verify it as legitimate, and, critically, respond by deploying to the scene. Officers also need the ability and motivation to cascade alerts.
Comparable considerations apply to voicebased systems. The text alert system is a simple robust system that should easily handle several thousand messages. The station is open 24-hours, seven days a week, and taking messages should not be a problem because 16 officers have been trained, with seven men and seven women covering the three shifts usual in small stations. The system is basic but reliable, which matters when only 1-2% of the population in Somaliland is thought to have an IT connection; it is built on a SIM box with cards that is connected to a laptop computer that creates groups and sends messages. Actions are logged on police mobiles and in a logbook, keyed into the system manually, and a message is sent over the computer system (Somaliland Nation News 2016). In theory, this should take about 30 seconds but in reality, the process is haphazard, illustrating the obstacles in the way of exploiting ICT.
The officer taking the call needs to take full details of the complaint or information before a more senior officer decides on its seriousness and the appropriate response. But not all officers have the necessary keyboard skills, senior officers are not always available, the information is lost if the phone is mislaid and there is a high incidence of hoax calls. Issues of confidentiality are a potential concern for both the MoI and EUCAP, as is data protection and the ways in which information is to be used in the courts. Also, although the information received sometimes results in police intervention, this does not happen systematically. It did not, for example, prompt police to break up a fight at a graveyard on the outskirts of Hargeisa early on 8 December even though a resident had rung to warn the police that trouble was imminent (Interview 2015b).
Explanations for the text alert system's failure include Somaliland's weak civil society, the unwillingness of President Silanyo's government to encourage community engagement, and failure on the part of the Somaliland authorities and EUCAP to provide Hills: Is There Anybody There? Police, Communities and Communications Technology in Hargeisa Art. 2, page 10 of 16 the resources and commitment needed for its success and sustainability. Practical reasons for people not using text alerts include the lack of an emergency response number (the MoI reserves 100 for crisis calls) and the complications created by the independence of Somaliland's main telecoms providers, Telecom, Somtel and Nationlink; some mobiles are accessed by three or four SIM cards. The cost of messages is a consideration, too, for while texts are free for officers, who do not pay for responding to or verifying a call, the public must pay. Also, the SIM card used must be pre-paid or in credit. And legacy issues intrude. The text alert system is an element within a broad approach to police-community engagement and can be described as community policing, but such policing reminds many Somalilanders of former president Barre's repressive system of local councils in the 1970s and 80s in which community policing was called 'hamuunta' or ' directing the people' (that is, connecting people to the state). Such policing was used to manage groups or clans that Barre saw as a threat and was linked to the military in terms of its monitoring and reporting mechanisms and community control methods. But the fundamental reason for the failure of the text alert system and, more importantly, the low take-up of ICT for communicating with the police may be local realities and preferences. Support for this explanation comes from the fact that Hargeisa's informal policing providers like to use mobiles to cascade verbal alerts, but not texts. And Somali realities mean that some form of supplementary community or informal areabased security is essential, especially at night, with speed of response the main criteria by which it is judged. Even residents living near Macalin Haruun's station value community groups or neighbourhood watches because they provide security more quickly than the police. The reasons for this include the groups being integral parts of their communities in a way that police are not; the youths, women, elders and businessmen contributing to the groups are known throughout the districts in which they operate and people trust them. As a respondent from Macalin Haruun observed in March 2016, 'guard men' minimise theft and are sometimes able to return stolen items. But not everyone is convinced and other respondents argue that informal providers do not necessarily offer better security; they are rag-tag groups without formal offices or contact points, which makes them inaccessible, and they cannot afford to pay for mobiles or transport. Others are concerned that the groups are managed by the state for its own purpose or may be infiltrated by al-Shabaab or ex-criminals. But for most it is an acceptable solution; Macalin Haruun may be open 24 hours a day but no one expects its officers to respond quickly, least of all at night. For now, mobile messages, verbal and text, play a minimal role in the everyday business of both the SLPF and the city's neighbourhood watches.

What local people really think
The text alert system failed to achieve its objective of receiving and cascading information because local people did not use it. The reasons for this have yet to receive systematic attention from EUCAP and the MoI but a partial explanation can be deduced from focus groups carried out with a demographically and socially representative range of residents from Macalin Haruun and five neighbouring districts in March 2016. 7 A team of ten Somali researchers from Transparency Solutions, a Hargeisa-based development consultancy, each conducted 18 interviews over the course of a week, with two or three interviews completed each day. Some of the researchers came from the area, which gave them easy access to potential participants, while the inclusion of male and female researchers meant that the team was able to reach men, women and young people. Drawing on their experience of working in Hargeisa, the team organised discussions with 180 volunteers from six districts around Macalin Haruun station. Fifty were interviewed in Macalin Haruun, 28 in Mahmid Haibe, 28 in Ahmed Dhagah (a separate enclave that became anti-government in the aftermath of shootings in 2012), 28 in 26-June on the other side of the main road to Macalin Haruun, 25 in Ibraahim Koodbuur, a district containing a well-known IDP camp, and 21 in Ga'an Libah. Of the 180 respondents, 96 (53.3 per cent) were male and 84 (46.6 per cent) female. Eighty-two (45.5 per cent) were married, 81 (45 per cent) single, eight (4.4 per cent) divorced and seven (3.8 per cent) widowed (the status of the remaining respondents was unknown). Twenty nine (16.1 per cent) were educated at madrassa, 13 (7.2 per cent) were educated to primary school level, 25 (14 per cent) to intermediate level and 43 (24 per cent) attended secondary school, while 31 (17.2 per cent) were educated at tertiary level and 11 (6.1 per cent) were self schooled. Twenty-three (12.7 per cent) were illiterate.
Respondents were asked if they had contacted the police and, if so, where, when and why. Those who had were asked if they had used mobiles, how they had been treated, and whether they would contact the police in future. Ninety said that they had contacted the police in the year before the text alert system was introduced, and 25 in the period since. But it is not clear why they contacted the police because 160 said that they did not report a crime. Although 140 said that the police treated them politely (25 said they did not), 174 said that they would go back to the police, with nine answering 'maybe'.
When asked how people normally communicate with the police, two-thirds of those questioned said that the youths, women, elders and businessmen involved in community-based groups visit their local station, which is the recognised site for engagement. They prefer to speak to officers face-to-face because only then can they develop or reinforce the personal relationship needed for an officer to respond. Significantly, 35 said that they had rung their local station using a mobile but the police had failed to respond to their calls. No one used the text alert system. Overall, respondents appear surprisingly tolerant of the gap between what is possible and what actually happens. When asked how the use of mobiles might improve security, 167 of those questioned agreed that mobiles allowed for information to be spread quickly and police to call for back-up from units away from the station, but none referred to ICT unprompted. ICT plays little if any part in their expectations or preferred form of policing.
Everyone is aware of the police's inadequate resources and flaws, and the implications of this were explored by asking how they would like to see the SLPF develop and what kind of police they would like their children to meet. Most (91) thought in terms of resources, stating that police should get more financial support, equipment and stations; 45 argued in favour of an improvement in officers' status in the community while 56 hope to see police reach the standards seen in developed countries. Respondents in Macalin Haruun said police would respond to crime more quickly and effectively if they received more equipment (21), a salary increase (13), education (five), better treatment within the police (two), stopped chewing the mild narcotic qat (two), and followed the law properly (three). But no one mentioned toll-free lines for text alerts, crime prevention or rape, though 28 stressed the desirability of a direct contact line for emergencies and improved communications tools more generally. The ideal for all respondents, and what they hoped their children would encounter, is honest police who perform their tasks quickly and do not harm civilians.
The foundation of police-community relations was addressed using questions about police work and whether other groups provide security more quickly. All respondents agreed that the police's job includes securing peace and stability, from the village to national level, and providing a quick respond when insecurity threatens. Significantly, almost all respondents said that local people should help the police to do their jobs more effectively though none explained how this might be achieved. But an idea of what the police's role is thought to involve is evident from the explanations given by a self-selecting group of 18 respondents from across the six districts. Nine described the police as a tool for punishing criminals and 'the guilty', five emphasised the police's responsibility for ensuring rights and property, while two said the police exists solely to hurt, arrest and restrain people. Two said that they were not aware of what police do.
Perhaps the most surprising result came in response to questions about the text alert project. When asked, only five of the 142 respondents involved in the discussion had heard of the project, even though 47 (38 per cent) had friends or relatives in the police and 74 (51 per cent) had been to Macalin Haruun station. The reasons for this are unclear, not least because the project's launch in August 2016 was marked by a highprofile ministerial visit and multiple TV and news reports, but it probably owes much to the filtering effects of local preferences and Somalilanders' experience of governance and state-society relations. People with access to mobiles or SIM cards are comfortable ringing the police, especially when they have been the victim of robbers or wish to 'tell the government' about a potential incident (the commonly used phrase is telling), but they do not use text messaging. This may reflect literacy levels, the strength of oral culture or distrust of the Silanyo government's security forces, or it may be no more than an acknowledgement of the police's inability to respond quickly.
Macalin Haruun's value here is that it offers insights into the everyday choices shaping low-level policing in one of Hargeisa's more secure districts. The composition of the focus groups also provides a snapshot of the social environment in which the SLPF conducts its business, representing a crosssection of the age, education and marital status of residents found in the district and surrounding neighbourhoods. Most people are satisfied with the police and wish to support officers in their job, arranging their lives to accommodate the police's limited resources and reactive role. Most recogniseand use -the station as their preferred site for engagement, with many having visited a station in the preceding year even if few used their visit to report crime. Indeed, few claimed to report crime to police, with elders playing a mediating role between the two. With the exception of low-key collaborative efforts to ensure that some form of nighttime policing is available, neither residents nor police seek to change current patterns of engagement; neither regards ICT as necessary or especially desirable. In other words, the ways in which residents respond to police helps to reproduce the current relationship.

Conclusions
Macalin Haruun's experience of the text alert system offers insight into the potential connection between ICT and policecommunity engagement. It illustrates the everyday choices shaping low-level policing in one of Hargeisa's safer districts while showing how local norms and preferences can negate the availability of a globalised technology. People may have access to mobiles but they choose not to send text alert messages even though officers have (in theory) the manpower, mobiles, radios, vehicles and training needed to respond to day-time calls. Meanwhile the SLPF's response to both ICT and community engagement is casual. In December 2015, one senior officer said that the system worked well in New Hargeisa though not necessarily elsewhere in the city, whereas his colleague said that the text alert system did not work anywhere because people did not understand it (Interview 2015b).
The reasons for this are debatable, but education and publicity campaigns are unlikely to make a significant difference to the assessment of either the SLPF or residents because officers already have the knowledge