Remembrance of Things Past: Somali Roads to Police Development

as much by legacy issues as by international standards filtered through local norms. This article uses the experience of Somalia’s three regional police forces to reconsider the relationship between past and present projects to build police authority and capacity, and what this says about institutional memory in the absence of documentation. In Somalia, as in other clan or tribal-based societies, police development is influenced by a blend of security levels, political imperatives, pragmatism, international resources and memories of past practices, with group experience playing a more significant role than institutional memory. The only identifiable general principle is the need for political settlements and tactical flexibility – that is, for stability.


Introduction
On 20 December 2013, Mogadishu's Somali Police Force (SPF) celebrated the seventieth anniversary of the founding of its namesake (Keydmedia 2013). Originally established in 1943 as the Police Corps of Somalia, it changed its name to the SPF in 1960 when it joined with the British Somaliland Scouts to form the national force of the newly independent Somalia. Fragmenting in 1991 when former president Siad Barre fled and two decades of civil war began, today's SPF re-emerged in the late 2000s. It celebrated its anniversary for the first time in December 2013.
Unlike its predecessor, the current SPF is confined to Mogadishu, a coastal city of some three million inhabitants. Furthermore, its claim to be the successor of the postindependence force rests primarily on the international support and protection that it and Mogadishu's government receive. Even so, the public police forces operating in Somalia's other main administrative regions -Somaliland in the north-west and Puntland in the north-east -accept the SPF's existence and inheritance. This suggests that, while political settlements and international resources play a significant role in Somali police development, so do memories of policing in the 30 years between independence and the wars of the 1990s. This article uses Somalia's experience to consider the relationship between past and present in areas of limited statehood with low literacy rates, and what this tells us about the role of institutional memory in the absence of documentation.
Police reform is thought to require a police force to break with its past. This is notably so in the aftermath of conflict or regime change. In practice, however, most police forces are selectively reconstituted, and their development is influenced as much by legacy issues as by international standards filtered through local norms. This article uses the experience of Somalia's three regional police forces to reconsider the relationship between past and present projects to build police authority and capacity, and what this says about institutional memory in the absence of documentation. In Somalia, as in other clan or tribal-based societies, police development is influenced by a blend of security levels, political imperatives, pragmatism, international resources and memories of past practices, with group experience playing a more significant role than institutional memory. The only identifiable general principle is the need for political settlements and tactical flexibility -that is, for stability.
These issues deserve attention because the principles on which a police is best reconstructed or reformed and its authority established are unclear. Police culture is notoriously resistant to change (Hills 2012) yet police development continues to be understood as a reform project designed to break with the past (e.g. Bayley 2006;OECD DAC 2007;Downie 2013). This is understandable in that the brutal or corrupt policing found in many fragile environments is often cited as a cause of instability and conflict. Nevertheless, most development projects aim for fundamental change, with police reform regarded as a means for social engineering. Take, for example, the ambitious and futureoriented goals of the UNDP's civilian police project for Somalia, which is one of the poorest and most fragmented countries in the world. Established in 2002 as part of a Rule of Law and Security in Somalia (ROLS) programme, with an annual budget of US$ 20 million in 2012, the project focuses on providing a professional civilian police service. It aims to meet the needs and expectations of all Somalis, with a special emphasis on inclusive and accountable policing. This will enable Somalia to 'make progress towards peace and the Millennium Development Goals through equitable economic development' (UNSAS 2010).
Police development projects also assume that the officers concerned depend on a police institution for support, opportunity and knowledge. It is thought that police as a group share an institutional memory; i.e. that an identifiable set of facts, concepts, experiences and knowledge exists. But what happens to the organisation and activity of policing when the institutions, processes, influences and officers on which it is seemingly predicated for existence fragment, and stay fragmented for 20 years? Somalia represents a textbook case in which to explore these issues.
Admittedly, Somalia's experience is extreme, and the lessons it offers may not transfer to other clan or tribal-based societies, such as Yemen or Afghanistan, let alone to other states like Guinea-Bissau that are usually classified as failed. Yet it offers insights into the interplay between memories of policing and the experiences and skills that shape the re-emergence of police in the aftermath of conflict. Two are noteworthy. First, it suggests that police development is an adaptive process reflecting a mix of contingencies, political imperatives, pragmatism and functional knowledge, and that group and individual memories play a key part in this. Indeed, this is markedly so in societies, such as Somalia's, that value genealogy and group relationships. It could even be that clan memory, which transcends the individual, plays a role similar to institutional memory. 1 Second, re-emergent forces such as the SPF are best understood as reflecting layers of functional and political knowledge and experience. 'Layers' is used in a descriptive, rather than a scientific manner, but it conveys the way in which police typically integrate technical experience and local norms and practices into a meaningful framework.
The discussion that follows is divided into three sections. The first offers an overview of Somalia's three public police forces before analysing their development in terms of a layered knowledge base, rather than an institutional memory. The second section identifies security and stability as critical variables influencing the re-emergence of police forces, with tactical flexibility the only identifiable general principle. The third section concludes that policing is selectively reconstituted, rather than fundamentally restructured or reformed. In Somalia, as elsewhere, the success or failure of police development projects is determined by the political skill with which local realities, contingencies and international expectations are balanced.

Layering Knowledge
Somalia is commonly regarded as the paradigmatic failed state (Fund for Peace 2013) yet it is actually managed by a variety of security and administrative entities and authorities. These are linked ethnically and economically but have different levels of stability and styles of governance. (Governance here refers to the rules, processes and interactions through which decisions are made and authority exercised). The international agenda for Somalia focuses on Mogadishu as the capital of a unified state, but in practice inter-governmental organisations (IGOs), bilaterals and donors divide Somalia into three main administrative areas: Somaliland, Puntland and Mogadishu (see UNDFS 2011). South-Central Somalia is best categorised as a remainder rather than an entity or region; its cross-hatching of clans obstructs the development of an entity with a common governance structure. This division reflects that Somaliland declared itself an independent republic in 1991, Puntland claimed autonomy in 1998, while the international community supports a government confined to Mogadishu. And each of the three main entities has a distinct public police force. 2 The nature of Somalia's complex and dynamic environment means that progress in police development depends on political settlements, which require relative stability. This has been a positive experience in Somaliland where, in 2011, the director general of the Ministry of the Interior described security and stability as the country's most significant achievement (Interview, Director General, Hargeisa, 7 September 2011). The development of Puntland's police has followed political settlements, too, albeit to a lesser extent. But the process has been much less successful in Mogadishu. Somaliland's police structure reflects the skill with which its power brokers have balanced political calculations, clan relations and modern and traditional institutions (Harris and Foresti 2011; Renders 2013), whereas the SPF's rudimentary system reflects the precarious position of Mogadishu's Federal Government of Somalia (FGS), on whose behalf it acts. Although the US formally recognised the FGS as the national government of Somalia in January 2013 (the IMF did the same in April 2013 and the UK opened an embassy in May), the existence of the FGS and SPF actually depends on the presence of thousands of African Union troops and on financial and humanitarian support from the United Nations and donors such as Japan.
The Somali environment also means that officers in all three forces lack the appearance and attributes commonly associated with police and police forces. Not all have uniforms, not all consider conventional forms of discipline and hierarchy (e.g. ranks) desirable or necessary, and the social acceptability and addictiveness of chewing qat, a mild narcotic, means that many stations are closed in the afternoon (Hills 2014: 97-98). Also, as classes at Somalia's regional training schools emphasise (Personal observations, Mandera, September 2011; private communications, international adviser, August 2011), many general-duties officers are unfit, elderly, illiterate or unable to understand the basic principles of international-style policing. Yet, this does not mean that they are not police or that ' a police force in the Western sense is...an alien institution' (Murphy 2011: 156; for Western policing see Manning 2011). Regardless of their appearance, or ability to read or provide order, Somalia's forces share occupational commonalities with police in other regions, are linked to international policing networks, 3 are aware of the internationally recognised policing that Somalia experienced during independence, and are in some cases (primarily in Somaliland) consciously developing the complex of practices, procedures and norms required for the capacity-building through which institution-building is thought to emerge and be expressed.
In other words, Somali officers are police, and they belong to police forces even if those forces have little in common with those described by, for example, Deflem, for whom 'police bureaucracies are hierarchicallyordered with a clear chain of command... and policework follows set rules and procedures...' (Deflem 2000: 744;Deflem 2002: 18). More importantly, all three forces have enough in common with police elsewhere in the world to offer a baseline for exploring the ways in which police authority is constituted and institutional memory developed in societies subject to chronic insecurity and high levels of illiteracy. Indeed, Somalia is a particularly valuable case with which to discuss these issues because its three forces cover a policing spectrum that spans an under-resourced SPF that is expected to provide civilian policing in the middle of an insurgency and humanitarian emergencies, Somaliland's relatively developed system, and Puntland's emergent force which is positioned between the two.
Somalia's police forces are also surprisingly conventional. They are structured on lines common throughout Africa and are familiar with international practices and procedures even as they filter them through local interests and dispositions. Admittedly, this assessment conflates diametrically opposed meanings of ' conventional' (of police relying on local practices while complying with Western structures), but it reflects the ways in which Somali police, like police around the world, respond to imported models or new practices in an adaptive manner, integrating aspects of international understanding with local realities and personal or group experience. It reflects the ways in which officers adjust to local realities (Menkhaus 2006) and international pressures (Höehne 2009: 253).
The best way to understand this situation is to see it as layers of knowledge, with the layers consisting of legacy issues, international influence, and local norms and practices.

Legacy issues
Since approximately 2008, when the UNDP began keeping records of officers receiving training, international notions such as rights-based policing and police-community partnerships have been promoted as a means to mitigate Somali policing's flaws. However, while Somali policing is typically minimal, predatory or oriented towards counter-insurgency and counter-terrorism, many older Somalis know something about the history of Somali policing in the decade after independence in 1960. This was when the SPF was created as a civilian police capable of counter-balancing the Soviet-supported army (Barre had been the Somali National Police's commissioner during the last decade of pre-independence). Anecdotal evidence suggests that some of the officers operating during those years felt a professional pride in their job, displaying willingness and self-reliance. In 1959, for instance, on discovering that only three of the 30 officers in a small town in today's Puntland were literate, a senior officer brought in teachers to make the remaining 27 literate in Italian (Interview, NGO official, Garowe, 14 September 2011).
More importantly, many of today's officers and officials are conscious of both Somalia's internationally respected policing during the 1960s (Perito 2002: 30) and their need to avoid the repressive practices of the Barre era of the 1970s and 80s, which many associate with centralised policing. Indeed, conversations with senior officers suggest that officers and officials in all three forces build on memories of independent Somalia's 30 years of national policing even as they reject Barre's brutality. Rather than trying to shed its colonial-era identity, Somaliland officers and officials say the legacy is a useful reference point (Personal conversations, Hargeisa, 7 September 2011). 4 This is understandable given that the police commissioner (chief officer) in post in Somaliland in the autumn of 2011 enlisted in 1942, and the commandant of Somaliland's police training academy at Mandera joined in 1958. There is no welfare provision so officers have no incentive to retire.
The extent to which an identifiable institutional memory developed during this period is debatable. On the one hand, the SPF's institutional base was initially strong. It was organised into northern and southern group commands, divisional commands (corresponding to the districts that existed then), station commands and police posts, with regional governors and district commissioners commanding regional and district police elements. By the mid 1970s, its approximately 15,000 officers (which included 6,000 from a clan-based militia called the Daraawishta) were divided into 18 regional districts, each with about 90 stations and 100 posts in smaller towns. Its officers carried out patrols, traffic duties, criminal investigation, as well as intelligence and counter-insurgency while the Daraawishta and a riot group (Birmadka Booliska) acted as elite mobile units capable of stopping clan conflict in remote areas and along the frontier. 5 It was overseen by the Ministry of Interior until 1976, at which point it came under the control of the presidential adviser on security affairs.
On the other hand, Somalia operated under a civil administration for only nine years (Barre came to power in 1969's coup d'état), and kinship and customary law (Xeer) were always more important organising principles for Somalia's pastoral-based society than formal institutions. Also, Barre distrusted and emasculated institutions such as the SPF, leaving only the National Security Service in a position of influence. Furthermore, by the late 1980s, his authority was confined to the main towns where police played a minimal role; the riots of 1989 were policed by troops with machine guns who cleared the streets by shooting anything that moved.
Somalia has not experienced central policing since 1991, though precisely what happened to the SPF's officers during the 1990s is unclear (for the early 1990s see Ganzglass 1996; Thomas and Spataro 1998;Perito 2002). Anecdotal evidence suggests that a handful of individuals offered a minimal and localised police presence, but most melted into their neighbourhoods and clan groupings. The practices and values associated with bureaucratic institutions were forgotten. As 45-year-old Mohammed Haroum (an SPF traffic policeman who had resumed his job after a two-decade break) said in 2012: 'These days ... people do not know what a drivers' licence is. They have never seen one. No-one has even heard of insurance ... We are learning to be normal again' (BBC 2012;Compare Kentucky.com 2014).
In practice, the foundation on which the new normalcy is built owes more to relative stability, clan calculations, tactical flexibility, pragmatism and memories of what is needed for international recognition than to international policing models and practices. And memories of the internationally-respected policing Somalia experienced during the 1960s matter because they influence Somali ideas about police today. Lund's observation that state-based notions and concepts such as stateness -or, in this case, police -are best understood as an amalgamation of public authority by local institutions ' conjugated with the idea of the state' is indicative (Lund 2006: 685).

International influence
While individuals have different experiences of life during the wars of the 1990s and the entities have differing visions of Somalia, there is a consensus on what police should look like that is reinforced by officers in all three forces having a history of international contact. The original SPF received equipment or training from, amongst others, China, Egypt, the German Democratic Republic, Italy, Sudan, West Germany, the UK and the USA. Additionally, Barre's advocacy of Marxism embedded certain Soviet approaches, and anecdotal evidence suggests that this still influences the attitudes of some senior officials. 6 Training projects supported or provided by IGOs such as the UNDP, the UN Political Office for Somalia (UNPOS) and the African Union Mission in Somalia (AMISOM) add another layer to Somali policing knowledge, and, while training projects rarely affect the way officers conduct their everyday business, they offer individuals a window onto international ideals and modern or 'professional' skills. Thus, the UNDP's civil policing project in Somaliland provides a comprehensive package of activities that includes not only basic recruit training and specialist training in criminal investigation, critical incident management and headquarters functions (such as financial management and logistics), but also offers courses in rights-based partnership policing, the Cairo Declaration on Islam and Human Rights, and the Somaliland police charter and constitution. Between October 2010 and 2011, 2,400 officers were trained at Mandera, while other international activities included the visit to Mogadishu in 2009 of an AMISOM police team with a remit to train, mentor and advise the SPF, mid-level training courses by the Uganda Police Force, and cadet courses held in Uganda and Ethiopia. In November 2010, UNPOS, funded by Japan and in partnership with the AMISOM civilian police unit, began a three-month basic training course in Djibouti for 501 new SPF officers based on a curriculum developed by the UNDP and approved by UNPOS, AMISOM and donors. More recently, a delegation from the SPF and AMISOM went to Sierra Leone for a study visit. Sierra Leone having been identified as a success story that Somalia could learn from (Hiraan Online 2013).

Local norms and practices
Of the three elements identified here, local norms and practices is the most influential on every-day policing and policework. This is because officers are recruited from, and represent, elements of Somali society. They must, for example, integrate traditional Somali approaches into their functional knowledge, because unlike the situation in Liberia for example, the social fabric did not collapse after 1991. Somalis look first to customary justice and local or traditional non-state actors such as elders or Shari'a, rather than to police; officers typically intervene only when requested by elders implementing customary law. 7 Also, officers (and general-duties officers in particular) engage with local elders, mayors and district security committees about issues such as land or water rights, and mayors in towns such as Puntland's Bossaso, Galkayo and Garowe say that land and water rights are the biggest issue they hear about on daily basis. They also settle many cases without recourse to either elders or courts, as when the five female officers attached to Hargeisa's women and children's desk settle informally nine or 10 of the 10-15 cases they receive each month (most involve children). In many cases, officers enlist the support of traditional authorities, but they also use their discretion to deal with minor incidents such as theft from market stalls (Interview, five female officers, Hargeisa, 6 September 2011).
The role of discretion in policing has long been acknowledged (Kelling 1999). However, Somali policing lacks the hierarchical and bureaucratic-based organization that international policing models use to offset discretion. Additionally, negotiation and tactical flexibility are underpinned by the threat of political or physical violence, for these are strong themes in Somali policing, just as they are in Somali society. This is particularly true of SPF officers who may be aware of international policing practices, but must repeatedly negotiate between the contradictory demands of donors' insistence on civilian policing and the federal government's need for counter-insurgency operations. They must also negotiate with a range of alternative policing providers, for security and justice are provided by militia and clan groups loyal to factional leaders, as well as by business men and Shari'a courts. Consequently, combinations of formal and informal security providers are common. In July 2011, for example, security in Mogadishu's Dharkenlay district improved significantly as a result of the combined efforts of an experienced cross-clan militia made up of officers who had served under Barre (the Hillac brigade), youth militias (madani), a Sufi militia (Ahlu-Sunna Wal-Jama, which is theoretically aligned to the government but actually operates autonomously) and the district's police (PAC July 2011).
As this suggests, local norms and practices play a critical role in policing. Somalia is the site of multiple conflicts over territory, trade monopolies, political power and scarce resources, all of which are fought by people with little if any interest in institutions or institutional development. Policing is a commodity and a business opportunity, rather than a community service, and is part of the same political dynamics as clans, conflict, entrepreneurialism and fragmentation (Interviews, officials, Nairobi, September 2011). And Somalis are susceptible to such pressures because clan-based commitments and obligations are the only guarantees most can rely on; security, protection, opportunities for poverty reduction -all depend on clan/group relationships, rather than on governmental directives or institutions (Luling 2006;Gundel 2009; Danish Immigration Service 2013).
Legacy issues, international support and local norms and practices have in this way provided Somali officers with a layered knowledge base that does not rely on institutional memory. Nevertheless, institutional development is now a fact. Significantly, in each police force it depends on political settlements, which in turn require relative security and stability. International advisers may regard Puntland's police as being technically and institutionally five years behind Somaliland's (Puntland was a backwater during Barre's era), but technical and institutional development has been possible in both because their forces were established after political settlements were agreed; their police are seen as legitimate elements in their government's control apparatus. But this is not the case in Mogadishu where the SPF are little more than militiamen in uniforms and the credibility of the government on whose behalf it claims to act is limited (Financial Times 2014). Hence the significance of the SPF's name and claimed lineage -and its utility as a tool for addressing the key contextual issue affecting institutional development in fragile states: security levels. Specifically, the insecurity and instability Mogadishu has experienced since 1991 allows consideration of the issue at the heart of this article: the significance for police development of institutional memory in the absence of written documentation.

A Re-emergent Force
Ironically, it is the SPF which best illustrates what happens to the organization and activity of policing when the institutions, processes and international influences, on which it is seemingly predicated for existence and opportunity, fragment -and remain fragmented for 20 years. Indeed, the SPF's experience adds a new dimension to police studies because a small number of former officers retained personal memories of policing and policeness even after formal state authority collapsed and policing was taken over by militia groups, warlords, traditional elders, Shari'a courts, and businessmen. Explaining this -and identifying its analytical and empirical implications -is, however, complicated by a lack of hard evidence. Not only is there minimal documentation available from Somali sources, but also there is little between the articles and reports written by advisers in the early 1990s (Ganzglass 1996;Oakley et al. 1998) and the records created by the UNDP's civilian policing project since 2008 -and record keeping remains challenging even for the purpose of paying today's stipends.
What currently exists reflects a combination of relative security and stability, and the influence of international advice and standards filtered through local norms and imperatives, rather than institutional records; the occurrence books recording events in stations are rarely kept up-to-date (Personal communications, Nairobi, September 2011). It also suggests that despite several decades of conflict, memories of the SPF as a centralised command organisation has helped to ensure that a baseline for public policing exists. Mogadishu remains a dangerously insecure city but the speed with which the SPF has developed over the last four or five years suggests that the force has re-emerged, rather than been created.

Coping with insecurity
In June 2013, the Islamist insurgent group al-Shabaab attacked a UN compound just outside Mogadishu's airport where thousands of African Union troops are based, killing 15, and in August 2013, Médecins Sans Frontières (MSF) pulled out of Somalia because the threat of violence had become intolerable: 'We have reached our limit' (Al Jazeera 2013). Elsewhere, suicide bombings, assassinations, mines, IEDs, stray bullets and hitand-run attacks erode traditional practices and values. Insecurity is fuelled by fears that veiled women bus-passengers and students with rucksacks could be suicide bombers, and is exacerbated by clan and community tensions, freelance militias, and disputes about land, livestock and water (especially in IDP camps). 8 There are no reliable statistics covering crime and violence in the city, but Somalia's dozens of (almost real-time) news services speak of high rates of theft, forced detention, sexual violence and gun-related murders (compare OCVP 2011: 2).
Unsurprisingly, many Somalis see such crimes as components of broader conflict dynamics, rather than as individual events (OCVP 2011: 21). Somalia receives weapons from Ethiopia, Yemen, Djibouti, Ukraine, Libya, Saudi Arabia, Eritrea and the USA, and it is thought that more high-powered weapons are in circulation now than in the 2000s. This, combined with memories of Barre's brutal policing, meddling by frontline states such as Ethiopia, and the FGS's weakness, destroys respect for state institutions. Somali politicians and power brokers think in terms of individual or clan interests, government authorities are unresponsive to people's needs, and social factors play out in the political sphere (OCVP 2011: 36). It makes for high levels of suspicion (Somalia Report 2012a), which undermines even minimal forms of institutions and governance. It also makes for hopelessness. Mogadishu is currently experiencing a building boom (Yahoo News 2013), but many share the bleak outlook of participants in OCVP focus groups in 2011. Participants spoke of a 'lack of faith in the ability of the community (let alone the state) to protect individuals, a sense of injustice, trauma, and general hopelessness for the future' while youth focus groups identified 'fear of the unknown' as a prompt for joining armed groups (OCVP 2011: 35).
Despite this (or perhaps because of it), a range of resources for coping with the consequences of insecurity and social fragmentation exists, with the most successful predicated on local ties, knowledge and informal responses (compare Renders 2012). This approach is reinforced by Somalis prioritising reconciliation and reparation over retribution and punishment, and preferring informal alternatives to formal sentencing and detention facilities. Customary law is more powerful than formal institutions and even governments use elders to solve, manage and negotiate issues; government (secular) law, and its associated institutions, is the weakest of all. Co-operative arrangements exist, too, as when militias loyal to district commissioners ensure that people returning to their neighbourhoods register at a police station, while police accommodate civil society initiatives providing surveillance and reporting: 'they tell us'. They must also accommodate militia operating under agreements with the government and AMISOM. In other words, police development cannot be discussed in isolation from broader trends.

Tactical flexibility
The long-term consequences of fragmentation for the police's state-based authority are more limited than international policing models suggest. It is undeniable that many SPF officers are little more than militiamen in police uniforms, yet this does not mean that they are not police any more than being former rebels precludes South Sudanese from being policemen. 9 Indeed, Mogadishu's environment makes this understandable.
The SPF's technical capacity has declined, and its operating procedures and discipline are minimal, but memories of how policing was organised before 1991 have evidently been influential. It managed to keep between four and six stations open throughout the war, and by 2011 had a (semi-functioning) headquarters and training academy. Organised conventionally in directorates of operations, finance, training, Criminal Investigation Department (CID), and administration, its activities range from counter-insurgency to traffic management and criminal investigation. Dharkeynley district may be managed by militia but even in April 2011 there were 30 uniformed officers present on the streets and in local markets (PAC April 2011).
It is true that this picture owes much to international resources, which are intended to reinforce the appearance of the SPF as an institution. For example, most of the city's officers are dressed in light blue shirts and dark blue trousers or in military-style light brown (colour is based mainly on where the officers are trained) even if their pay is delayed and they lack weaponry, handcuffs, batons and the pick-up trucks needed for mobility. Similarly, the SPF's line ministry may be little more than an inexperienced newcomer in an empty office, but approximately 5,300 of its estimated 6,000 'police-soldiers' (the term used by the city's police oversight body, the Police Advisory Committee (PAC)) are given a three-month police training course by the UNDP before being sent out to Mogadishu's operational police stations (the rest are trained in the region under bilateral agreements between the FGS and its partners). Nevertheless, internationals can only exert influence when Somalis allow it (Hills 2014: 97-98), so the situation seen in 2011, for example, indicates Somali views on what a police force should look like: officers should be conventionally organised, maintain a visible presence and provide area security.
By mid-2011, after two decades of conflict, the FGS's predecessor, the Transitional Federal Government (TFG), and its police controlled approximately 98 per cent of Mogadishu, which was divided into 16 districts, each of which had a headquarters and four divisions (there is a station in each). 10 In addition to stations at the airport (70 officers), port (94) and the CID (150), there were stations in Dharkenley (65 officers), Hamar Weyne (91), Hamar Jabab (250), Shangani (40), Waberi (72) and Wadajir (92) (PAC July 2011). Dharkeynley, for instance, had 30 uniformed officers at checkpoints and in markets, while officers in Wadajir manned a checkpoint on the district's main road and patrolled the nearby market and main bus stations armed with AK47s, talking to local people in a 'friendly' manner (PAC April 2011). Reports by the PAC noted that in April 2011 security was tight at Wadajir station (several inmates had recently tried to escape), but hygiene in the cells was generally good, with water and electricity available, and inmates separated by sex, age and crime (unusual for Somalia).
The range of activities typically undertaken by SPF officers can be seen from the PAC's Monthly Report on the Activities of the Somali Police Force for July 2011. (It is consistent with that offered in 2013 by international assessments such as those provided by the Danish Immigration Service). Written primarily for the UNDP/ROLS, and, therefore, arguably presenting an ideal picture, the PAC report notes that the SPF's workload is organized around divisions and stations, and is based on information from street patrols and informants and on government directives or orders. Activities are divided into operations intended 'to enhance community safety' and 'street patrolling as part of wider counter terrorism efforts', but most relate to deterrence. Officers manned checkpoints, deployed at major intersections, and conducted random stop and search operations in an effort to increase the police's presence. They targeted individuals or small groups suspected of armed robbery, kidnapping and looting (the latter involved militiamen responsible for providing night-time neighbourhood watches) though non-terrorist prosecutions were rare. No mass arrests were carried out in July but March had seen a major security operation involving 300 people charged with having unlawful weapons and violating security rules (the use of mass arrests is noted in Danish Immigration Service 2013).
The PAC's July report emphasises the SPF's role in counter-terrorism and/or counterinsurgency, rather than civilian policing or formal governance. But this is appropriate given that the SPF is a faction in the continuing conflict in which its lightly-armed officers confront insurgents armed with heavy machine guns and light anti-aircraft artillery. Indeed, PAC members routinely refer to 'policemen-soldiers' targeting 'sleeping insurgents', particularly among IDPs. It noted that Dharkenlay police had had a particularly difficult week, fighting youth militias (armed with hand grenades and handguns) suspected of being both armed robbers and covert insurgents, and arresting seven men (three of whom were military) who had set up illegal check points to extort money. The PAC's members, like respondents to focus groups conducted by the Hargeisabased Observatory on Conflict and Violence Prevention (OCVP), were clear about the distinctions between police and military enforcement agencies, but were less concerned by those between police and militia. And Mogadishu's environment makes this understandable: policing is subject to shifting loyalties, tactical fluidity and clan calculations, and a significant proportion of officers have close links to militia. Younger and fitter officers probably find the swarming tactics employed by militia and fighters (Kilcullen 2013: 81-86) more attractive than the community-based policing promoted by donors.
Whatever the case, the PAC report implies that Somali policing places a premium on flexible and pragmatic approaches that are at odds with international-style policing's reliance on a formal and bureaucratic institution. Discretion is a feature of most police systems, but Somali officers are expected to align their practices to international standards while operating in a rule-based society characterized by the legal pluralism of formal, customary and Shari'a law. 11 This suggests that Somali policing structures are, like Somali social structures, best understood as flexible social and political processes, with the emphasis on accommodation, interpretation and knowledge based on a range of resources (compare Little 2003: 3), all of which is overhung by the threat of actual or potential violence.

Conclusions
Surprisingly little is known about the ways in which the past and present interact to affect police development in areas of limited statehood. International police projects are designed to ensure that local forces break with their past whereas local realities help to ensure continuity. This is notably so in conservative clan-based societies where the absence of welfare arrangements means that there is no incentive for unsuitable officers to retire or reject the political economy they are part of. Tensions arise as the futureoriented agenda of international reform projects collides with the more immediate expectations of local recipients, but the interface is rarely analysed.
Somalia is a particularly interesting case in which to explore these issues because significant number of its officers were members of the original SPF and are now, after 20 years of conflict and fragmentation, part of the re-emergent SPF or the new regional forces. Hard evidence is missing, but anecdotally, the experience of serving officers suggests that police development is influenced by a blend of variables. These include contingencies, international resources, and political and/or entrepreneurial imperatives, all of which are superimposed on technical skills, experience and memories of past practices. The critical issue for IGOs and donors, therefore, concerns identifying the optimum balance -or tipping point -between past and present, and between locally and internationally acceptable norms and practices.
Taking this as its point of departure, this article offers five findings, which are also relevant for police reform in other clan or tribal-based countries: First, police knowledge is layered. Personal experience of policing in the 1960s, an aversion to Barre-era practices, a lack of formal education in the 1990s, or life as a militiaman in the 2000s, combine with local norms and exposure to international approaches to provide Somali officers with a functional knowledge base that is more reliable than institutional memories or formal documentation. This is reinforced by low literacy, respect for genealogy and the resilience common to many clan-based societies.
Second, the Somali approach to policing is tactically flexible because officers operate in a rule-based society characterized by legal pluralism. In most countries policing is characterised by discretion and pragmatism, but Somali policing is, like Somali politics, particularly subject to shifting loyalties, clan calculations and opaque decision-making; pragmatism and informality are more characteristic of its governance than routine or precedent. Officers protect what they value, manipulate what they can use and subvert approaches that offend the sensibilities of their conservative society. This tendency is reinforced by police reform attracting international resources, which make it a business opportunity for senior officers, as for politicians and businessmen. The real challenge for reformers is how to protect the institutional development some now associate with professional and modern policing from the predations of other officers and politicians.
Third, Somalia's record shows that while institutional development (and technical progress) depends on political settlements based on an agreed set of rules, the re-emergence or restructuring of police forces is not contingent on institutional structures or institutional memories. Neither is it dependent on the existence of a formal criminal justice system. Indeed, the last decade emphasises that minimal forms of policing are sustainable in the absence of formal government structures and institutions.
Linked to this is a fourth point, which is that even token forms of institution building and technical development require the broader political-security environment to be relatively stable; formality is not required for police development, but relative security and stability are. Stability has allowed Somaliland to reconstruct a locally acceptable police that combines elements from the past with an eye to the future, whereas Mogadishu's police and government are dependent on the political agenda of external actors. Consequently, the SPF's primary concern is to acquire the heavy weapons, vehicles, fuel and communications equipment it needs to survive today (Interview, SPF commissioner, Nairobi, 19 September 2011. Compare Reuters 2014. Hence the prospects for police development are favourable in Somaliland and, to a lesser extent, Puntland, whereas the SPF's future is uncertain.
Lastly, police institutions may disintegrate and their documents may be destroyed, but police systems are remarkably resilient. Somalia's experience offers insights into the nature of institutional memory in the absence of documentation and line ministries, and into what happens to the organisation and activity of policing when the institutions and processes on which it is seemingly dependent for existence and opportunity fragment. But, general principles are hard to find because the re-emergence of police is driven by a blend of aspirations, opaque decisionmaking, political settlements, memories and, critically, levels of security and stability.