Cross-case patterns of security production in hybrid political orders: their shapes, ordering practices, and paradoxical outcomes

ABSTRACT Examinations of substate security and everyday peace in hybrid political orders are mostly limited to single-case studies or statistical analyses. Seldom are qualitative methods applied with a comparative aim that can unveil patterns of security production. I attempt such an approach by studying 12 cases across the Central African Republic, Haiti, Somaliland, and South Sudan. I investigate (1) where hybrid interactions take place, (2) how they happen and (3) what this means for people’s security. I argue, first, that hybrid ordering shapes socio-geography by separating a rigorously controlled inner from a securitised outer circle. Second, I find that actors clash over the use of contrasting ordering principles on a spectrum from stable to fluid. Third, measured security indices, paradoxically, often diverge from how safe people feel depending on public support for the socio-geographical shape and ordering principles applied. These cross-case patterns of hybrid political orders underscore the importance of comparing political ordering processes.


Introduction
Security studies in the political sciences are a vibrant field. Emphases on geopolitical war 1 have been extended down-and upwards to intrastate and suprastate issues 2 and recently to bottom-up perspectives on daily concerns. 3 Today's empirical security studies enrich themselves by drawing from multiple disciplines and methodologies. 4 Yet, Robin Luckham and Tom Kirk lament that 'existing studies [of security in hybrid political orders] still tend to be geographically scattered, thematically and methodologically diverse, and in many cases lacking in empirical vigour'. 5 Indeed, the production of security is studied comparatively in quantitative analyses, 6 while qualitative scholarly work emphasises context particularities mostly through single-case studies. 7 In this article, I demonstrate the importance of a comparative lens rooted in empirical fieldwork that challenges what we know about hybrid political orders (hereafter HPOs) by unveiling cross-case patterns of security production. HPO literature studying instances 'where the state and its monopoly of violence are contested and diverse state and non-state security actors coexist, collaborate or compete' 8 delineates a strong theoretical framework for analysis, 9 which is then used and refined for single-case studies. 10 However, it lacks an appreciation of which theoretical assumptions reverberate across cases. In this article, I wish to contribute to the debate by comparatively investigating where hybrid interactions shape the interaction space, how they are carried out and what outcomes they produce.
Single-case studies have analysed where hybrid interaction shape space by 'overstretching' realms of authority 11 or forging 'boundaries' between different zones of influence. 12 Through comparison across cases in this article I argue that such delineation attempts are a recurrent pattern seen in the differentiation actors make between a tightly controlled inner circle from a more loosely approached, and often securitised outer circle.
Furthermore, a key deficit of the HPO debate are its, albeit often qualified, primordial assumptions on the difference between the involved actors -be it international versus local or bureaucratised state versus customary non-state -preconceiving their respective ways of interaction. 13 Newer single-case studies, challenge this ascription by pointing out how state and international actors can act beyond legal and bureaucratic bounds or how even non-state actors can create rigid administrative systems. 14 The everyday peace debate has been particularly fruitful in challenging top-down, elite-centred approaches to studying conflict and putting people's daily engagement's in building peace 15 or more recently also creating everyday conflict 16 centre stage. Everyday peace describes 'routinized practices used by individuals and collectives' in places that are 'prone to episodic direct violence in addition to chronic or structural violence'. 17 Taking both elite and all inhabitants' actions together, the literature on ordermaking during conflict encourages scholars to abstract from individual acts or strategic combat decisions. 18 Actors engaged in ordering want to shape the ways in which actions in the future can happen; they strive to give form to the political order, seen 'as a set of predictable behaviours, structured by widely known and accepted rules which govern regular human interactions and behaviours'. 19 Questions of ordering and security are particularly interconnected, because violence transforms order just as much as order is used to tame violence. 20 In this article's mid-N transnational comparison, I argue that the key ordering struggles surrounding security production in HPOs concern actors (of any kind) choosing either fluid ordering, which is marked by the openness of processes and results, versus the more rigid and predictable use of stable ordering. A recurrent pattern across cases was that actors align or clash over the appropriate use of ordering forms.
Finally, in many security studies publications, 'security' is often only vaguely defined 21 or merely narrowed down to the absence of physical threats to the individual or to a societal group. 22 Equally important is the study of people's perceptions of security. This article contributes to a recent turn towards vernacular security and everyday peace, emphasising the 'subaltern perspective' 23 and experiences by "non-elite communities". 24 To limit the scope of this analysis, I mostly look at physical incidents of security but as viewed from respondents' perspectives and thus define security as 'the felt durability of physical integrity and insecurity as the felt threat of physical harm'. 25 Hybridity is often used as a prescriptive policy tool that assumes positive security outcomes through hybridisation particularly in intervention contexts. 26 Somaliland is thereby recurrently referred to as an example of the positive impact HPO can have on security. 27 Through comparison I argue that security levels vary from HPO to HPO and even, paradoxically, within one HPO when we separate subjective and measurable aspects of the security definition mentioned.
Clearly, single-case studies are crucial to the HPO debate and to the theoretical arguments discussed in this article, such as the place of hybrid interactions, the distinct forms they take in each case, and the varying security outcomes HPO produce. The HPO literature can benefit from a methodologically sound discussion of a possible recurrence of observed instances as patterns across cases. To achieve this, the Comparative Area Studies (hereafter, CAS) approach 28 allows scholars to combine area knowledge of each case through elaborate field visits, with the aim of challenging theory-building based on European and North American assumptions -thus gathering comparative empirical observations from across world regions. 29 Mid-N qualitative comparisons face two key methodological challenges: a tight schedule per site visited and gathering comparable data points for each case. A two-step process led to my selection of cases: first, countries with a marked absence of a national monopoly on the use of force and a recent insecurity history, and second, subnational localities with a plurality of security actors, of peri-urban make-up (tertiary towns with around 10 to 50000 inhabitants) and varying on an external force presence. Whereas I attempted to preselect these localities in a non-random fashion, the reality of security studies is that scholars -including myself -have to navigate access constraints for reasons of safety. 30 I moved back and forth between home institutions and varying field-research localities. Between late 2014 and the end of 2016 I conducted a one-month-long field trip to South Sudan (Raja, Buseri, and Mundri) and two two-month stays in the Central African Republic (Paoua, Obo, and Bangassou) and Somaliland (Baligubadle, Zeila, and Daami) each. In 2017, I conducted short follow-up visits to the capitals of the latter two as well as repeated field trips to the Central African Republic (hereafter, CAR) in 2018 and 2019. Finally, a three-month stay in Haiti in late 2018 (focusing on Jérémie, Anse Rouge, and Mirebalais) further cemented my analysis.
In each locality, I used site-intensive methods inspired by an ethnographic approach 31 as the basis of data-collection efforts. Most important were semi-structured interviews with at least a dozen stakeholders per locality that inhabitants described as important to their production of security. When possible, I met these interview partners where security was produced and asked to also observe such processes (e.g. peace committee 26  meetings, court trials, check point crossings). To avoid an elite bias in the study, I conducted multiple focus group discussions with varying parts of the community, emphasising participation by inhabitants underrepresented in security institutions. In most cases I conducted three of these focus group discussions as actor mappings -that is guided discussions on naming local security actors, drawing their interactions and recounting productions of (in)security. Finally, in CAR I also lead a nonrepresentative, random walking sample survey with 240 respondents in four localities. I use anecdotal citations as illustrations of larger trends I found in this breadth of data. Drawing from fieldwork across these 12 localities in 4 countries, this article advances the still-nascent domain of qualitative mid-N transnational comparisons, encouraged by the CAS approach. Jeffrey Sellers argues that the combination of within and between country variation 'can ultimately yield more accurate and reliable conclusions' all the while posing 'numerous additional challenges to theory, testing, and inference'. 32 Transnational comparisons have, to the author's knowledge, been confined to a few cases -small-N -whereas statistical methods usually come in to play beyond four or five cases. What I attempt here is a qualitative mid-N transnational comparison -that is comparing around six to twenty cases across at least three countries solely using qualitative methods. To do so, I thought up my theoretical ideas through recourse to conceptual and empirical HPO literature embedded in reflections stemming from my own recurring fieldtrips, a process others have termed building 'hunches' 33 or 'analytical movements'. 34 I tested these evolving hunches by listening to respondents' insights through the array of methods and spaces mentioned above. 35 Finally, I tell the reader (within this article and others) a select number of stories from the field studies that I believe speak for respondents' experiences, to one another in the form of building patterns in the data, 36 and to the academic debate. What is left wonting in terms of national context or subnational detail, is gained in the form of shared experience beyond national or subnational context. 37 In what follows, I will present the three findings that I discovered by using this third way between quantitative large-N and qualitative single-case studies. First, respondents interpret and security actors shape their interaction space by evoking a distinction between an inner and an outer circle within it. Only by comparing this finding across cases can I show where actors similarly both differentiate and bridge the borders between the two spheres, and how rigid differentiation breeds felt insecurity from ominous threats; strong interrelations lessen this danger ascription meanwhile. Second, I delve further into how people and actors affect security and find a wide spectrum of ordering options from fluid to stable to be available to them. While stable ordering works well for some cases, and likewise fluid for others, in comparing the cases I show how neither of these forms of ordering is per se more prone to insecurity. Security originates from how actors allow varying forms of ordering to cohabitate or clash over them -and less so of the type of actor in charge, or the political and economic interests at stake. Third, security on the ground is not always described as what I expected it to be from my advance information based on statistical databases. Rather paradoxically, objective and subjective security often diverge. Only using measurable data as an indicator for security outcomes thus would do injustice to respondents lived experiences. On the other hand, people's responses do suggest differences in levels of security from one place to the next. I find the positive adherence to the way security is ordered to explain the variation in how secure people felt and how far this aligned with measured security.

Where: inner and outer circle
The distinction between an inner and outer circle in HPOs is something that stood out immediately and repeatedly when entering towns situated in conflict environments. Strong security actors, such as the state military or peacekeeping forces, put up barriers at the town's entrances to physically bar passage to perceived external dangers. Inhabitants often described their town's surroundings as harbouring incalculable threats. These physically and narrated distinctions of an inner and outer circle implicitly shape the methodological approach to studying conflict, further to impacting on the everyday lives of inhabitants of course. Theorising about this distinction can improve what we can and cannot learn about the workings of local security. For instance, to learn how actors connect across multiple levels of different arenas 38 ; or, as discussed here, to grasp on the local level the nature of the relationship between a given town and its surroundings.
The inner circle is 'simultaneously a geographical and a socio-political space of actors that are within reach' whereas the outer circle 'represents the flip side of the inner circle, as it lies beyond the immediate space of observation, forming the projected and imagined world'. 39 Gatekeepers will often depict the inner circle as comparatively more peaceful and 'orderly'. For the researcher, it thus lends itself to the use of multiple methods and indepth discussions. Here, actors are accustomed to talking to foreign non-governmental organisations, state employees, or journalists, and will thus be more familiar with debating their societal roles with a researcher. The individual in question her-or himself can seek out actors similar in their foreignness to the researcher, such as those working with UN missions, international NGOs, or as domestic academics. Within it, a researcher can delve into the ephemeral practices of the everyday, 40 and study the relations between seemingly differing types of actors as hybrid governance systems.
The inner circle is a conflict researcher's comfort zone. Humanitarian air services or armed convoys will bring the researcher from one inner circle to the next even in ongoing war contexts, granting little insight into the in-between spheres. When travel is possible 38  to the outer circle, it is often sporadic and mediated by certain actors that 'allow' that movement to occur or who even accompany the researcher to surrounding villages or farming areas. Whereas studying the inner circle seems mundane, visiting the outer circle is an exception. It is a sphere inhabited by seemingly uncommon actors with whom gatekeepers -often the state or international actors -form exceptional rather than everyday hybrid relations. Most conflict data hence stem from within the inner circle, whereas the outer one is relatively less known about. Methodologically speaking, few local security studies thus engage the full scope of actors, perspectives, and livelihoods, and not with the same expectation of the everyday.
What we know about the outer circle is thus often shaped by the perceptions of the people inhabiting the inner one. These perceptions are mostly negative and, in that way, empirically relevant. Compared with the organised inner circle, the outer one is described as chaotic -being where violence predominates. In CAR's Paoua, a chief prohibited youths from killing a thief in his town quarter; he added, however, that: 'If you want to kill him, you will have to do it in the bush, not here'. 41 In Haiti's Mirébalais, motorcyclists organised and started wearing gilets with their unique insignias to differentiate themselves publicly from motorcycle-riding thieves who entered from outside, robbed people, and then left again. 42 Considering the violence seemingly characteristic of the outer circle, it becomes logical that security actors claim legitimacy by differentiating (and protecting) the inner one from its surroundings. Using this narrative, people can also evoke the outer circle to pressure a leading inner-circle actor. In Somaliland, the state security apparatus was often dominant in town, but when it wanted to apprehend someone in the countryside the mediation of local elders was needed -or the officer risked being shot at. 43 In South Sudan, a youth leader in Mundri suggested that the state should cease its repressive behaviour in town if it wanted to stop young people from joining the rebellion in the bush. 44 The circles are also constantly evolving. Urban centres might expand and thereby enlarge their inner circle 45 or marketplaces in contested border zones -outer circlesmight create new meeting points that themselves can originate an inner circle. 46 The empirical manifestations of the inner and outer circle call for theoretical explanations of their differing characteristics. The physical proximity inherent within the inner circle increases frequency of interaction and the interdependency of livelihoods. This will more likely produce accepted rules of engagement. It also means that overlapping power claims regularly create frictions, thereby necessitating their lasting resolution. In the outer circle, spaces are wider; interaction and interdependence are less likely. Selfsufficiency and the possibility of avoidance allow for horizontal relations, ones in which security issues are continuously (re)negotiated. Thus while everyday peace is more easily observed by researchers in inner circles, the common practice of avoidance 47 in the outer circle make it more likely there. And whereas the need for hybrid order is ascribed by dominant gatekeepers as being a necessity of interacting with those in the outer circle, it is more relevant to the overlap of responsibilities in inner circles. In brief, inner circles grant more opportunities for what I term in the next section 'stable ordering' whereas outer ones tend towards 'fluid ordering'. The borderlines between the two spheres are often focal points of conflict because it is here where ordering principles clash.

How: clash of ordering principles
Looking at ordering helps differentiate between acts that at first sight resemble one another closely. For instance, around Paoua (CAR) two armed groups extorted money from civilians via roadblocks. However, while one of them, Révolution et Justice (hereafter, RJ) had fixed rates of FCFA 1,000 to 2,000 (EUR 2-3) depending on the vehicle, the other, Groupe de Patriotes (hereafter, GP) would take at times FCFA 500 from a motorcyclist and at others the motorbike itself. 48 An actor does not simply choose one unique form of ordering that it always applies. For instance, the police commissioner of Paoua was both rigid in some decisions and bent the rules in other instances: he was adamant about registering me officially during my second visit to Paoua, to ensure my own protection in his own words, all the while knowing very well that he had not the means of protecting even himself against recalcitrant delinquents and relied on flexible recourse to unmandated police 'auxiliaries' and fees.
I characterise ordering principles on a spectrum from fluid to stable to emphasise how vastly different varying actors can interact with one another in HPOs. On the latter side of the spectrum, actors hierarchise and fixate relations between actors, seek regularised resource extraction, make claims of control over security, and thereby create structures that allow inhabitants more predictability in the security domain. On the fluid ordering end of the spectrum, meanwhile, actors and people relate horizontally to one another within shifting relations, acquire resources in an ad hoc manner, evade each other's control, and thereby continuously modify the way in which security is governed. Neither stable nor fluid ordering is per se better for inhabitants' security: the predictability in stable ordering can also mean foreseeable violence by a repressive regime, whereas the adaptability to changing circumstances in fluid-ordering contexts can also mean that unforeseen violence is used by multiple actors in pursuit of their aims. These findings reinforce the critique that everyday peace actions are not always positive since fluid daily interactions below the elite-level can also reinforce structural and physical violence, especially against women. 49 As the inner and outer circles each tend towards one of the two ends of the ordering spectrum, so one could assume that different types of actors inherently prefer stable over fluid ordering and vice versa. Indeed, the example of Paoua seems to point in this direction: A non-state actor, the GP armed group, preferred using fluid ordering because it enabled them to remain flexible, make a modest profit, but not strain their very limited resources and capabilities by asserting control over a certain territory. By so doing, they were able to eschew defeat by the UN forces that used rather typical procedures of stable ordering: regularised patrols, territorial control, and modern warfare. Another non-state armed group however, the RJ, used relatively stable forms of ordering by controlling territories and regularising extortion. Interestingly, the UN collaborated with the RJ but combatted the GP despite neither holding significant levels of legitimacy in the eyes of the population. A mid-N comparative view on this single-case finding shows that, across cases, actors prefer collaborating with others who share their ordering principles.
John Karlsrud 50 has analysed the growing bias towards stabilisation in peacekeeping operations that I find reverberates throughout the cases studied here too. The collaboration in the Paoua case stemmed from the UN appreciating the RJ's principles of ordering -such as hierarchised command structures, public political ambitions, and territorial control -while denouncing as 'banditry' 51 the GP's ways of acting -loose structures, unclear aims, and local embeddedness. The UN Mission in South Sudan, having a similar bias towards stabilisation, has long adhered to the hierarchical primacy of the host state, despite the latter continuing its abuse of the population and even scapegoating the mission -thereby overall putting international soldiers and humanitarians in danger. 52 Although eschewing control and seeking negotiated relations with other security actors (rebel, community defence, armed groups) might have been more fruitful for both the citizens' and the mission's security in this situation, the strong international pressure to unilaterally engage in stable ordering forced South Sudan's UN mission into a dangerous predicament. For instance, when Sudan bombed a rebel camp near Raja in South Sudan in late 2014, the South Sudanese security forces prohibited peacekeepers from investigating the scene and negotiating a de-escalation of tensions between the warring actors. Stabilisation as a security objective thus poses a risk to actors' safety and interests when they hinder them from eschewing repressively dominant actors through the turn towards fluid ordering.
In comparing state security forces' tendencies towards stable and fluid ordering across cases, it stood out that in those where the state was relatively well-resourced and hierarchically controlled -for instance in Mirebalais (Haiti), Baligubadle (Somaliland), or Wau (South Sudan) -they resorted mostly to stable ordering. On the other hand, where control was infeasible due to limited resources -for instance, all CAR cases, Zeila (Somaliland), Anse Rouge (Haiti), and Mundri (South Sudan) -fluid ordering contrariwise prevailed. State actors are not naturally keener to use stable ordering, nor are traditional authorities more inclined to the fluid ordering end of the spectrum -ordering principles can clash even within each actor group. In multiple localities of CAR, state actors narrated a pursuit of stable ordering but recurred to fluid versions in their daily activities. Actors choose forms of ordering not only based on what they see as the right way, but also according to what they believe themselves realistically capable of pursuing and per how they expect others to behave. If an actor thinks it can dominate other security stakeholders, they might seek hierarchization through stable ordering even when this can breed conflict -for instance, the repressive control state security forces exert in Raja (South Sudan) and Daami (Somaliland). Even a weak actor might strive for stable ordering, should the stronger actor be expected to use its authority in their favour -thus in a qualitative survey in four CAR localities, an overwhelming number of inhabitants desired the re-establishment of a benevolent and strong state despite the presence of trusted but relatively weak local institutions. 53 If an actor thinks it is unable to control others, or if it fears violent authority being exerted by other actors -such as the weak GP near Paoua (CAR) or the politically marginalised Isse ethnic group near Zeila (Somaliland) -, then it is contrariwise more suitable to engage in fluid ordering. This allows for the negotiation of stakes according to changing interests and strengths, enabling everyday peace beyond official laws or top-down commands.
Differing ordering principles are known to the local population, and these can be used to resolve daily security concerns or to make a threatening statement designed to challenge the existing order. In Somaliland, security actors in Baligubadle and Zeila have divided up their responsibilities in a typical HPO manner: state actors are seen as in charge of town (the inner circle) and dealing with matters according to protocol (stable ordering), while clan representatives address issues arising in the countryside (the outer circle) through a mix of stable and fluid approaches. The key, however, is that the relationship between the two actors and their respective spheres is deliberately left illdefined to allow for fluid collaboration and to eschew rigid notions of control thereby enabling everyday peace practices, such as avoiding pursuing crimes that could cause wider contestation or keeping the known political causes of certain incidents ambiguous. 54 Collaboration of fluid and stable ordering, of state and clan actors in the inner and outer circle, has led to high levels of both felt and measurable security in many localities of Somaliland.
While divergent forms of ordering can be used to mutual benefit, we should not go as far as to romanticise their cohabitation. 55 Indeed, ordering principles often violently clash with one another. In South Sudan's Mundri, many youth in late 2014 joined the rebellion that was mounting in the bush around town to protest the state's repression of local opposition. 56 The state wanted to impose its hierarchical vision of security production, while local inhabitants wished to decide for themselves on security mattersespecially because the rest of the country was then engaged in a devastating civil war. The confrontation between imposed stability and fluid self-rule made Mundri one of the hotspots of the war in 2015. This local resistance -which in a similar sense could be witnessed during election protests in Jérémie in 2016 and 2017, 57 or in Zeila in 2012 58caused conflict casualties to rise; however, it also allowed people a sense of possibly gaining security in the future. This leads to the last point: the paradoxical differences between objective and subjective security.

What: paradoxes of security
The widely used Armed Conflict Location and Event Data Project (herafter, ACLED) counted zero conflict-related deaths in CAR's Ndélé 59 in 2019. Nevertheless, a housekeeper said that same year: 'We are deprived of peace'. 60 Of 60 respondents randomly selected for an open questionnaire, more than half felt 'rather insecure' or 'very insecure'. In Bangassou, on the other hand, ACLED recorded 21 conflict-related deaths in 2019, while a farmer said: 'Peace is beginning to return'. 61 Only a handful of the 60 randomly selected respondents in Bangassou felt 'rather insecure', and none 'very insecure'. Recurring either to conflict databases -as is often the case in statistical analyses -or to individual responses -as is often done in single-case studies -changes what is observed. Looking only at one of the two sides to security is to miss out on the intricate relationship between measured and perceived security.
Paradoxes of security describe instances where improvements in measurable security are accompanied by decreasing subjective security, or vice versa. Objective security can be ascertained through police records, civil society, and international organisations' reporting to create a triangulated database. Objective security is menaced by specific threats, such as attacks by armed groups or highway robberies. Subjective security is found by asking respondents in varying formats -interviews, focus groups, and surveyswhat their expectations concerning their physical integrity are. Anything mentionedeven immeasurable supernatural forces -is part of this side of security.
The sub-prefect of CAR's Obo described security as follows: 'Security, it is the absence of trouble, of acts that create confusion among the population, to flee, to be worried about your life and family. To live in peace, there is nothing more important than peace'. 62 Women in South Sudan's Buseri described their uncertainties similarly: 'People do not feel really secure, not even in the house. You have the feeling that anything can happen any time'. 63 Finally, in Hargeisa, youth from a minority group said: '[Security] can change at any time, any time people can fight and it can deteriorate, one moment it is good, another it is bad'. 64 The above statements, however, are in line with a typical problem of security studiesthey describe security as the absence of conflict. The sultan in Somaliland's Baligubadle explained why inhabitants of that area felt very secure -indeed, respondents vocalised this feeling of security throughout my two rounds of research there despite incidents of violence being far from absent on-site 65  , and the problem was solved'. 66 Other respondents described further incidences of violence with clans across the border in Ethiopia and over government encroachment on communal lands. However, a sense of ownership -or at least comprehension of how security is ordered -led most respondents nevertheless to respond they felt secure.
A participant in a youth focus group in CAR's Paoua, which has witnessed intermittent fighting for over a decade now, remarked: 'Since 2003 there has been no peace, there is a lot of disorder'. 67 In the town of Mundri, at the time not yet engulfed in -but as ensuing events would prove, on the brink of -violence, the acting commissioner specified the exact disordering elements: 'Insecurity is caused by cattle keepers [. . .]. When they came they were supposed to see the authorities [. . .], but they just sit wherever they want and give a hard time to our people [. . .], so people fear'. 68 People thus strongly relate security to how they perceive it to be ordered, and feel insecure when familiar ordering processes that enable everyday peace break down, leaving malevolent forces -often originating from the outer circleto cause 'trouble' by infringing the often-unwritten rules of cohabitation. This preference for known customary security institutions can even continue when these entrench a social order that exacts violence, as Marjoke Oosterom's study on women's support for male-dominated protection institutions in South Sudan showed. 69 Rumours exemplify how important it is to analyse both the measurable and the subjective aspects of security developments. Louisa Lombard and Sylvain Batianga-Kinzi note the spread of rumours in a CAR town from a teaseller, who accused a customer of being a 'penis thief'. 70 Since there exist no measurable instances of one person through magical abilities actually stealing the male sexual organ, it is a purely subjective form of insecurity and therefore does not figure in an objective security analysis. Nevertheless, rumours shape security regardless. First, they popularise the use of violence and, if based on a shared belief system, can mobilise a large crowd. Second, since the instigators of rumours remain unaccountable it undermines the authority of the actors who claim to dominate security production. On the other hand, third, acting upon rumours can bring legitimacy to a given individual, while neglecting them can reduce support for an actor. 71 In the above example, a military officer decided to have the alleged penis thief imprisoned, which increased the officer's authority. In other cases, the UN peacekeeping mission pressured the CAR police to liberate criminals accused of witchcraft. This undermined trust in the peacekeepers' security production, who did not take a subjective security threat seriously. It also undermined state authority, because administrators worried that people would pursue the liberated alleged sorcerers themselves and kill them. 72 The contrasting norms of this HPO involving international, national and local frames challenged everyday peace.
Security paradoxes can be explained by recurring to the two prior findings: First, the distinction between an inner and outer circle. The outer circle is a sphere of projection to inner-circle actors, often remaining opaque to them and to the researchers that cannot access it. Threats are therefore easily attributed to actors in the outer circle and are only very difficult to dispel as long as contact and movement between the two spheres remains limited. For instance, the inhabitants of Obo in CAR's south-east imminently feared the Lord's Resistance Army (LRA) although 'objectively' speaking they were last sighted hundreds of miles away; in South Sudan's Mundri, youth leaders spoke of a rebellion in the bush despite few verifiable incidents of insurrection having yet been recorded; many respondents in CAR's Bangassou or Paoua denounced the alleged violence emanating from nomadic Mbororo herdsmen in the countryside; taxi-moto drivers of Haiti's Mirebalais and Jérémie complained about armed robbers on motorbikes arriving from outside, tarnishing their own reputation.
However, second, the paradox is often a methodological artefact. 73 This means that objective and subjective security might realign, when analysed in depth and with a view to the political demands behind such seemingly 'paradox' exclamations: In Obo, inhabitants were unhappy with the work of the anti-LRA intervention force -speaking of an LRA threat was a way of denouncing the external forces' legitimacy. In Mundri, there were few acts of rebellion when the statement was made, but soon after a massive insurrection indeed did originate in the area to resist government repression. The subjectively ascribed violence of 'the Mbororo' in CAR is, in fact, a narrative used by armed groups to manipulate and in many cases extort 'their' respective constituencies, a typical everyday conflict action. 74 Local self-defence groups can use protection against herders' armed groups to challenge existing authorities that used to negotiate peaceful cohabitation. Armed groups linked to the Mbororo, on the other hand, often force herders to remunerate them for revenge attacks against farmers, even if the herders or the cattle owners never asked for such retribution. The motorbike thieves in Haiti were indeed also sometimes locally based, but the spatial differentiation delineated their legal practices from the illegal means some drivers might resort to in times of need. It is thus crucial to look both at objective and subjective security descriptions to understand paradoxical divergences between the two -and the contestations of local ordering that create them.

Outlook and conclusion
I discovered three patterns of security production in hybrid political orders that concern their shapes (where), their ordering principles (how), and their outcomes (what). By studying hybrid ordering across a moderate number of cases I respond to the recurrent plea for more qualitative comparisons, so as to better understand informal institutions. 75 These patterns build on insights from other scholars' single-case studies and point out avenues for theory-building of security production in HPOs especially in relation to the everyday peace and conflict debate. 76 First, security actors differentiating between an inner and an outer circle points out where crucial hybrid ordering interactions take place. Where inner circles are rigidly delineated, respondents fear their everyday peace practices to be interrupted by ominous threats emanating from the outer circle; strong interrelations between people in the two circles, meanwhile, lessen this danger ascription. In comparison, the outer circle of CAR's Paoua might share ordering characteristics with the outer circle of South Sudan's Mundri, even if both inner circles strongly diverge -and their respective national contexts vary even more widely.
Second, how people and actors produce (in)security varies on a spectrum from fluid to stable ordering. This finding counters the literature that suggests there is a natural tendency towards stable ordering by rulers over time, 77 and that people always prefer predictability. Fluid ordering that emphasises modifiability over predictability can be a meaningful strategy for those living in a repressive environment to enact everyday peace. Ignoring or subduing one form of ordering due to an inherent bias towards another (for instance a stable ordering preference is common for international interventions) can heighten insecurity by fuelling clashes over ordering principles. These different ordering principles should, however, not be attributed in primordial ways. All actors use a range of ordering actions on a spectrum from fluid to stable based on what they believe is right, feasible, and regarding how others will react. In HPOs neither ordering practice is per se prone to more security. Predictable security brought forth by stable ordering can also signify constant repression, while fluid everyday peace practices are mirrored by the possibility of structural violence (especially against women and minorities) in the form of everyday conflict.
Third, what security outcomes are observed during fieldwork in HPOs can at times be unexpected or even paradoxical, when arriving with hypotheses based on seemingly 'objective' security measurements. It is thus indeed meaningful to study subjective aspects of security from the bottom up, in line with the current trend towards 'vernacular security' 78 and 'everyday peace'. 79 Nevertheless, 'objective' security still matters, especially when it -seemingly paradoxically -differs from subjective security feelings. Such security paradoxes can teach us about the workings of security production and people's support for or resistance to ordering processes. Acknowledging people's felt insecuritydespite the absence of measured conflict -can have real-world consequences: if people disagree with how security is currently ordered, clashes can arise and lead to blatant insecurity.
Although the methods applied here are common in qualitative research, the question remains whether employing them across more than a few cases within a mid-N transnational comparative design allows researchers to gather sufficient information to make statements about the larger workings of security production rather than just about the few data points amassed. In other words, how to achieve a meaningful level of saturation. 80 Working in teams on the ground rather than alone vastly accelerated achieving saturation for this study. This goes especially for teams wherein the researchers differ on perceptions of the dimensions relevant to the study at hand. Diversity in gender, origin, age, and language ability can speed up the process by reducing the likelihood of overlooking key categories in the data. 81 Furthermore, funding permitting, splitting research stays into a short pilot, a long main, and a brief follow-up visit improved saturation's attainment by building trust, testing expectations, and making long-term trends observable. Finally, continuous event logs filled out by local knowledgeable people such as journalists or teachers, even when untrained, are a final tool to test for saturation. In sum, given the same time constraints as single-case or small-N studies, applying these and other tools to one's research design can provide a meaningful level of saturation even for mid-N comparisons.
The article raises questions concerning hybrid political orders, everyday peace and conflict, and the security studies field in general. More outer-circle research on hard-toreach places that are subject to travel restrictions, and which are lightly populated by inhabitants who often converse in languages rarely studied at international universities calls for a greater appreciation of the work of domestic researchers -in line with the everyday peace agenda that 'confront[s] the dominant narrative that associates peacebuilding expertise with outsiders and essentializes "locals" as insular and passive'. 82 Further research is also needed on relating everyday peace and ordering activities to the fluid and stable ordering spectrum, wherein the initial proposition presented herein is that everyday peace strongly relates to the fluid ordering side. The lens of 'ontological security' 83 can be used to investigate identity stability in conflict contexts, which could explain parts of the security paradox. With the Critical Security Studies lens applied, 84 researchers might focus on how security is constructed as a priority public good and how subjects decouple (or not) their demands from the security prerogative and particularly towards the outer circle.
The findings can also inform policy. Well-intentioned security actors should ask what security means to the people they are being sent to protect, how and why they subdivide their intervention space into an inner and outer circle, and what existing ordering principles are currently employed cooperatively or competitively -as well as why. They should carefully gauge how their intervention impacts not only conflict incidents but also security perceptions. Rather than trying to securitise a seemingly threatening outer circle, they should support building bridges between the inner and outer circle (as Somaliland actors have successfully done in many localities). Finally, external, institutionalised intervention forces should acknowledge their stable-ordering bias and seek to balance their activities to facilitate a locally sustainable mix between fluid and stable ordering that enables everyday peace as well as calling for wider structural reform.