Exploring the Backstage: Methodological and Ethical Issues Surrounding the Role of Research Brokers in Insecure Zones

ABSTRACT The contribution and situation of research brokers problematically tend to be shrouded in silence in most research texts. In this article we probe into the particular ethical and methodological challenges that we may encounter when working with brokers in conflict settings, drawing upon existing literature and contributions of this special issue. Reposing on post-colonial perspectives, we problematize both the increasing securitization of conflict research with its one-sided focus on researcher safety and the notion of researcher responsibility. Moreover, we argue that the inequalities marking researcher-broker relations are often particularly glaring in conflict settings, thus increasing the risk for exploitation.


Introduction
While local brokers, such as research assistants, interlocutors and a variety of 'fixers' play a crucial role in most research contexts they tend to be particularly important in violent theatres. Not only do they have the in-depth knowledge that enables them to gather data on sensitive topics, they are also crucial in enabling researchers to navigate safely through 'dangerous fields' (Kovats-Bernat 2002) or 'frontlines' (Hoffman and Tarawalley 2014) whilst doing what Nordstrom and Robben call 'fieldwork under fire' (1995). From dealing with reluctant state agents and fostering sufficient trust to gain access to isolated groups (e.g., Norman 2009), to managing perceptions of the researcher and obtaining updated security information on the security situation: locally based brokers are key resources and gatekeepers (Campbell et al. 2006, Luning 2013, Middleton and Cons 2014, as well as security and crisis managers (Mandel 2003, Hoffman and Tarawalley 2014, Jenkins 2015). Yet, the role of many brokers goes much further than merely facilitating research or gathering certain data. They often become the eyes and ears of researchers, thus exercising a large influence on the latters' grids of intelligibility, shaping not only the way in which they make sense of certain phenomena, but also what they see in the first place (Rabinow 1977, Olivier de Sardan 1995, Metcalf 2002, Borchgrevink 2003, Paluck 2009, Turner 2010, Caretta 2015. Brokers could, therefore, be considered as full-blown 'coauthors' of research without writing a single word. However, the contribution and even existence of research brokers problematically still tend to be shrouded in silence (Sanjek 1993, Molony and Hammett 2007, Turner 2010, Gupta 2014)including in our own (the authors') research. Often not making it further than the acknowledgements section, the various ways in which brokers co-construct research rarely feature as the subject of systematic reflection in the final research 'results'. It is the (often Northern) researcher that takes centre stage in research accounts as s/he reflects on his/her positionality and the ethical and methodological difficulties encountered (Barnes et al. 1966, Salamone 1977, Geertz 1983, Bleek 1987, Sluka 1995, Herbert 2001, Mandel 2003, Vlassenroot 2006, Venkatesh 2008, Luning 2013, Lee 2016. As Bloor et al. puts it, the researcher is often portrayed as 'the lone gun' (Bloor et al. 2010, p. 46), reflecting gendered imageries of 'the lone, white male trampling about in his "stout boots," overcoming obstacles, "objectively" recording observations and revealing truths' (Campbell et al (2006, p. 97). Given the crucial contribution of brokers to the research performance, such silencing, 'ghost-workering' (Turner 2010), or outright 'ghost production' (Sukarieh and Tannock 2019) is thus problematic both from the perspective of research ethics and research quality.
Better knowledge about the role and situation of local research brokers appears particularly urgent at this point of time as many research institutions in Europe and the US are increasingly regulating and restricting the fieldwork access of their staff due to security concerns (Peter and Strazzari 2017). This, in turn, leads to more outsourcing to locally based research assistants, increasing the risk of exploitation (Sukarieh and Tannock 2019).
This special issue explores the methodological lacuna briefly outlined above by letting a range of researchers and former brokers engaged in conflict research reflect upon their experiences. Clearly, brokers are commonplace in all types of research and the ethical and methodological dilemmas encountered in conflict settings are certainly not unique, particularly given the porous boundaries between secure and insecure spaces, as well as war and peace more generally. Yet, while the dilemmas reflected upon here are also prevalent in other research areas, the aim here is to highlight how research in conflict settings presents additional challenges to working with and through brokers. In short, what particular challenges may conflict and insecurity pose to the broker-researcher relationship?
In this introduction we probe into the particular ethical and methodological challenges that we may encounter when working with brokers in insecure and conflict settings, drawing on the contributions to this special issue, as well as the existing (yet limited) literature from various disciplines. Not surprisingly most accounts of relations between researchers brokers/ assistants come from anthropology (for early accounts see e.g., Griaule 1948, Powdermaker 1966, Turner 1967, Grindal and Salamone 1995, as explaining and qualifying social ties and relations with the field are part of what you are expected to do in methods sections in this discipline. Yet despite such insights, anthropologists still devote relatively little space in their writings to research assistants and other brokers. In a hackneyed way the psychological character and the social capital of the researcher her or himself is often described as the main asset, where the researcher is the 'a sole vessel of observation and analysis' (Cons 2014, p. 376). More so, we propose, a special lacuna can be found in anthropological studies on armed conflicts. A reason for this may be that up until rather recently war and conflict have not been much studied within anthropology (Sluka 1990, Nordstrom and Martin. 1992, Nordstrom and Robben 1995, Avruch 2001, Waterston 2009).
However, far less attention has been directed to research brokers in other conflict research fields such as IR and Political Science. This is hardly surprising and reflects the methodological tools used within these disciplines, which tend to be less fieldwork heavy and also geared more towards quantitative methods. Yet, there is an emerging literature addressing ethical dilemmas in conflict research coming from political science, IR and social geography and some of these also addresses questions around research brokers and assistants (Cronin-Furman and Lake 2018; Vlassenroot 2006). Moreover, in a rather unique forthcoming article Hoover Green and Cohen (2019) address ethical dilemmas in quantitative conflict research, touching upon research assistants and brokers. Yet, while there is an emergent literature on ethics and methodology in IR and political science, much of this literature seems quite oblivious of insights and debates within anthropology. In short, there is a visible need for more consultation and debates between the various disciplines engaged in conflict research. This introductory article will proceed as follows. First we will more carefully define what we mean by brokers. After this overview, the article is divided into three major themes, reflecting what we identify as the particular challenges attached to the research-broker relationship in conflict settings. The first part focuses on the challenges posed by the predicaments of insecurity and violence, while the second theme addresses one of the most common themes in the wider (yet limited) literature on brokers and research assistants, namely whether the researcher-broker relationship can be one of mutual benefit (and even friendship) or whether it is intrinsically marred by inequality and exploitation. The third theme, addresses more methodological challenges: how may brokers shape the research process, and ultimately the research results?

Brokers?
As will be further clarified in our conceptualisation below, in our notion of brokers we include what in the literature is often named 'research assistants' or 'local research collaborators'. In short, we define a research broker as a key agent being in-between the researcher and the researched who regulates the access and flow of knowledge between them. We acknowledge that the notions of research brokers and research assistants carry different connotations. In some contexts the concept of broker (similarly to the concept 'fixer') has a derogatory connotation when compared to research assistant (see the contribution by Parashar). Yet, by using the term brokers we have no intention of minimising the role or importance of 'research assistants' or 'local research collaborators'. On the contrary, our aim is to (as hopefully reflected in the definition) highlight their crucial role, something that arguably is not well captured in the term research assistant.
Moreover, we chose the notion of research brokers in order to be able to better acknowledge the wide repertoire of actors whom are vital to research endeavours in conflict settings. By only focusing on those participating in the collection of the data (e.g., conducting interviews or surveys) a range of other crucial actors who are vital to regulating the access and flow of knowledge in research are made invisible. While such actors may not actively participate in the collection of the data, they might play an equal or even more important role in regulating access and flows of knowledge and ultimately shape the research results. As addressed in the contribution by Alfred Banga et al., brokers do not only come in the form of people/ individuals but as institutions and organisations, as academic research in conflict settings often relies on support from UN-missions and humanitarian organisations. Importantly, the concept of broker here should not be understood in opposition to the notion of researcher. Rather the boundaries between brokers and researchers are inherently blurred. As noted initially, brokers could be considered as full-blown 'coauthors' of research even when not writing a single word. In addition researchers who originate from, or who have spent considerable time in the research setting often also become brokers for new researchers (or journalists, humanitarian workers etc.) arriving to an unknown field (Sangarasivam 2001, Luning 2013, Hoffman and Tarawalley 2014, Middleton and Cons 2014; see also contribution by Utas) or act as brokers in relation to other members of a larger research team (see the contribution by Myrttinen and Mastonshoeva).
Yet, while several contributions in this special issue address institutions as brokers (Alfred Banga et al.) and the researcher as broker (Myrttinen and Mastonshoeva; Utas) the main focus of the articles (and the remainder of this contribution) is on brokers as people who are originating from, and most often still live and work in the research settingoften in other literature named 'local'. Clearly, even with this delimitation brokers are a highly heterogeneous group, performing various forms of tasks (e.g., translation, collection of data, analysis and interpretation, brokering access to local communities, finding safe places to stay, providing security, mediating conflicts), occupy various levels of socio-economic privilege and have a range of occupations (e.g., journalists, university lecturers, students, NGOworkers, politicians, combatants in armed groups or soldiers/officers in state armed forces). Moreover, while some engage in brokerage on a more temporary basis, othersparticularly in post-conflict settings where there is an 'oversaturation of research' (Clark 2008) and a strong presence of humanitarian actors and peace-keeping forcescan turn brokering into a more regular occupation (see below).
Reflecting colonial history and the continued stark inequalities of North-South relations in and outside knowledge production (Chakrabarty 2000, Spivak 2004, Boshoff 2009) many researcher-broker relationships are deeply embedded in, and shaped by, colonial memories and identities. Yet, while North-South inequalities in knowledge production are indeed strong, particularly in relation to research on Africa (Boshoff 2009, Mwambari andOwor 2019), much conflict research is also conducted by researchers based in, or originating from, the global South (D'Costa 2011, Parashar 2014; see also Henry 2003) who also make use of brokers in their research. In short, imagining the researcher-broker relationship simply as reflecting a North/ South divide problematically downplays the research conducted by scholars from 'the Global South', in turn reflecting the problematic trend of Northern white navel-gazing marking much of post-colonial studies (Chakrabarty 2000, Eriksson Baaz andVerweijen 2018). As highlighted particularly in the contributions by Alfred Banga et al., Parashar and Myrttinen and Mastonshoeva conducting research in 'native areas' (see Parashar) pose particular dilemmas and challenges. As Parashar shows, being seen both as foreign and an insider complicit in the violence and oppression of the state had serious implications during her fieldwork and also affected the research output.
Given the immense fluidity and variation in possible researcher-brokers positionalities and assemblages highlighted above, a general theory of researcher-broker relations is clearly not only impossible (Cons 2014, p. 389), but dangerously misleading. Ethical and methodological challenges will always vary depending on the various (and multiple) positionalities of brokers and researchers at hand, as well as the shifting dynamics in the conflict setting in which they work. While we hope to capture some of this great variation in this special issue we are inevitably doomed to fail. The contributions, while excellent and varied, still provide fairly limited examples of possible researcher-brokers relations and constellations. Yet, the most serious failing in this special issue is the ways in which it, with a few notable exceptions, reflects the silencing of the voices of research brokers themselves in academic textssomething we hope to correct further on along the research project of which this special issue forms a part. 1 Let us now develop the three themes mentioned initially. We will start with the most obviousyet still markedly biaseddiscussions of the challenges posed by in/security.

Whose (in)Security Matters?
Clearly, in/security is not only a concern for researcher-broker relations in conflict zones: the boundaries between insecure and secure spacesas well as between war and peace more generallyare certainly blurred (Richards 2005, Beck 2012). In particular, much research conducted on 'sensitive topics' (e.g., corruption, politics of repression etc.) is attached to similar security concerns as research in conflict settings. Yet, one of the defining aspects of conflict research is precisely insecurity and this, unsurprisingly, stand in the centre of the general methodological literature within conflict research.
As we will emphasise below, a defining feature of the methodological discussions on the predicaments of insecurity is the focus on the in/security of the researcher. Lately we have witnessed how '[a]dditional resources are put into ensuring the physical safety of researchers, and into securing data along the research supply-chain' (Peter andStrazzari 2017, p. 1532). An additional defining feature of the literature is that most researchers admit that others in the field (informants, brokers etc.) are more knowledgeable in terms of assessing and managing insecurity and risks. Hoffmann (2014, p. 5) when researching a particular wing of the armed group Mai Mai in the DRC, concludes that 'During my research I quickly became aware that my collaborators and interviewees were much better equipped than I to foresee the potential hazards of the research'. In a similar manner, Kovats-Bernat (2002) in his fieldwork recalls how he soon realised that his informants where much better at identifying and dealing with dangers than he was. Nordstrom (1997, p. xvii) similarly states that it was the awareness of those around her that had ultimately protected her from violence.
In acknowledging their relative inability to foresee risks and dangers many researchers take a critical stance towards the notion of 'researcher responsibility', repeated in most guidelines on research ethics. For instance, Kovacs-Bernat argues that instead of researcher responsibility we should talk about 'mutual responsibility': This relationship should be one of mutual responsibilityand not just the validity of data reported: all participants in the research must also willingly accept the possibility that any involvement in the study could result in intimidation, arrest, torture, disappearance, assassination, or a range of other, utterly foreseeable dangers (Kovats-Bernat 2002, p. 214).
Learning from this literature (as well as our own experiences) a researcher taking on the role of judging and managing insecurities risks jeopardising not only his/ her safety, but particularly those of others. Yet, and arguably, the notion of 'mutual responsibility' is also problematic as it somehow connotes an equality of risk and privilege. As noted above, crude distinctions between a secure researcher and insecure broker are clearly simplistic as brokers and researchers hold various and fluid positionalities. Yet, particularly in contexts where the researcher is white and based in the global North and the broker is based in a conflict setting in the 'global South', insecurity and risk are clearly not equally distributed between researchers and brokers. In contrast to the researcher, brokers (through their embeddedness) form an integral part of the (often) particularly complicated and volatile power fields in conflicts settings. As such, they tend to be already named and seen as belonging to/representing a particular conflicting party in the eyes of research subjects and others in the surroundings. This is clearly reflected in a joint piece by Middleton (researcher) and Pradhan (broker) (Middleton and Pradhan 2014, p. 364) where they reflect upon their work on 'tribal politics' in India. As Pradhan concludes, his positionality became a risk for him as local leaders 'would look at my involvement from a personal point of view and also as between two ethnic groups: the rivalry between their ethnic group and my own group'. Yet, as reflected in Parashar's contribution in this special issue, the same predicament often faces researchers conducting research in 'native areas'.
Moreover and related to the point of inside/outside raised above, interpretations of 'risky/suspicious actions' that may increase insecurity are also unequally distributed between researchers and brokers, as pointedly raised by Pradhan in the article cited above. As Pradhan explains to Middleton (ibid, p. 366): . . . white people, because of their skin color, do have a license in India to do things which normally the locals would find outrageous, right? Even if you climb up on the dais or try and sit next to the biggest leader being felicitated, you would still be ok. Someone would show you the way out (laughter). Or if you just shoved the camera in front of [political supremo] Subash Ghisingh or somebody, you could still get away with it . . .
Here again, researchers conducting research in 'native areas' are often not granted the same generous interpretation of suspicious/out of place actions.
Most importantly, in cases when risks such as 'intimidation, arrest, torture, disappearance, or a range of other, utterly foreseeable dangers' (Kovats-Bernat 2002, p. 214) appear, the options available to secure oneself often differ considerably between researchers and brokers. Researchers frequently have access to a range of security measures that are out of reach for most brokers. In addition to having better access to financial resources that can 'solve problems', researchers can regularly access support from his/her embassy or the wider research/humanitarian/peace-keeping community and, ultimately, simply leave if the risks appear too high.
Relatedly, while researchers' insecurity predicaments tend to be limited to fieldwork, brokers often live with their involvement in research projects for a long time after (Middleton and Pradhan 2014). Jenkins (2015), for instance, offers an account of how, a year after leaving the field, she received a message from one of her assistants/brokers that someone had begun to spread rumours around his village that he had worked with her to bring Kenyans to The Hague. Yet, in many cases where relations with brokers are less long term the researcher might not even be aware of the risks and violence brokers are exposed to after fieldwork. For instance, while conducting a research project among Congolese refuge victims of sexual violence in Uganda 2 the research team (involving one of the authors) encountered a deeply traumatised former broker. As a punishment for his facilitation of a US-based larger research project on sexual violence in the DRC, soldiers from the government forces gang raped him and killed a family member after the researchers left. Yet, as he contended, the researchers in charge were probably not aware of what happened to him. While he initially, similarly to the other brokers/assistants had received the researchers' contact details, these were lost while fleeing to Uganda after the retaliatory violence committed towards him and his family.
In short, while 'mutual responsibility' clearly seems more accurate and preferable to the notion of 'researcher responsibility' it arguably risks to problematically downplay that the risks are seldom equal. Moreover, it hides that brokers often consider themselves as the host whose role is to secure the researcher (Jenkins 2015, Deane andStevano 2016). Ironically, such sentiments may put the broker in even more risk as he/she tries to safeguard the safety of the researcher, at times through efforts at mitigating effects of researchers suspicious and risky behaviour. 3 Surely, there are also different (less gloomy) stories that can be told in terms of researcher-broker security. Hoffman and Tarawalley (2014), for instance, argue that travelling together meant that they safeguarded each other. Tarawalley with his deep network of rebel commanders took Hoffman to places where he would not have reached himself. Yet, Tarawalley was on other occasions protected because he travelled with an outsider and someone who had the power of giving accounts in case something would happen to Tarawalley. Yet, these 'formidable' relationships are quite rare and appear mainly possible or beneficial to brokers when theyfor some reasonharbour and have an interest to travel and engage in relations otherwise not accessible to them.

Gendered Notions of In/security
While brokers tend to harbour feelings of responsibility to 'secure the researcher' more generally, such sentiments are clearly also gendered. Such gendering appears particularly evident in conflict research, where (drawing upon the existing literature and the contributions in this issue) an overwhelming majority of brokers are menwhile many researchers are women. As Jenkins 2015, p. 158) concludes in her reflections on researching post-electoral violence in Kenya: As a young, female researcher, my assistants and local fieldworkers adopted a very protective stance with regard to my security and it was not uncommon for Hassan to profess, "What will I tell your mother or your father if something happens to you? If something happens to their daughter, what will I say?" It seems likely that my assistants were more cautious than they perhaps would have been with a male researcher, and I suspect that they did not actively communicate their tactics to me in the same way that they would have had I been male.
Much of the scholarly literature on in/security in research highlight the ways in which women researchers are at (sexualized) riskincluding by brokers (Sharp and Kremer 2006, Mügge 2013, Hanson and Richards 2017; the SAFER Research Handbook). For instance, as concluded by Saliba et al. (forthcoming) in the SAFER Research Handbook (draft manuscript) addressing research in conflict settings: 'Being a female researcher is sometimes associated with an increased exposure to uncomfortable situations while interacting with male counterparts (if not actual sexual harassment)'. Some, like Sharp and Kremer (2006) even argue that the attention to protect research subjects has led to a silencing of the possibility that female researchers may also be at risk. Without in any way questioning that women researchers may be exposed to particular (including sexualized) risks (ibid) we want to draw attention to how assumptions about risks in conflict research tend to be apriori gendered in specific ways that also needs to be problematized. As reflected in the contribution by Eriksson Baaz, such apriori gendered assumptions risks downplaying the security predicaments of male (even militarised) brokers. Moreover, constructions of risk/insecurity are not only gendered in a problematic manner, but shaped by colonial racialized and sexualized narratives of the innocent white woman exposed to the threats of brutal, uncivilised and oversexual Other men (Eriksson Baaz and Stern 2013).

Mutual Benefit or Inequality and Exploitation?
One of the most common themes in the wider literature on brokers and research assistants revolve around the nature of researcher-broker relationships with regards to inequalities in power, privilege and benefits. Below we first recap the more general debates and then situate the discussion in the context of conflict research.

Overview of Debates outside Conflict Research
As emphasised above, the model of a privileged researcher using or even exploiting a non-privileged broker is clearly simplistic (as brokers hold various positionalities and levels of privilege). Moreover, efforts to assess broker-researcher relationships are complicated by the fact that reciprocity or mutual benefit can come in various forms. A broker, depending on positionality, can have various gainsother than financial remunerationfrom the work. Brokers could benefit from co-authorship, though this as highlighted above, is quite rare in social science conflict research (and see Utas in this issue). Yet, brokers may also gain experience and establish relations with influential people who could facilitate further careers (in or outside academia). As will be highlighted in the coming section, brokers may also attain political gains when/if they manage to get the researcher to voice their grievances (Hoffman andTarawalley 2014, Hoffmann 2014). Moreover, brokers and researchers may sense the emotional benefits of friendship. In addition, researcher-broker relations are not static but evolve and change over time (Campbell et al. 2006). What may start as a more distanced and contractual agreement with fixed remuneration can later develop into friendship, and even love affairs and marriage (Ajwang and Edmondson 2003).
The general literature (outside of conflict research) offers various suggestions. As touched upon above, one strand of thought highlight the colonial legacy of researcher-broker relations (Sanjek 1993, Middleton andPradhan 2014). Clearly, the images that researchers hold of themselves and research brokers, and vice versa, as well as how such identities are manifested in research practice are often shaped by colonial history (see below). As Sanjek (1993) points out, the very practice of systematic silencing of brokers forms part of that history (see also Spivak 1988, Turner 2013. Consequently, and hardly surprisingly, scholars acknowledging the importance of post-colonial perspectives tend to paint a quite gloomy picture of researcher-broker relations as inevitably unequal, even exploitative. Middleton and Pradhan (2014) for instance, talk about the 'subalternity' of research assistants, framing the relationship as an 'extraction surplus' and acknowledge the inevitable failure in their efforts: Despite our modest successes in achieving a more inclusive and equitable mode of ethnographic engagement, we can make no claims to fully transcending the coloniality of the researcher-assistant relationship. Structural inequality and difference remained (and remain) integral to our relationship. (2014, p. 358) Middleton and Cons (2014, p. 284) also emphasise the inherent inequality of the relationship highlighting how fieldwork labour 'cannot be divorced from the logics of capital'. As they conclude: While these fieldworkers may become key informants, cultural brokers, coauthors, and even friends, they remain employees. As such, the commodification of ethnographic labor proves integral to the researcherassistant partnership. [. . .] With rare exceptions, ethnographers are in the business of transforming the work of the hired assistant into the cache of intellectual capital and acumen.
Yet, other scholars paint a less gloomy picture of the relationship, often evoking the agency of the broker/assistant. Schumaker (2001, p. 12), for instance, argues that '[t]he model of exploitation fails to capture the assistant's own motives and goals'. Parts of this literature highlight the powerlessness experienced by the researchers in an unknown environment. Another example is provided by Deane and Stevano (2016, p. 217) where they underscore their sense of vulnerability by citing an assistant who talked about the researcher as 'having a child to look after'.
In other accounts, a picture of a reverse exploitation, where brokers use researchers appear to emerge. While there are several types of accounts of feelings of 'researcher exploitation' in the literature, much is centred on demands for material and financial resources. In such accounts research assistants are often portrayed as somehow forcing themselves on researchers to 'extract cash from the research funds' (Molony and Hammett 2007, p. 294) and as 'simply after the money'. (Ibid, p. 296). Molony and Hammet further conclude: In the eyes of somenot least the research assistant, whose close work with his/her employer provides a good idea of their cash flowthe researcher may be regarded as a squanderer or a soft touch, an easy opportunity to obtain money" (ibid).
In short, images of brokers as greedy and canny and thereby somewhat unreliable in contrast to an honest, well-meaning but somehow gullible researcher Self emerge in some literature (Cons 2014). In our own experience, accounts like this tend to emerge particularly frequently 'off text', when researchers get together and share field experiences. Such images are familiar from the colonial library and resonate with post-colonial white identities elsewhere, not the least within development work (Eriksson Baaz 2005). Echoing such familiar representations, some scholars also engage in discussions whether the researcher 'have a role to play in advising the research assistant in how to save regular earnings?' or perhaps should even 'hold back a percentage of the research assistant's income, or forward it to the wife' when she/he sees that 'payment is being squandered on alcohol and prostitutes?' (Molony and Hammett 2007, p. 297). According to Molony and Hammet, such questions (as financial issues more generally) 'can weigh heavily on the researcher' (2007( , p. 287, see also Jenkins 2015. Molony and Hammet do not provide an answer to the questions they pose, and they also do acknowledge that answering yes might be regarded as paternalistic. Yet, representations such as thesepromoting images cunning, prodigal brokers, unable to use money in a wise mannerare clearly problematic. By emphasising how financial questions 'wear heavily on the researcher' and assuming that the problem is that assistants don't know how to use money and save, such accounts certainly fail to recognise researcher privileges, as well as the realities of acutely insecure livelihoods. In general, accounts of research-broker relations as friendship are quite rare and mostly apparent in the general anthropological literature. This is hardly surprising given that friendships often take longer time to develop. Spending a year or more in the field, often sticking to the same brokers, can establish not just deep ranging friendships, but also mutual dependencies and expectations that may turn into lifelong relations further extending to relatives and so on. Some texts bear witness of such experiences of research-broker relations as true friendships and mutual dependencies, rather than exploitation (Jackson 2004).

Conflict Settings: Mutual Benefit or Inequality and Exploitation?
The inequalities marking researcher-broker relations in many other research contexts touched upon above are arguably often especially glaring in conflict settings. Brokers in conflict settings, compared to other contexts, are frequently in particularly precarious situations. One defining feature of conflict settings is that while resources can sometimes be more easily accumulated (particularly by armed actors), resources are also more easily lost (e.g., through various forms of retaliation, pillage, armed group taxation, increased risk of expensive sickness and death and more). Above all, providing security for oneself and one's family is costly and it can become particularly expensive when associated with a wealthy researcher.
The precarious living conditions also appear to be reflected in that brokers are forced to accept work for very little pay. A recent article by Cronin-Furman and Lake that touches upon research brokers/assistants highlights that assistants are often badly remunerated. As they conclude: '[o]ut-of-work or underpaid professionals may affiliate with foreign researchers for little or no pay in the hope that doing so could lead to future employment, educational opportunities, or open other doors' (Cronin-Furman and Lake 2018, p. 5). In a recent blogpost, David Mwambari and Arthur Owor talk about a black market of knowledge production where Northern researchers 'rely on local human resources to collect knowledge' separating foreign 'experts' from adjunct local experts due to abysmal differences in financial situations (2019, p. 1). Thus at times ghostworking assistants endure lives under bare survival, where employment is perceived more as a hopeful investments in an uncertain future shaped like patronclient, or rather a Big Man-follower relationship (Utas 2012) based on ideas of extended forms of debt. This vulnerability, in turn often pushes brokers to expose themselves to considerable risk, fearing that the (prospects of) income offered will otherwise disappear.
Pervasive insecurity might make co-authorship, another aspect of reciprocity, more difficult as this might expose the broker to further risk. Yet, and importantly, this argument can also conveniently be used as a pretext for not including collaborators as co-authors. A call amongst anthropologists more broadly to co-publish work with 'assistants' has recently lead to a few interesting co-publications. In addition to the article by Middleton and Pradhan (2014) cited above, the text by Hoffman and Tarawalley (2014), although written by Hoffman rather than jointly, offers a poignant conversation which pinpoints the mutual benefits of a relationship with regards to 'frontline' anthropology. However, these remain rare examples in conflict literature. A most glaring call for a change is Sukarieh and Tannocks article (2019) on what they call the Syrian refugee research industry where they note a structural erasure of research assistants. Sukarieh and Tannock talk about the bifurcation of the research environment into Northern 'research capitalists' and a southern 'research proletariat' (2019, p. 668). Their cases from the vicinity of the Syrian warzone come across as extremely troubling exploitation with cases where many assistants have university degrees from Europe and US and play central roles not just in collecting date but also in the writing process. An assistant's voice in the paper makes this point lucidly clear: 'I was paid to be a researcher, but not paid to write a paper to be published not in my name, while it is 90% written by me. What is the PI being paid for?' (Sukarieh and Tannock 2019, p. 675). Another assistant summed up her experience: '[t]hey steal our labour, they steal our future, they steal our self worth' (2019, p. 673).
However, while the inequalities marking researcher-broker relations in many other research contexts appear particularly glaring in conflict settings, such settings can also offer particular benefits/opportunities for brokers (see Caretta 2015). The demand for brokerage tends to be particularly high in conflict and post-conflict settings, created by the often massive influx of international resources through peacekeeping missions and humanitarian aid (Hoffman andTarawalley 2014, Hoffmann 2014), frequently marked by an 'oversaturation of research' (Clark 2008). Such settings have turned brokerage into a livelihood strategy for many (mostly men) who are engaged by a range of actors such as intervening militaries and peacekeeping forces, humanitarian actors, journalists and researchers (see contributions by Käihkö och Utas). Hence, the brokers engaged by researchers often play multiple roles, simultaneously facilitating the work of a range of actors (de Jong 2018). These multiple engagements and entanglements and their possible implications for research have received little attention in the literature.

Brokers Crucial Impact on Research Results in Conflict Settings
Let us now turn to the more methodological question of how brokers shape the research process, and ultimately 'research results', in conflict research. Most researchers have encountered and tried to make sense of totally disparate narratives of events, people and places (not only by different groups of people, but from the same person encountered in different contexts and time periods, see contribution by Käihkö). Yet, while such experiences are not unique for conflict research, they tend to be particularly pronounced in such research settings, reflecting both high levels of politization and rapidly shifting conflict dynamics, allegiances and identities.
How we make sense of, and handle, such divergent and contradictory narratives and accounts depends, of course, on our (different) epistemological positions as researchers, but also the varied aims of our research projects and the truth claims we seek to make. Yet, regardless of our various epistemological positions and research aims, we (as conflict researchers) need to acknowledge that the brokers we work with have a particularly marked impact on the research processes and ultimately the 'results' however we define that. This particular pronounced role and impact stems partly from 'researcher incapacity' and consequently the exceptional dependence on brokers in navigating complex and volatile power fields discussed above. However it also reflects the highly politicised nature of conflict zones. While we as researchers always have to be attentive to our and others positionalities when we carry out field research, as well as when we write our texts) (Blaufuss 2007, Hoel 2013, the political power fields we study and that we, and research brokers, make use of are particularly politicised in zones of conflict. There is thus a risk of 'diplomatic counterinsurgencies' (Kostic 2017) also in small wars. Hence, in such heavily politicised settings not only research informants, but also brokers tend to form an integral part of political games and conflict narratives, and they are at times also targeted by people with political motives. As the relationship between brokers and researchers is asymmetrical and in a variety of ways politicised, managing different versions of stories are essential. When we deal with the triple subjectivity of researcher, research broker, and researched we need to address the versions in all relations as different relations form different sets of problems (Deane and Stevano 2016).
In some cases, the positionality and political agendas of brokers appear more straightforward. Some researchers have accounted for the ways in which they clearly sensed that brokers tried to make them see and emphasise one particular conflict narrative over another. For instance, in his research with the Mai-Mai (as brokers and informants) in eastern Congo Kasper Hoffman concludes that: Whereas I was interested in their narratives for scientific and ethical purposes, but also for the advancement of my career within academia, they were interested in collaborating with me because they saw me as someone who would be able to help them get their message through to the outside world, but also as someone who might be able to help them practically" (2014, p. 11).
He continues, explaining how even this kind of obvious positionality/partiality, can be seductive: In the narratives of the Maï-Maï they represented the will of the Congolese masses, and were the vanguard of Congolese resistance to foreign exploitation and domination. It was tempting to become absorbed in this self-glorifying narrative, with its divisions between heroic victims and villainous perpetrators. It had an instinctive pull and it seduced me with its air of authenticity, which accrued to it precisely because it was the 'local point of view'. I did not fully realise at first that in my eagerness to build a good rapport and in my desire to empathise, sympathise and understand them, that I had to some extent abandoned my sense of criticism. (ibid, p. 12) Yet, many times the positionalities of brokers in the complex and volatile power fields of conflict are far less distinct or obvious. Moreover (and again), positionalities are never fixed or unitary, but multiple and shifting, partly (but not simply) reflecting changing conflict dynamics and allegiances. Moreover, there is a risk particularly in conflict settings to equate broker positionality with ethnic background, language, religion (or other supposed markers/drivers of conflict), downplaying other aspects such as social class. Utas, for instance, experienced this when he conducted a minor consultancy in a diamond area in Sierra Leone together with two research brokers. The study was about livelihoods amongst diamond diggers (many former combatants), right after the civil war. One of the brokers/assistants was an MA student from the same ethnic group as the bulk of diamond diggers, whilst the other was a former combatant from a minority group that was partly referred to in a negative way. Yet, while the former combatant brokered the study in a very smooth and professional way, the student failed to do so, partly by repeatedly alluding to his elite background. Hence, in this case, as many others, the social class and background of the brokers was clearly more central than ethnic background and affiliation in shaping the research process (see Utas in this volume).
The increasing securitization of conflict research addressed earlier accentuates the need to be more attentive to the ways in which brokers shape research processes and results. One aspect of this, addressed in the contribution of Banga et. al. is the impact of humanitarian organisations as brokers. Yet, the main consequence of the increasing securitization appears to be that an increasing amount of research is conducted at a distance and out-sourced to people living at the site of research. Käihkö describes how he has tackled the limited access to the Ukrainian front with the development of new techniques. One of these is what he terms 'chatnography': the use of social media as the central platform or tool, although without making a study of the social media itself. 'Chatnography' enables a certain kind of proximity without sharing the same physical space. Quite the contrary to the article he presents in this special issue, many of his informants in his Ukrainian 'field' he will never meet, some he will meet only briefly and on neutral grounds. Yet, such approaches create additional ethical (as other) dilemmas, as reflected in his account below: I was enjoying a long breakfast and a book when Vadim needed to talk. He had been sick and miserable for a few days, but now the enemy had been lucky and scored a direct hit on their positions with a mortar round. One of Vadim's comrades was seriously wounded in his throat. (. . .) Every so often such interactions interrupted my everyday life, which I had not anticipated would include casualties or indirect fire. How could I do so, considering that while Vadim was fighting the war in an undisclosed place in eastern Ukraine, on this particular day I was sitting in my kitchen in Uppsala, Sweden? (Käihkö 2018, p. 1-2).
Hence, long-distance approaches certainly require a re-centering of current debates and concerns around insecurity and risk, putting the security concerns of locally based brokers at the centre. Yet, the increasing outsourcing or 'long-distance approaches' are, as Myrttinen and Mastonshoeva conclude in their contribution, not necessarily negative but may offer new opportunities for local researchers and brokers and ideally lead to new forms of more emancipatory ways of co-producing knowledge.

Concluding Reflections
Compared to other (less marked by violence and conflict) research settings, it is clear that pervasive insecurity makes researchers particularly dependent on brokers. While we (as researchers on conflict) frequently acknowledge that conducting fieldwork in conflict areas is edgework that comes, at times, with considerable risks (Lee 1995), we rarely talk about the challenges and security concerns, if not dangers, brokers face while trying to facilitate our research. Moreover, we often neglect the obvious: that brokers in a variety of ways influence and potentially bias our data and that we must take their social and political positions into account as well. This silence is problematic from the perspective of both research ethics and research quality.
Drawing upon a broad definition of brokers that acknowledge the fluid boundaries between brokers and researchers, this special issue (including this article) seeks to highlight the particular challenges that conflict and insecurity pose to broker-researcher relationships. As concluded, brokers are often better equipped to assess and navigate risk and often assume the role of protector of the researcher. Yet, ironically the subject most at risk, who also suffers from more long terms threats of violence and intimidation, is most often the broker. This in turn warrants a critical interrogation of the one-sided focus on researcher safety and troubles notions of researcher responsibility. Moreover, we have argued how the (various and interrelated aspects of) inequalities marring researcher-broker relations in many other research contexts are often particularly glaring in conflict settings. The pervasive insecurity also makes othernon financialaspects of reciprocity and compensation such as co-authorship more difficult as this might expose the broker to even more risk. Yet, there is a persistent danger that this argument may be used as a pretext for not including collaborators as co-authors. Finally, we have emphasised how brokers often have a particularly strong impact on research results in conflict settings, due to the crucial role they play in mediating particularly complicated and volatile power fields.
Our review of existing literature in this article clearly calls for more consultation and debates between the various disciplines engaged in conflict research. In particular, much of the literature within the field of conflict research within IR and political science seems quite oblivious to the insights and debates within conflict anthropology. Another problematic tendency in the existing literature highlighted in this special issue is the associations attached to the notion researcher (i.e., white, based in the 'Global North') and brokers (based and originating in fieldsites in the 'Global South'). While we clearly need to recognise the continued marked inequalities in North-South relations in conflict research (as in other fields), this implicit assumption problematically neglects research conducted by scholars based in or (labelled as) originating from the 'Global South'. Moreover, as highlighted in this special issue the boundaries between brokers and researchers are inherently blurred. Finally, let us end by again highlighting the immense variation in possible researcher-brokers positionalities and constellations. As suchand though we have tried to capture some of this variation in this special issueour reflections inevitably remain partial and incomplete. Notes 1. This article and special issue forms part of a research project conducted by Maria Eriksson Baaz, Swati Parashar and Mats Utas funded by the Swedish Research Council. One part of this project is to, in workshop settings, bring research brokers from three conflict settings together in order to in an open-ended manner discuss their experiences as brokers, both in the actual field situation and afterwards. The outcome of these workshops will be a book in which research brokers 'tell their own story', partly inspired by Gullestad (1996). 2. This project was conducted together by Maria Eriksson Baaz, Maria Stern, Chris Dolan and Harriet Grey in Uganda, exploring narratives of particularly male survivors of sexual violence. 3. Assistants to conflict journalists work under similar conditions. Norwegian journalist Morten Ströksnes (2010) has a very good section about imprisonment and other problems that occur to a broker/fixer of his in eastern DRC.

Disclosure statement
No potential conflict of interest was reported by the authors.

Funding
This work was supported by the Vetenskapsrådet [2017-05575].