Conducting effective research into state complicity in human rights abuses

ABSTRACT This paper explores how to conduct effective research into state complicity in human rights abuses. This type of research is challenging: the secretive nature of state violence presents considerable difficulties for the researcher, in terms of both access to evidence, and the safety and security of the researcher and victims. Recent developments in the methods and types of data available present new opportunities for strengthening research. Drawing on our own experience, specifically our work to map the CIA’s rendition, detention and interrogation programme, we aim to show how we have navigated the difficult terrain of human rights investigation. The paper begins by exploring a number of challenges involved. We then discuss recent developments in human rights investigation techniques, as well as the emerging body of critical scholarship that is beginning to shape this kind of work among practitioners and academics alike. We consider some of the imbalances of power that affect this type of research. We then demonstrate how we tried to embrace new opportunities, while being mindful of the risks involved and the limitations of what we can achieve. We close with some reflections on ways forward for this type of research.


Introduction
This paper is aimed at exploring how to conduct effective research into state complicity in human rights abuses, in ways that can help contribute to state accountability. This type of research is challenging: the secretive nature of state complicity in abuses presents considerable difficulties for the researcher, both in terms of access to evidence, and in terms of safety and security for the researcher and for the victims. However, recent developments in the methods and the types of data available present new opportunities for strengthening the quality and scope of research that is possible. While these developments are welcome, old challenges remain, and new ones have emerged. Risks to safety are still considerable, especially for researchers in the Global South, who often lack the resources and protections that scholars working in the Global North enjoy. New methods also give rise to new challenges, since they pose important ontological and ethical questions that need to be navigated thoughtfully. New methods can also be resource-intensive, and they often require considerable expertise if they are to be used effectively. There is a risk that they may play a role in further entrenching global inequalities in human rights investigation.
Drawing on our own experience of research into state complicity in human rights abuses, specifically our work to map the CIA's highly secretive rendition, detention and interrogation (RDI) programme, we aim to show how we have tried to navigate the difficult terrain of human rights investigation. The RDI programme resulted in the kidnap, incommunicado detention and torture of at least 119 prisoners in dozens of countries over several continents, between 2001 and 2008. 1 The highly secretive nature of the RDI programme presented certain challenges; the trauma suffered by prisoners presented others. But the availability of an unexpected source of data, and new methods for analysing and presenting data, meant we were able to provide the most comprehensive picture to date of the workings of the CIA's RDI programme. Our work involved collating and analysing flight data relating to dozens of aircraft known or suspected of involvement in rendition operations, over the period 2001-2008, presented through an interactive map with various in-built search tools. Flight data is not an obvious source of data for investigating human rights abuses, yet proved invaluable in uncovering considerable details about states' covert operations involving extensive human rights abuses. In that sense, it is illustrative of new methods and approaches that can help strengthen human rights investigation in quite powerful ways. By triangulating the flight data with multiple other sources, including victim testimonies, we have been able to significantly enhance the factual record regarding the detention and torture of many prisoners caught up in the programme. This work was made possible through collaboration with a small, international network of human rights investigators and litigators, and we believe that, moving forward, this sort of collaboration is an essential ingredient to undertaking effective research on secretive state practices involving the violation of human rights.
The paper begins by exploring a number of the challenges this type of research entails. We begin by discussing the epistemological and methodological difficulties, before exploring the risks this type of work presents both for researchers and victims of state violence. We then discuss recent developments in human rights investigation techniques, and the opportunities these present, as well as the emerging body of critical scholarship that is beginning to shape this kind of work among practitioners and academics alike. We consider some of the imbalances of power that affect this type of research. We then demonstrate how we tried to embrace new opportunities, while being mindful of the risks involved and the limitations of what we can achieve. We close with some reflections on ways forward for this type of research.

Researching state violence and terror: challenges and opportunities
Where states, particularly liberal democratic ones, are perpetrators of, or are complicit in, human rights abuses, they tend to go to considerable lengths to conceal their activities. The secrecy surrounding such actions inevitably presents difficulties for researchers seeking to investigate human rights violations, particularly when these are contemporaneous to the research. Almost all academic work on human rights violations occurs after the event, and has tended to be heavily reliant on the work of non-academic investigators, such as UN bodies and human rights NGOs. There has been a proliferation of organisations engaging in human rights investigation, all of which deploy different methods and approaches. As Philip Alston and Sarah Knuckey point out, while some approaches are common to many of these organisations, there is little standardisation, or indeed shared written guidance or expert repositories to inform this type of work. Transparency about methods is also in short supply (Alston & Knuckey, 2016a, pp. 4-5;12). In bringing together a range of leading human rights practitioners and academics from various disciplines, Alston and Knuckey have produced an impressive volume which begins a timely debate about how academics and practitioners might better work towards establishing best practices in human rights investigation (Alston & Knuckey, 2016b). In this paper, we draw on some of the key questions they raise, and offer some insights from our own work on how we have attempted to address them.

Uncovering 'facts'
For practitioners and academics having to contend with state secrecy, fundamental epistemological questions further complicate the process of human rights investigation or 'fact-finding'. At one end of the spectrum are those who would suggest there are incontestable facts that can be found, and who argue that certainty is possible. By contrast, and in line with postmodernist approaches, there are those who contest the notion that we can determine any kind of 'truth' or 'facts' (Alston & Knuckey, 2016a, p. 8). There are two issues at stake. The first concerns questions of evidence. In some cases, it may be possible to attribute a killing or human rights violation to a particular perpetrator in light of clear evidence of their culpability. However, 'the scope for contestability grows as the situation becomes more complicated and involves more actors and a broader range of investigators' (Alston & Knuckey, 2016a, p. 8). The second has to do with the politics of the investigation: 'The decision to launch an investigation, the choice of investigators, the actual focus of the inquiry, the resources devoted to it, and so on, all contribute to constructing the "reality" that will emerge from any given fact-finding exercise' (Alston & Knuckey, 2016a, p. 8). Or to put this another way, there are two possible views of fact-finding. First, factfinding relies on a strong notion that facts exist. Second, that fact-finding is a form of strategic practice 'oriented not only toward the production of truth claims but more generally the creation of social consensus' (Mégret, 2016, p. 29). With Mégrét, we share the view that even though human rights investigation has substantial epistemological limitations, that 'facts' are often politicised, that those seeking to establish 'facts' are motivated by certain agendas, and that 'facts' may be used to justify particular actions, maybe even furthering human rights violations, there are still good grounds for seeking to rescue facts from relativism, 'precisely by safeguarding their relativity ' (2016, p. 45). This involves being honest with ourselves and our readers that there is a power struggle over facts, and that human rights investigation as a practice includes 'the ability to legitimize, communicate and strategize about facts' (Mégret, 2016, p. 46). Transparency about how we arrive at our 'facts', and upholding rigorous standards of social scientific inquiry, including using public or transparent methods and recognising and reporting uncertainty, is key to this process. It also allows others to judge our conclusions on the basis of our best efforts (Satterthwaite & Simeone, 2016, p. 330). There is also, of course, an ethics to this: those who seek to deny the Holocaust, for example, do so by distancing themselves 'from the ability to listen to witnesses, hear sad stories of pain and grief, and, fundamentally, empathize with humanity' (Mégret, 2016, p. 47). By contrast, there is a lot to be said for the moral position of cautiously believing in the best attempts at producing facts, because not to do so may mean we become their enemy (Mégret, 2016, p. 47). State accountability for human rights violations depends on a normative commitment to providing the most accurate account possible of state complicity. It also demands a willingness to struggle against state power that would silence inconvenient truths, not simply from our ivory towers, but in collaboration with those practitioners who are best placed to use our scholarship in meaningful processes aimed at securing redress for victims (Blakeley, 2013).

Methods of investigation
Human rights fact-finding traditionally involved the practice of witness interviews by practitioners, and this tended to be the focus of academic literature that grew up around it (Alston & Knuckey, 2016a, p. 12). It was a minority of scholars within the discipline of anthropology who led the way in undertaking ethnographic work on state violence. As Jeff Sluka demonstrates, this occurred not because those anthropologists were seeking to study state violence and terrorism, but rather, because while studying unrelated topics particularly among Latin American communities they witnesses how these communities became targets of state violence during the Cold War (Sluka, 2000a, p. 10). As a result, that minority of anthropologists developed 'a considerable literature on the lived experience of human violence, and the ethnographic study of terror and resistance has become a sub-field of its own' (Sluka, 2000a, p. 11). A number of these have made important contributions to Sluka's edited volume (Sluka, 2000b). Nevertheless, as with other disciplines, those scholars prepared to undertake this kind of work remain a considerable minority, not least because of the considerable risks involved for researchers. Recriminations against academics who have taken on this kind of research are not uncommon. As just one field experience among many, Myrna Mack, a renowned anthropologist who spent years investigating the destruction of rural communities in Guatemala under the US-backed military regimes between 1954 and the early 1990s, was stalked for two weeks by a military death squad before they assassinated her on 11 September 1990(HRF, 2003. Of course, this type of research also increases the risk of recriminations against those individuals who are prepared to provide testimony. Indeed, a key finding of the Guatemalan Truth Commission was that 'for fear of reprisals, a large number of people continue to remain silent about their past and present suffering' (Tomuschat, Lux-de-Cotí, & Balsells-Tojo, 1999, p. 27). There is an important literature that reflects on the role of witnesses and the ethics of engaging them. Paramount concerns include: ensuring adequate protection of informants; ensuring that victims are not re-traumatised by the process of investigation; protecting victims' dignity, privacy and welfare; being sensitive to cultural norms and ensuring that victims are not simply treated as pawns for wider purposes. 2 There are excellent examples of work to investigate human rights abuses through interviews with perpetrators of the violence as well (see, for example, Huggins, Haritos-Fatouros, & Zimbardo, 2002), although as Sluka points out, this, too, is not without its own challenges: senior military officers often ingratiate themselves to the researcher, through shared class, educational and cultural experiences, which can result in a dulling of the researcher's critical faculties. This can, in turn, sometimes lead to empathetic portrayals of perpetrators (2000a, pp. 24-27).
The Truth Commissions carried out in Latin America, including in Chile, El Salvador and Guatemala following years of state terror and violence by repressive military regimes, marked a significant moment in the development of human rights investigation. These processes did not only rely on witness interviews, but drew on a wider range of expertise, particularly forensics, to exhume mass graves and other sites of state violence. More recent developments include the use of surveys, focus groups, GIS and mapping information, and analysis of secondary sources materials such as criminal justice and demographic data. 3 Where possible, some investigators are turning to satellite and remote imagery where available, which can help with tracking the destruction of buildings, such as schools and hospitals. Experts from fields beyond political science, international law and anthropology are also drawn on increasingly, including from environmental science, pathology, statistics, public health and ballistics. Social media and user-generated content are also gaining prominence (Alston & Knuckey, 2016a, pp. 12-13).
We share Alston and Knuckey's view that increasingly sophisticated techniques offer huge potential in facilitating evidence-based advocacy work around human rights abuses. But they also present new challenges. These developments often require a substantial investment of resources, complex coordination of a wide range of experts, and adequate expertise and training in the fast-evolving techniques. As we indicate later, where we lack a sufficient grasp of the strengths and limitations of specific techniques, there is a considerable risk that rather than provide clarity, poor deployment of new methods can muddy the waters, and present a hindrance to producing the most reliable evidence in relation to human rights abuses.

Entrenching imperialism
As we alluded to earlier, motivations for undertaking human rights investigation are many and varied. The US State Department has for years produced annual reports on the human rights performance of all other states in the world. This continued even through the period when the George W Bush administration was engaged in efforts to redefine human rights law to legitimise the use of torture, incommunicado and arbitrary detention, and the denial of habeas corpus rights. This raises an important point about how we approach the state when we undertake this kind of research. Treating the state as a unitary actor means we lose sight of considerable contestation between different functions of the state with regard to human rights. It also means that we risk reifying the state at the expense of exploring the role, whether direct or indirect, of non-state actors in state-sponsored violence, including private corporations. There is a growing critical literature that explores the ways in which the practice of human rights might constitute or further imperialism, and might be used to maintain unjust hierarchies both within and across state boundaries, or preserve the political interests of executive power or other elites within and beyond the apparatus of government. Human rights investigation may also fail to adequately address the underlying causes of human rights violations, and may be poorly suited to addressing global economic inequalities (Alston & Knuckey, 2016a, p. 15).
One remedy to this can be found in a small body of critical-oriented scholarship. Firstly, this work questions a state-centric framing, preferring instead a historical materialist approach which pays attention to capitalist relations of production that structure the international system and have a significant bearing on the exercise of foreign policy. Secondly, it seeks to shine a light on how the use and sponsorship of state terrorism and violence by various agents of liberal democratic states, whether the executive, the military, the security services, or indeed private actors hired in for specific operations, serves a purpose in shoring up capital and maintaining particular configurations of power in world politics, especially US hegemony (Blakeley, 2009;Chomsky & Herman, 1979b, 1979aGeorge, 1991;Raphael, 2009;Sluka, 2000b;Stohl & Lopez, 1984, 1988. It is in this vein that we sought to investigate the workings of the CIA's RDI programme, as well as the involvement of other (neo-) imperial powers in 'War on Terror' abuses, situating their involvement within the context of their historical uses of violence and terror, including torture. 4 A further risk of the development of new methods is that it reinforces a divide between fairly well-resourced academics and human rights practitioners operating in the relative comfort and prosperity of the Global North, and academics and practitioners in some regions of the Global South. The latter often face far greater risks to their safety and security for the work that they do. Institutions functioning to defend their freedom of thought, expression, movement and (in extreme cases) right to life tend to be weaker. They generally have fewer material resources and more limited access to networks within which innovative practices might be shared. Yet they are often those best placed to access first-hand the consequence of state violence and terror. Further reflection and action on how best to build networks that support and include scholars and practitioners from the Global South is needed, a theme we return to in our conclusion.

Researching the CIA's RDI programme
The CIA's use of torture in the War on Terror has received a lot of attention following the publication of the heavily redacted 500-page Executive Summary of the US Senate Intelligence Committee's investigation into the use of torture by the CIA (SSCI, 2014). Details of the extensive torture that prisoners were subjected to were laid bare for the world to read, in a level of excruciating detail that only a small community of litigators, human rights investigators and academics had previously understood. Like many of the techniques used, as the Senate report concludes, these practices rarely had anything to do with securing credible intelligence in the fight against terrorism. Instead, they are intended to degrade, humiliate and dehumanise, irrespective of whether or not the victim has any credible connection to Al Qaida or affiliated terrorist groups (Bellamy, 2006;Blakeley, 2007Blakeley, , 2011MacMaster, 2004). The Senate report also corroborates conclusions that human rights investigators, litigators, and academics working on rendition had long ago reached: very few of those caught up in the rendition programme had anything to do with international terrorism. With the exception of a very small number, most were incarcerated, incommunicado, for years on end, and tortured repeatedly, simply because they were in the wrong place at the wrong time, were victims of erroneous so-called intelligence, or were sold for bounties by corrupt agents of third party states.
Details of the RDI programme were first published in November 2005, when The Washington Post revealed that US and foreign officials had admitted the CIA had secretly detained and interrogated al Qaeda suspects at a facility in Eastern Europe, and that this was one of a number of Eastern European states that were hosting covert prisons for use by the CIA to secretly detain and interrogate suspects (Priest, 2005b(Priest, , 2005aRoss & Esposito, 2005). The story was picked up and covered in the press across Europe, leading to calls for high level, formal investigations into the role of European states in the CIA's RDI programme. Within just a few weeks, investigations were launched by the European Parliament, led by rapporteur Giovanni Claudio Fava (hereafter, EP), and by the Council of Europe, led by rapporteur Dick Marty through the Parliamentary Assembly's Committee on Legal Affairs and Human Rights (hereafter, PACE). These two investigations provided the first comprehensive, yet partial, accounts of the involvement of dozens of European states. The PACE investigation showed that these states were complicit in all sorts of ways, through hosting CIA black sites and Department of Defense prison facilities, or detaining prisoners in their own detention facilities on behalf of the CIA. What connected the complex network of detention facilities were flights by aircraft known to have been used for rendition operations (PACE, 2007).
As we describe elsewhere (Raphael, Black, Blakeley, & Kostas, 2015), we began a project in late 2011 with UK legal action charity, Reprieve, to collate various sets of flight data secured by the EP and PACE investigations, as well as additional data obtained through various Freedom of Information requests and litigation on behalf of specific victims. Through collating and analysing the various sets of flight data, we were able to develop the Rendition Flight Database and interactive map. The database includes 11,000 flights by more than 200 aircraft. In bringing the flight data together, in this way, we have been able to order individual journeys by specific aircraft into 'circuits', and in doing so, have been able to establish which of those circuits include confirmed or suspected rendition operations. 5 Specifically, the database has enabled us to provide the most detailed, publically available account to date of more than 60 rendition operations, where specific circuits have been matched to known prisoner transfers between particular detention sites. In addition to these 60, we have identified more than 200 additional circuits which have included a landing in at least one location where a detention facility is known to have operated as part of the RDI programme.
To determine which suspicious flights formed part of a specific rendition operation, we matched these with various pieces of contracting paperwork between the CIA and a complex network of private contractors, who each provide various logistical services connected to the journeys of each aircraft, in order to be able to link specific flight circuits to the RDI programme. This painstaking work to reconstruct the journeys of individual aircraft has meant we have been able to offer a much more complete picture of the CIA's role, that of other states, and that of numerous private corporations that were paid handsome sums of money to facilitate rendition operations. Outsourcing aspects of the programme, in this way, was key to attempts to conceal evidence of the programme. Uncovering these connections helps to tell a more nuanced story of the international political economy of violence that gets beyond a purely state-centric focus.
In some cases, our work confirms rendition operations identified by previous investigations, but our efforts have also led to us being able to evidence a large number of previously undiscovered rendition operations. We have been able to provide a much more detailed account of when and how the black sites operated, how they were connected to each other and in many cases, who was transferred in and out of them at specific points between 2002 and 2006 . We have also been able to provide a detailed account of Britain's role as a senior partner in the programme. In that case, we explored the extent of involvement and knowledge of different branches of the UK government. This is important for identifying where resistance to UK involvement was strongest, providing insights into how accountability might be best pursued (Blakeley & Raphael, 2016).
Triangulation of the flight data with our analysis of other key data sets, such as victim testimonies from individuals that were caught up in the RDI programme but have since been released 6 and official inquiries such as the Senate Select Committee on Intelligence (SSCI) investigation has been particularly important in providing compelling evidence about the treatment of specific prisoners TBIJ and TRP, 2015). Individuals' accounts often give detailed chronologies of the various detentions and mistreatment that they were subjected to. On their own, however, they are weak as evidence that the detention and treatment occurred in a particular location, not least because prisoners were often extremely disoriented, and deliberately so. US-declassified government documents, such as the SSCI report, the 2004 report by the CIA's Inspector General and a set of official memoranda describing the detention and treatment regimes within the programme, are often heavily redacted, with important information such as details of prison locations withheld throughout the documents. This is particularly the case in the SSCI report, where each of the CIA black sites is pseudonymised, and given a reference colour (e.g. BLUE, GREEN, COBALT). All the information which might enable the identification of the countries in which these black sites were located has also been redacted.
Despite these considerable challenges, it is possible to test the veracity of these sources through triangulation with the flight data. Analysis of the flight data can help determine whether inferences in the declassified documents can be corroborated, and whether the testimonies of victims can be upheld. We have also been able to successfully match the flight data with the content of the SSCI report to confirm the locations that specific prisoners were held, and the movements of prisoners between prisons at specific times. As we describe elsewhere, it has also enabled us to geographically locate the torture of specific prisoners in specific jurisdictions, which has incredibly important implications for documenting complicity of a range of actors complicit in the abuses, as well as for seeking redress and legal remedy . We have, overall, been able to provide what is, without a doubt, the most comprehensive public account of the fate and whereabouts of each of the CIA's prisoners, including capture dates and locations, dates of detentions at specific black sites, rendition operations, and the use of torture at each location.
In turn, our academic work has had significant impact beyond the academy. This has included the provision of expert analysis in a number of ongoing legal cases, on behalf of victims of CIA torture. This has included the triangulation of flight data with analysis of other data sets, as described above. Indeed, flight data analysis such as ours has been used in a number of cases before the European Court of Human Rights, where the Court has accepted arguments drawn from this analysis in various judgements on specific cases of rendition and secret detention. In the case of el-Masri v. the former Yugoslav Republic of Macedonia, for example, the Court found that flight data had 'enhanced the applicant's credibility' in relation to his allegations of secret detention in Macedonia and subsequent rendition from there to Afghanistan, and formed 'prima facie evidence in favour of the applicant's version of events' (ECtHR 2012, paras 156-165, emphasis in the original). In the cases of Abu Zubaydah v. Poland and al-Nashiri v Poland, the regime of secrecy at Guantánamo Bay meant that the applicants were prohibited from giving firsthand accounts of their detention and torture. Nevertheless, the Court determined that the evidence drawn from the flight data was 'sufficiently convincing' to place the applicants in Poland, and accepted that there was 'no alternative explanation' for the flight that had transported them there (ECtHR 2014). Additional cases are on-going at the time of writing, at both the ECtHR and elsewhere, using our analysis of flight data and other documents as independently verifiable evidence to support the central accusations of rendition, secret detention and torture.

Ethics
In undertaking this research, we were confronted with two main ethical challenges. The first was how we engage with rendition victims who had suffered considerable trauma, including prolonged detention, often in solitary confinement, torture and other inhuman and degrading treatment. We, therefore, took a decision not to re-interview rendition victims. Many of them had given their testimonies to human rights investigators who are trained to undertake this kind of work in ways that limit the harm that might ensue from investigative questioning. This is not an area in which either of us is trained, so we opted instead to build relationships with victims' lawyers and advocates, and draw on their knowledge, expertise, and carefully prepared victim testimonies to ensure that we handled victims' stories carefully and protected them from further harm.
The second ethical challenge that emerged related to our relationships with the lawyers representing prisoners subjected to rendition who are now being prosecuted in closed military tribunals at Guantánamo Bay for their alleged involvement in terrorism. These lawyers are forbidden from divulging any details of the tribunals. Should they do so, they would themselves face prosecution. Therefore, while they are permitted to receive evidence from our research, and can ask us questions about our findings, they may not at any time provide any information to us about why they are asking certain questions and what implications this would have for their clients. While they and we share serious concerns about the many ways the tribunals fail to meet due process standards, breaching these conditions would endanger both the clients and the lawyers whose advocacy efforts our work is intended to support. Over years we have built up trust with those legal teams, and our close guarding of this commitment has been critical to the sustainability of those relationships. 7

Using data responsibly
Relating to questions of ethics, we have also had to think carefully about how we use the flight data and the sorts of claims that can be made about it. As indicated above, we were not always able to conclude with certainty that a particular flight circuit was a rendition operation, especially where there is insufficient evidence to confirm that a particular prisoner or prisoners were on board. It has also been important to be clear about the limitations of the data, and the certainty or doubt that surrounds our findings. Therefore, a key feature of the Rendition Flights Database and interactive map that we developed involved categorising and tagging each flight according to the level and quality of evidence we had to conclude that a flight was part of a rendition operation. In being very clear about our degree of certainty, we avoid the risk that we over-count the numbers of rendition operations that occurred.
One of the unforeseen challenges that arose from our development of the Rendition Flights Database was the interest it attracted among several quantitative social scientists. Containing as it does 11,000 individual rows of data, a number of political scientists approached us to subject it to various forms of statistical modelling, on the grounds, they argued, that it could help prove that many more renditions took place than we have been able to prove. This is understandable, since, as Brian Root has argued, there is a growing preoccupation within the social sciences with quantifying the scope of human rights abuses (Root, 2016, p. 355). There are good examples of the type of modelling available (Landman & Carvalho, 2010). Whereas such analyses can provide distinct benefits, helping to 'demonstrate the scope, distribution (over geography and/or time), or variance of a human rights problem' (Root, 2016, p. 356), there are also dangers inherent in deploying such techniques in isolation. As Root argues, a particular challenge lies in understanding 'what can be said with data or statistical findings in regards to description, inference, and attribution of association or causality' (Root, 2016, pp. 356-357). This manifested itself in our case through the assumption by some researchers that there were many more rendition operations lurking in the data than we had found, and that these could be uncovered through modelling. There are serious limitations to this approach, not least because a great many flights our database had no connection to rendition, and one can only determine a positive match where the movement of known rendition aircraft can be matched with non-flight data information, such as black site operational dates and the fate and whereabouts of individual prisoners.
The risks of applying such modelling techniques to the Rendition Flights Database without appreciating the nature of the data and the importance of triangulation is that wildly speculative claims can be made about the scope of the rendition programme. This has, indeed, occurred (Cordell, 2017). In turn, there is a risk that such analysis could undermine the painstaking work that has been done over many years to carefully compile evidence of sufficient quality that it can be used in legal proceedings on behalf of victims. This, we think, provides a cautionary tale about the opportunities and risks that arise from the fast-evolving techniques available to human rights researchers. Ultimately, those who suffer if research is not undertaken cautiously will be those victims of human rights abuses in whose name these social scientific enquiries are ostensibly being undertaken.

Concluding reflections
Those who undertake research on covert state practices which involve human rights violations tend to be motivated by a normative commitment to producing knowledge in order to hold to account those individuals or institutions of government that are implicated. Many state institutions, including those in liberal democratic states, as well as a range of private and semi-private actors are increasingly involved in secretive counterterrorism operations that result in a whole range of human rights abuses, from detainee abuses to extrajudicial killings. The need for high quality and impactful research in this area is urgent, and can best be produced in collaboration with practitioners who share a commitment to upholding universal human rights for all, and to deploying experimental methods to produce the most robust findings possible.
We propose that there are two key obstacles that might deter the sort of work referred to in this paper. The first concerns data, and whether appropriate sources of data can always be identified that can facilitate this sort of research. The second concerns resourcing, both human and material.
As outlined at the outset of the paper, research on human rights violations is particularly challenging because of the secrecy surrounding the perpetration of such acts. A critical reading of our work might lead to the conclusion that we were unusually fortunate, in that a particular chain of events resulted in the air traffic control data being available for our use, and in some senses, the flight data was the 'smoking gun'. Without this, there would have been little we could have done to research the workings of the RDI programme. An alternative view is that research into human rights violations depends on triangulating a whole range of data sources, each of which might speak to different aspects of operations that result in human rights abuses. The RDI programme illustrates how covert programmes that involve such abuses now tend to depend on complex networks of actors, relying on equally complex logistical sets of arrangements, precisely with the aim of evading detection. 'Traces' of such arrangements can often be detected in the public record, if one only knows where to look and has the resources (and tenacity) to begin digging. 8 In our work, obtaining the flight data was significant, but only alongside the engagement with careful on-the-ground investigative work carried out by journalists and human rights organisations. Much was learned through these endeavours before all the flight data were obtained, and in many cases, the acquisition of the official flight data simply confirmed what had already been ascertained. This is important, since it shows that much can be achieved through the collation of findings by journalists, human rights investigators, leaks and official investigations, and that these are necessary in order to convert seemingly innocuous data sources into 'smoking guns'.
The second potential obstacle to this type of research is one of resourcing. Our experience of this work is that it is incredibly labour-intensive, requiring months of careful relationship-building with collaborative partners outside of academia, and years of meticulous and painstaking work to collate the data, to compile appropriate databases and test their validity, and to analyse hundreds of documentary sources to help build a detailed picture of the workings of the RDI programme. In our case, this was facilitated by a research grant, but this covered just a fraction of the human hours invested. We consider the investments into those relationships, and into the compiling of vast amounts of data and translating them into user-friendly tools, to have been extremely worthwhile. The research has had considerable reach, well beyond our usual academic audiences, and has been used for endeavours that we are normatively committed to. We have also learned a great deal about how to ensure that the research we undertake leads to outputs that are sufficiently robust for use in litigation in the highest domestic and international law courts.
There are a number of ways in which the resourcing issues can be overcome. The most obvious are to work collaboratively, building teams of researchers and simultaneously developing a shared mode of working and sets of skills that lend themselves to this type of research. In terms of academics' contributions to such research, more experienced colleagues often have invaluable experience and knowledge, yet tend to be encumbered with a range of institutional commitments beyond research. It is, therefore, incredibly helpful to bring on board more junior researchers, who in additional are often at the cutting edge of the latest research techniques, as well as being well-placed to experiment. Furthermore, given the risks of entrenching the divide between the Global North and South, there are good grounds for looking at how human rights researchers from the Global South can be incorporated into grant applications and research projects. Funding councils in the UK are increasingly making funds available for this type of research. For those of us with commitments to overcoming global inequality within and beyond academia, this is one means by which we can do so. 9 We are also of the view that the most effective teams will include practitioners, from the Global North and South, both to ensure excellence in academic terms and legitimacy and credibility in non-academic circles. In our experience, the best way to approach this was simply to identify relevant NGOs and litigators and enter into a dialogue about how our work could support their efforts. As the Alston and Knuckey volume makes clear, practitioners are very keen to collaborate with social scientists who are able to help them deploy methods that produce high-quality evidence in support of litigation and advocacy efforts. Our experience was no different, for it was the expertise, time and resources that our institutional positions afforded that meant we had something to offer. Our experience has been that this work is incredibly rewarding, and that collaboration with skilled investigators in the third sector has enabled us to produce work that is likely to have far more impact, and be far more robust and compelling, than it otherwise would have been. Notes 1. www.therenditionproject.org.uk. 2. For excellent examples, see Muzigo-Morrison (2016, pp. 175-190), Boutruche (2016, pp. 131-154), Atrey (2016, 155-174, Rothenberg (2016, Tomuschat et al. (1999, pp. 213-230), and Malkki (1995). For more general texts on researching difficult topics with vulnerable research participants, see, for example, Lee (1993) and Davidson and Layder (1994). 3. Forensic Architecture use methods of spatial analysis to gather evidence relating to a range of human rights violations, including attacks on hospitals and reconstructions of torture sites: http://www.forensic-architecture.org/. 4. See, for example, our work on Britain's role in the RDI programme: Blakeley and Raphael (2016). 5. 'Circuits' are comprised of multiple flights by one aircraft from A-B, then B-C, then C-D, and so on, with the aircraft typically returning to A. For access to the database, go to: http://www. therenditionproject.org.uk/flights/index.html. 6. See the various detainee accounts compiled by The Rendition Project. Available at: www. therenditionproject.org.uk/prisoners/index.html. 7. We do not have space to elaborate, but researchers have similar obligations to ensure they do not endanger whistle-blowers involved in exposing human rights violations. There are, of course, also ethical dilemmas relating to whistleblowing that researchers need to consider, including the impacts that leaking of state documents might have for specific named individuals in those documents. Indeed, one concern surrounding Julian Assange's publication of unredacted versions of the Wikileaks files, after his falling out with the New York Times, was that it endangered vulnerable Iraqis and Afghanis who had worked, for example, as translators for the US military. See: http://www.nytimes.com/2011/01/30/magazine/30Wikileaks-t.html? pagewanted=all&_r=0.
8. For an excellent example of scholars able and willing to engage in this sort of research in the context of 'War on Terror' national security operations, see Moore and Walker (2016, pp. 686-716). 9. The Somali First initiative, sponsored by the University of Bristol and with funding from the UK's Economic and Social Research Council, offers an excellent example of the kinds of North-South collaborations that are possible: http://www.bristol.ac.uk/global-insecurities/ transforming-insecurity/.

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

Funding
This work was supported by Economic and Social Research Council [grant number RES-000-22-4417].

Notes on contributors
Ruth Blakeley is Professor of Politics and International Relations at the University of Sheffield. She has published widely on state violence and human rights. Together with Sam Raphael, she directs The Rendition Project which provides the most comprehensive, open-source research and analysis of the CIA's programme for the rendition, detention, interrogation and torture of terror suspects.