Digital Accountability Symposium: Whose Stories Get Told, and by Whom? Representativeness in Open Source Human Rights Investigations

[Yvonne McDermott is a Professor of Law at the Hillary Rodham Clinton School of Law, Swansea University, UK; Daragh Murray is a senior lecturer at the University of Essex School of Law and Human Rights Centre, Deputy Workstream Lead on the Human Rights, Big Data and Technology Project, and Co-I on the Open Source Research for Rights project and Alexa Koenig is Lecturer-in-Residence and Executive Director of the Human Rights Center, University of California, Berkeley School of Law.]

It is no exaggeration to suggest that open source information is changing the shape of human rights fact-finding. Two key benefits are typically identified. First, and as noted by the Special Rapporteur on Extrajudicial, Summary or Arbitrary Executions, open source information has the potential to ‘democratize the process of human rights fact-finding’, by enhancing the ability of ‘ordinary people’ – both in affected communities and from around the world – to participate in fact-finding processes. Second, it is hoped that open source information can mitigate against, or overcome, biases associated with traditional investigative methodologies. For example, witness interviews are an invaluable source of detailed information but are vulnerable to selection biases. The availability of alternative sources, such as user generated photographic and video footage, may facilitate a more objective examination of the facts as they occurred.

However, an overly optimistic rush to adopt open source information risks overlooking some very real biases and blind spots. This post reflects on some of the issues of representativeness that can arise, including: the types of violations reported, the victims and witnesses who have the opportunity to have their voices heard, and how narratives of mass human rights violations are constructed. We also discuss the impact of open source evidence on advocacy, and how it may be inappropriately prioritising certain types of violations. Much of this post is based on qualitative research conducted as part of an ongoing project into the use of open source information for human rights fact-finding.

As shall be seen, open source research has the potential to profoundly affect not just whose stories get told, but also who gets to tell those stories, and who will listen to them.

Mind the gap(s): Open source evidence’s blind spots

It is a truism that not everyone has access to the tools necessary to capture and disseminate evidence of human rights violations. A number of factors are relevant in this regard, including gender, income, age, and internet penetration. As such, while user generated content can broaden participation in fact-finding beyond traditional professional investigators, large groups of people – many of whom are among the most vulnerable – will continue to be excluded.

As well as the ‘who’ of user-generated content, there is a real risk of global bias in relation to the ‘where’ of open source investigations. The affordability of smartphones and mobile internet is subject to drastic global variability, and is often compounded by politically-motivated internet shutdowns (such as in Kashmir). In some countries, access to 1GB of data can cost up to 20% of the average monthly income, while many areas simply do not have the infrastructure necessary to enable local people to share information online.

There is also a gap when it comes to whatviolations are documented. Videos and images (generated by both perpetrators and eyewitnesses) tend to focus on particular – often visually dramatic – incidents, such as killings, the use of force at protests, or air or artillery strikes. Other incidents, such as gender or sexual based violence, torture, or starvation as a method of war, are less likely to be documented using open source methods.

There is evidently a risk that investigations which focus primarily on open source information may ignore important violations that are simply not visible to open source investigators. This, in turn, can further silence individuals – such as women, persons with disabilities, or poorer or less educated persons – that are already under-represented. Even in situations where user generated content exists, getting that information out can be difficult. As Jay Aronson has observed, having your voice heard in a cacophonous social media environment can depend heavily upon tech-savviness, having an extensive social network, or just plain luck.

Shaping narratives

Ari Gandsman has written of the risk that victim interviews can entrench conventional narratives, resulting in a resistance to anyone sharing an experience or perspective that deviates from that norm. One might expect that the relatively objective and independently verifiable nature of video or image evidence, which decouples evidence of the violation from witnesses’ accounts, could play a powerful role in protecting against the reproduction of these accepted or dominant narratives. Interestingly, however, investigators we interviewed reported that, at times, open source evidence has in fact shaped witnesses’ accounts and impressions of what they saw: if a photograph or video has been widely disseminated, it can – consciously or unconsciously – influence and inform a witness’s testimony. For example, some witnesses have expressed an expectation that investigators will want to talk about a particular incident that has been widely shared on social media, or will indicate that an incident is important and worthy of further investigation precisely because of its wide dissemination. This means that open source information can also contribute to the emergence of a dominant narrative, in a similar vein to witness interviews. Investigators need to be mindful of the impact the social media environment can have upon witnesses’ perceptions of events.

Overcoming issues of (under-)representativeness

This leads us to consider the extent to which open source evidence’s blind spots can be overcome. It could be argued that, because no investigation is purely based on open sources, but instead works hand in hand with traditional investigative techniques, there is nothing to worry about. On the other hand, open source research is often used as ‘lead evidence’, which can then be followed up on in field investigations and corroborated by evidence gathered through more traditional means. This leads to a risk that investigations will be inadvertently shaped in a biased manner.

Of course, human rights investigators, mindful of (often context-dependent) ‘hidden’ human rights violations, can come up with counter-strategies. For example, the report of the United Nations Commission of Inquiry on the 2018 protests in the Occupied Palestinian Territory noted that, while the proportion of men and boys killed or injured in the demonstrations was much higher than the number of women or girls killed or injured, there was a huge impact on women in the form of domestic violence, an additional burden of debt or becoming breadwinners, and so on. Those aspects would have remained hidden had investigators not recognised that in a patriarchal society such as Gaza, there was a need to go beyond the visually dramatic evidence to determine the gendered impacts of human rights violations.

Analysis and Advocacy

A notable finding of our research to date relates to the impact that open source evidence has borne on advocacy and policy-making. In a sense, open source research risks becoming a victim of its own success. The widespread sharing of content like this video of summary executions in Cameroon – which was viewed over 1.5 million times – has led to something of an expectation that advocacy on human rights violations will be accompanied by illustration of those violations through open source evidence. It could be that this is a product of our time-poor, information-dense, age, where attention spans are short. As one of our interviewees put it, “we’re all getting pretty tired of narratives”. In that context, the ability to use open source information to visualise violations (as in the powerful work of Forensic Architecture and SITU Research), to aggregate and disaggregate big data, and to present these findings in compelling ways can engage a broader cross-section of people. On the other hand, as both Rebecca Hamilton and Ella McPherson have noted, there is a risk inherent to this, in that humans have a cognitive tendency to process visual content as being more convincing or credible than written or oral evidence. It goes without saying that this gives cause for concern, in an era of ‘deep fakes’ and deliberate misinformation. There is also a danger that the new advocacy environment will drive investigators and organisations to focus more on the types of violations and the geographic regions where this supplementary evidence can be sourced, and this risks further perpetuating the issues of representativeness (or lack thereof) outlined above.

More broadly, the ability of open source information to strengthen advocacy efforts and engagement with policy makers is challenging, at a fundamental level, the epistemology of human rights fact-finding. Where in the past, the testimony of a credible eyewitness (especially if corroborated by another piece of evidence or testimony) was seen as the gold standard, investigators tell us that there is a growing feeling that ‘just’ testimony, no matter how powerful, will not be enough. This is deeply concerning, particularly for those organisations that do not have the capacity (in terms of time, money, expertise, and technological capacity) and inclination to engage with the difficult processes of collection and verification of open source evidence. There is a risk that this will cleave the community of activists and organisations into those wealthier, larger NGOs who can meet the expectations of the new advocacy environment and those that cannot. In an era where human rights are under increasing threat, and collaboration is key, this could be a tragic unintended consequence of the type of evidence that was once hailed as the great democratiser of human rights investigations.

The authors are investigators on the OSR4Rights project, funded by the UK’s Economic and Social Research Council, which seeks to examine the extent to which open source information is used in practice for human rights investigation and documentation, and the challenges and opportunities posed in dealing with this type of material.