Power and Privilege Investigating Sexual Violence with Digital Open Source Information

Digital open source information — including the videos and photographs that people post to social media and other publicly accessible platforms — is increasingly valued as a critical source of evidence. While investigators have repeatedly established the value of open source information for researching a range of crimes, there is a subset of crimes that investigators have struggled to address with digital open sources — namely, sexual violence. In this article, we report on findings pulled from our interviews with international investigators and gender experts with regards to the perceived strengths and weaknesses of integrating digital open source information into international criminal investigations of sexual violence. More specifically, we elabor-ate on three insights into how open source investigations may be refined to better respect and protect the interests of survivors: by considering contextual issues related to ethics, power, and privilege, including the identity of the investigator and of the victims; by integrating a gender analysis and an intersectional analysis into online investigation planning; and by being thoughtful about consent, privacy, trauma and control — including who determines what happens with open source information and how such information is used in courts. We conclude with a discussion of what is needed to strengthen the efficacy and ethics of sexual violence investigations through the use of digital open sources.


Introduction
It's 2014. The video opens onto a colourful scene in Kenya. Blurred footage obscures a woman being grabbed, her clothes ripped from her body as she's assaulted by a pack of men who accuse her of 'indecent dressing'. The video quickly makes its way to YouTube, then to Twitter, sparking outrage and igniting the hashtag #mydressmychoice. Donning purple ribbons, both women and men --many of them students --take to the street to protest the attack. Soon after, the director of public prosecutions orders a criminal investigation. Three years later, based on evidence that includes the video, the defendants are sentenced to death. 1 Digital open source information -including videos and photographs that people post to social media and other publicly accessible platforms -is increasingly valued as a critical source of evidence for investigations of atrocities and other crimes. 2  information, a number of challenges specific to such cases have raised concerns among investigators. 6 This includes additional and heightened ethical challenges given the often-sensitive nature of the crimes, gendered access to and use of online platforms, the likelihood that not only the perpetrator but the victim may be stigmatized by the crimes, and the often-coded language used by sexual violence survivors, given those cultural stigmas.
In Section 1 of this article, we introduce the results of an ongoing study into international investigators' experiences with researching sexual violence, and the perceived strengths and weaknesses of integrating digital open source information into their investigations. 7 In Section 2, we provide additional context and pose the question: 'What are the challenges and opportunities for using digital open source information to strengthen international investigations of sexual violence?' In Section 3, we outline our findings regarding how open source investigations may be refined to better respect and protect the interests of sexual violence survivors. This includes considering contextual issues related to ethics, power and privilege, including the identity of the investigator and of the victims; by integrating a gender analysis and an intersectional analysis into online investigation planning; and by being thoughtful about consent, privacy, trauma and control -including who determines what happens with open source information, and how such information is used in courts. In Section 4, we discuss what is needed to strengthen this area of practice, including what our findings suggest for how international investigators can strengthen the efficacy and ethics of sexual violence investigations.

Background
International criminal investigators are increasingly integrating digital open source information into their workflows. 8 This is evidenced in everything from a heightened emphasis on such information in the International Criminal Court's (ICC's) Office of the Prosecutor's most recent strategic plans, 9 to the increased use of digital open source information in ICC cases, 10 to the use of digital information in national war crimes cases, 11 to the development of new resources for the legal investigations community. This includes WITNESS' video as evidence guide 12 and the recently launched Berkeley Protocol on Digital Open Source Investigations, 13 both of which are designed to empower frontline documenters to more effectively produce, capture, analyse and present digital information for courts.
But as such methods have become more common, academics, critics and investigators themselves have increasingly questioned the utility of digital open source information for fact-finding and analysis related to frequently stigmatized crimes, including sexual violence. Such crimes, for example, may not have as obvious an online presence as more 'visible' crimes, 14 and raise numerous ethical concerns, both in terms of how they are investigated, and how that information is used. 15 Practitioners have often indicated that 'rape, amongst other crimes of a sexual nature, is a disproportionately difficult offence to investigate'. 16 Several relatively recent resources seek to provide guidance on documenting and investigating such crimes, in order to help mitigate any perceived or actual disadvantage, and improve ethical interaction with survivors as well as ethical handling of the information they share. Designed for a variety of potential users who may investigate sexual violence as an international crimeincluding international, national, and human rights investigators -such guides include the Second Edition of the International Protocol on the   21 Despite the support these resources provide for strengthening investigations of sexual violence, they provide relatively limited guidance specific to digital open source information. As a result, several practitioners have begun turning their attention to the strengths and weaknesses of using such information in sexual violence cases, merging insights of how to more effectively use digital information with how to more effectively investigate sexual violence. For example, Libby McAvoy and Kelly Matheson of WITNESS have provided a new chapter in the organization's video as evidence guide that outlines how video might be used for documentation of sexual violence. 22 The authors of this article have also reported preliminary findings from their interviews with international criminal investigators and gender experts, pointing to the urgent need for tighter communication and collaboration between those communities in order to strengthen digital investigations. 23 New attention is also being paid to the acute ethical challenges that frequently arise in digital investigations, including issues of power and positionality. This has prompted several resources, including an ethics section in the Berkeley Protocol  To answer that question, this article pulls from 25 semi-structured interviews conducted with international criminal investigators and gender experts between Fall 2019 and Summer 2020. 24 Interviewees were selected based on their experience either investigating and documenting sexual violence using field-based methods, or using digital open source investigations methods for researching a range of potential international crimes, including sexual violence.

Findings
Our interviews produced a number of insights into how digital open source investigations may be refined to more adequately respect and protect the interests of survivors of sexual violence. In this section, we discuss a handful of the most salient. 25 Our findings underscore the need to identify and address contextual power-related issues, including who has the privilege of accessing and using digital technologies. They also emphasize the need to integrate a gender lens and an intersectional lens into investigation planning, as well as the sourcing, verification, analysis and presentation of digital information.

A. Context: Ethics, Power and Privilege
Several of our interviewees discussed the need to carefully plan digital investigations, including where one looks for information and whose voices and experiences are likely to be reflected there, framing that need as both a logistical and an ethical issue. Ethical approaches to digital investigations is a vast topic, the full extent of which cannot be covered here. 26  Researchers Zara Rahman and Gabi Ivens recommend adopting a human rights rather than a utilitarian approach to decision-making, including with regards to using digital materials in investigation processes. 28 They argue that human rights -and thus a sensitivity to the inherent dignity of all involved -should be 'respected and protected' at all stages, including data collection, verification and presentation. 29 An ethical approach also involves understanding a situation's context, who produced relevant information, how the information was obtained, and how it will be used. 30 This point is particularly relevant to questions concerning the risk to, and security of, citizen documenters and the important ethical questions that encourage such actors to engage in investigative activities that may place them, those nearby, or those associated with them, in danger. 31 Several open source investigators who were relatively new to sexual violence investigations were struck, once into their research, by the additional ethical challenges that such investigations raise, noting they require 'much more thought and care' than investigations into other crimes. 32 The extreme and personal nature of the violations and the fact that related media can provide 'a diversity, power dynamics, evaluating risks and harms, cherry picking information, data minimization, sharing and storing data, informed consent, harm minimization, the rights of the victim and the victim's family, the rights of the accused, online scraping, transparency, the use of pseudonyms to access private spaces, hidden labour, the ethics of 'not acting ' lot of information about the survivors and about the perpetrators' had to be factored into ethical decision making. 33 One investigator had been researching a mass atrocity on social media for a major human rights organization and came across several social media posts related to a rape that had allegedly occurred during the atrocity. The investigator explained: 'I think because [the sexual violence content we were finding] was so personal, it made us really start to think ok, what are the ethical considerations behind this? And when are we crossing an ethical line by using this type of material?' 34 Given the power dynamics inherently implicated in an investigationincluding the investigator's power to decide what is worthy of investigation, to determine what sources of information are most relevant, and the vulnerabilities that victims and witnesses may experience when inserted into legal processes -it is important to start by considering the identity and thus the positionality of the investigator and of the victim/survivor.

The Identity of the Investigator
To investigate is to exercise power -to determine which crimes are worthy of the time, money and other resources needed to determine the facts underlying those crimes, and whose voices, experiences and perspectives should be prioritized. Power dynamics are present both between the investigating entity and those investigated, and more generally in the situation under investigation.
Related to this, several interviewees stressed the ethical and logistical importance of paying attention to the demographics of the investigation team in terms of gender, language, expertise and cultural and geographic backgrounds. Interviewees underscored the different lenses investigators bring to their investigations, including inherent biases and the effect those biases may have on cases. 35 They warned that investigations that exclude insights from a genderdiverse team may result in gender blindness. One investigator noted what she perceived as gendered gaps or blind spots in the sexual violence investigations of some of her male counterparts: 'Did they think of going to the birthing clinics? Of course not. Of course not. It just didn't come onto their radar, right? You need to go to the midwives, and you need to go to whatever birthing clinic might be there. You go to them, you're gonna find out all about the sexual violence.' 36 Structural and cultural inequalities in investigation teams can, of course, reflect inequalities that exist in the wider world. Investigators can bring attitudes based on their own life experiences that then crystalize gender constructions that frame and constitute world views, which ultimately affect their approach to the investigation of sexual violence. 37 For example, male investigators may especially focus on non-sexual violence crimes when determining what to investigate using digital sources, which may be much more visible online, while female investigators may be more likely to consider sexual violence. 38 As open source investigations teams are often relatively small, it is especially important for investigators to be sensitive to their possible biases and subjective attitudes to sexual violence, gender, race, violence and even how they approach and use technology. Of course, this also extends to a need to be aware of technical biases: biases may be exacerbated through the use of machine learning for tasks like object detection, which can help with identifying possible war crimes, for example, by automatically detecting prohibited weapons. 39 Sexual violence crimes may not as easily be captured by such tools, thus investigators need to think outside the proverbial box to prevent further marginalization of sexual violence crimes and survivors, as review of large datasets of digital open source information is increasingly facilitated by algorithms. 40

The Identity of the Victim
There are gendered aspects to how people report their suffering and how investigators perceive those reports. Michelle Jarvis, previously Principal Legal Counsel and Deputy to the Prosecutor of the International Criminal Tribunal for the former Yugoslavia (ICTY), has noted gender differences in how male and female survivors tend to discuss sexual violence as well as how that is interpreted by investigators, which can result in differential treatment. She gave an example from the ICTY: [O]ne of our most significant challenges [was] to more accurately recognize the violent nature of rape and similar acts directed against females. The natural tendency was to focus on the sexual component [of what they suffered], rendering the violence component invisible. Arguably, with male sexual violence, there is a danger of the reverse problem. Gendered constructions sometimes resulted in a failure to accurately reflect the sexual nature of the crime. . . . The natural tendency seems to have been to focus on the violence component, rendering the sexual component less visible. For example, the [Office of the Prosecutor] did not take all opportunities to characterize sexual violence against males as rape in appropriate cases and, sometimes, to reference the sexual component of the harm inflicted on male victims at all. Gendered assumptions that only women are raped or subjected to sexual violence in conflict may account for these outcomes. of male victims not to be labelled as 'sexual violence' victims due to adverse community reactions was also a factor. 41 Investigative teams must also have a sensitivity to where people tend to report sexual violence and how that may vary based on gender. The focus should not only be on where weapons were fired or bombs dropped, for example, but whether there was a spike in sexual-related health complaints, whether there are reports on medical or health-related blogs, websites or rape crisis centre websites (if they exist), and which local hospitals, medical centres, doctors and midwives may be potential leads for on the ground investigators.
Thus, integrating considerations of how to counter the biases and other limitations of the investigator with considerations of who the victims and survivors are and how they are communicating, as well as how information about them is being communicated, is critical to incorporate into investigations from the earliest stages.

B. Investigation Planning
Interviewees stressed the need to think critically and creatively about how to find relevant information when planning for digital investigations of sexual violence. Engaging in investigation planning typically means asking a series of questions designed to focus the scope of inquiry. 42 These include: (i) clarifying the investigators' mandate, (ii) determining the strategy and scope of the investigation, (iii) identifying relevant languages, history and geographical knowledge, (iv) thinking through how culture and religion affect behaviour and communications patterns, (v) mapping any relevant political and military groups, (vi) assessing security concerns and (vii) mapping other investigatory actors (who was there before you, who is there now, and who may be coming next?). 43 Of course, all of the non-sexual violence specific uses of digital open source information are also relevant to investigating sexual violence crimes. Any prosecution into which an investigation feeds will have to establish who was most responsible for the alleged crimes as well as satisfy the underlying and chapeau or contextual elements of relevant international crimes. 44 Such context may include facts that establish a nexus with an armed conflict (for war crimes), the systematic or widespread nature of any attack against a civilian population (for crimes against humanity) or the intent to destroy in whole or in part a national, ethnic, racial or religious group (for genocide). This may 41 Jarvis, supra note 37, at 13. 42 For an investigation planning template, see Annex I in the Berkeley Protocol, supra note 2, at 83. 43 This list of questions is for example purposes only and is not exhaustive; Institute for International Criminal Investigations (IICI), Investigations Planning Presentation (copy on file with authors). See also ibid. 44 SGBV Protocol II, supra note 17, at 41-59. require mapping such phenomena as military structures, including communication and operating methods; connections between political and military armed forces and police structures; any preparation for an attack, including public statements on TV, radio broadcasts, internet chat rooms and on social media posts; online review of any publicly accessible military orders, pronouncements, records, logs or orders; satellite or video footage of vehicle and troop movements; announcements of military or police personnel promotions or disciplinary actions; a review of online markets (in human trafficking, weapons, vehicles, etc.) including those present on the surface, deep and dark webs; the origins of any military assets used in an attack, including who could authorize that use; crime patterns (for example, known links between village burnings and sexual violence); and financial investigations, including payments and transfers of assets. 45

Integrating Creativity into Investigation Planning
Some commentators have argued that direct digital evidence of sexual violence, such as online videos or photos, is quite rare, and because of this, the utility of online investigations relative to these crimes is quite limited. However, several have argued the opposite, stressing that the information is out there if the investigator knows where and how to look. 46 This ranges from which platforms or websites hold relevant information, to which part of the web. With only four percent of information indexed to the surface web, versus 90% on the deep web and another six percent on the dark web, 47 the majority of webpages on the internet will not appear in search engine results, suggesting the value of extensive training of traditional investigators in searching 'deeper and darker' digital spaces. 48 One gender expert, in referencing the dark web, explained, 'We do know there's a whole part of open sources in the web . . . that just is banal and dirty, and the horrible side of human nature, right?' 49 It is within these often-skimmed over spaces that investigators may find especially useful information regarding sexual violence crimes. temptation to solely focus on the low-hanging fruit of information posted to the surface web and thus more easily accessed online. 50 The interviewee also noted the desirability of diligence, creativity and imagination in approaching investigations planning that incorporates digital technology: [ Another of our interviewees agreed that digital information can be limited in its utility but also noted the internet's potential for providing supporting or corroborating information: '[T]here's not always a video depicting someone being violated, but [digital information] can be used to . . . lend credibility by corroborating other kinds of evidence. . . . I think, at least for me, it'd be really helpful to be mindful of that aspect of it because I'm not looking for every answer within [online spaces].' 55 The genocide against the Yazidi population in northern Iraq provides numerous examples of how even direct evidence of sexual violence may surface online. As Patricia Viseur Sellers and Jocelyn Getgen Kestenbaum have pointed out, 'ISIS auctioned Yazidi women and children online, replete with registration information, photos, and minimum purchase prices' 56 -a wealth of data potentially relevant to a criminal investigation of sexual slavery. In addition, reporters discovered videos online of ISIS fighters joking about the potential purchase price of Yazidi women and additional information posted to an online magazine published 50  Ultimately, our interviewees seemed to feel that investigators have an ethical obligation to know where and how to find relevant online information to strengthen their work, but to do so in a way that does not distort representation of the crimes as they actually occurred. One especially needs to know what cannot be found online because it is not there, and how to find the information that is there but may be relatively invisible without careful planning and knowledge of where and how to look. This requires and benefits from integrating gendered and intersectional perspectives into the building of an investigation plan.

Gendered and Intersectional Differences in the Use of Digital Technologies
One of the most common pleas from interviewees was the need for investigators -whether using traditional or digital methods -to do a more thorough job of integrating a gender analysis into their investigation planning. 59 To take a gendered approach to an investigation means to think through the ways in which crimes affect people differently based on gender as well as how gender mediates how people report about their experiences, alongside other evidentiary traces. In this context, a gender analysis is a lens through which to examine the gendered nature of sexual violence crimes by analysing structural and other power relations, inequalities and dynamics that determine and shape gender roles in a society. Investigators should adopt an approach that encompasses an understanding of the gendered experiences of individuals and communities within that society. 60 The analysis should include a further examination and understanding of the gendered nature of technology in terms of, for example, use and access.
Applying a gendered analysis approach will be especially valuable for effectively researching crimes under consideration, and even determining which crimes should and can be researched. Much as online investigators ideally conduct a digital landscape analysis to better identify which platforms hold information related to relevant crimes, 61 as well as who shares information about 57  those crimes online, all investigators should design and review their investigation plan with a particular sensitivity to the gendered challenges that come with investigating sexual violence.
One interviewee explained: '[Y]ou need to have an intersectional approach built into your open source investigation plan. In the same way that you would consider the digital mapping environment . . . an intersectional framework needs to be applied to the bias and the questions and the search terms that could go across gender, class, race-ideally, in a participatory way.' 62 Thus, a gender analysis should be complemented with an intersectional analysis that expands beyond gender roles to issues that may further complicate online access and representation. Such intersectional approaches to investigations can include an analysis of, for example, not only race, class, ability, sexuality, education and age, but also other, perhaps less obvious, axes of analysis that may perpetuate vulnerability, oppression, invisibility and marginalization online -and in society more generally. These include location, such as any urban/rural divide in access to technology and therefore access to digital representation. 63  Given how many countries legislate sexual practices, investigators often step into a highly legalistic space when investigating sexual violence. They need to be aware of how stigma and law modify what individuals can communicate, which directly affects what information may be available online. One example would be male sexual violence in situations where reporting would risk being accused of being homosexual, which if illegal, may carry the risk of serious societal and legal consequences for the victim (including death in extreme situations). In such a situation, survivors may be less likely to post online. Thus, investigators should analyze any relevant laws, especially regarding sexual practices and sexual identities, to know what information may be missing or masked online. 65 Investigators should also research norms around families and communities, including cultural and legal responses to rape, pregnancy, adultery and marriage. 66 They will need to understand the extent of the erasure and marginalization of LGBTQIAþ 67 communities in conflict-affected societies but also the potential to further perpetuate the relatively invisible nature of crimes committed against LGTBQIAþ individuals and the marginalization of these communities by a lack of inclusivity in investigation planning. 68 As one investigator noted: [As an] SGBV investigator, for me, anytime you're talking about SGBV, it has to include marginalized communities such as LGBTQI. For me, there's a long way to go to have a holistic approach in relation to SGBV. [That's due to] bias but also [because] in many parts of the world, [homosexuality is] illegal still. It's a crime, so you have the conflict that happens even in trying to investigate in that area. 69 The International Protocol on the Documentation and Investigation of Sexual Violence in Conflict provides a useful documentation plan for conflictrelated sexual violence, parts of which can be helpful to online open source investigation planning. The documentation plan is comprised of three main elements: (i) preliminary issues to research and assess, (ii) an information collection strategy and (iii) procedures. 70 While some of the information contained within the documentation plan may be outside the scope of a digital open source investigation (e.g. interviewing, dealing with intermediaries, etc.), it contains useful questions that may help open source investigators apply gendered and intersectional approaches to their planning. Regarding gender dynamics and the LGBTQIAþ community, for example, the documentation plan recommends considering gender roles, cultural and traditional beliefs, gender expression and sexual orientation, and how these affect reporting. 71 It provides that a documentation plan should include preliminary research into the nature and scope of sexual violence, the patterns of these criminal acts, and how they fit into the overall pattern of an attack. 72 The plan also stresses the importance of considering the 'community understanding of, and attitudes towards, different forms of sexual violence, including against children, the disabled, racial minorities, indigenous communities or members of the LGBTI community.' 73 Investigators should also know what the legal requirements are for proving sexual violence, whether the legal system provides for equality before the law, and whether 'rules of procedure and evidence facilitate or obstruct access to justice for survivors'. 74 Open source preliminary research may also help establish what crimes are reported and to whom, as well as where and how individuals are targeted. 75 As discussed in the next section, all of this will help with figuring out what does not exist online versus what may, and where to look. Thus, the following includes guidance and insights for figuring out what does not exist online (and why), versus what may. Subsection (a) provides questions investigators may need to consider to establish what factors -including cultural, social and legal norms, use and control of digital hardware, architectures and infrastructures, and multiple intersecting modes of oppression (such as ability and access to education) -may lead to the relative invisibility of sexual violence crimes, and how sexual violence survivors communicate online (if at all). Subsection (b) underscores the argument that information relating to sexual violence crimes is available online and outlines an approach to better detect information that may be hiding in plain sight. Finally, we underscore the importance of applying an inclusive approach to investigations, incorporating local documenters, women's rights groups and others.  In order to establish what sexual violence information is not online, investigators need to be aware of multiple, often, intersecting factors that may contribute to online invisibility. They should undertake a comprehensive analysis of these factors and integrate it into investigation planning (including pre-investigation research and analysis). As part of such analysis, investigators should especially consider the relationship between gender roles and access to technology. Do women have access to smartphones at similar rates as men? What are the ruling authorities' attitudes to LGBTQIAþ individuals and their online representation? What are the censorship laws concerning sex, gender, sexual acts, violence, 'offensive' acts and hate speech, and how do they affect representation of women and LGBTQIAþ individuals on the internet? What platforms and forums are most likely to be used by sexual violence survivors?
An assessment of the kinds and level of technology available pre-, during, and post-conflict may be useful as well as any socio-economic barriers to access to those technologies. For example, investigators should analyse the technological infrastructure, including how and by whom it is controlled and whether there are any correlations between this control and the perpetration of atrocity crimes. 76 Such an analysis would include, for example, investigating whether the internet was shut down during the conflict and, if so, in which parts of the country this occurred and when, and whether there are anecdotal reports of sexual violence taking place in those areas during those times.
Moreover, a pattern analysis of population movements can inform whether there may be relevant information online (e.g. noting whether people have been transferred in large groups, under what circumstances those transfers occurred, over what length of time, in what locations, and whether the victims had access to technology during that period, etc.), as well as information regarding the presence/lack of presence of the internet in places like refugee and internally displaced persons camps. An analysis of other possible modes of oppression such as race, ethnicity, education, ability and class can also provide useful information for investigations, such as whether individuals within the affected communities are further stigmatized due to these identities, which may further affect access and online representation, and ultimately access to justice. Trafficking cases provide a particularly rich example of what kinds of information may be found online and what may not, including how that varies by gender. With regards to perpetrators, one investigator noted a strong gender differential in who uses digital technologies and how: '[With] human trafficking [you need to understand] the use of . . . phones and social media by men [and] access of women to those social media platforms.' 78 She explained how digital technologies are often used to facilitate recruitment of women for trafficking, and how other women were often the ones required to use those technologies, so if caught, they would be the individuals targeted --not the men at the top of the hierarchy. 'Those women being edged towards the recruitment process . . . they would be, perhaps, the fall person. . . . The men in the higher levels of business have been removed from that responsibility.
[For example, with labor trafficking] the women were very active on social media, recruiting men, recruiting young [laborers].' 79 The interviewee also pointed out that silences and absences on social media can be important clues pointing to the gendered nature of violence: 'For me, it's been the absence of information [that points to a problem that I need to investigate more carefully]. The real absence of reporting or the absence of women reporting on particular conflict zones . . . is [key].' 80 Ultimately, several interviewees stressed that digital investigators need to think not only about gendered access to digital technologies, but gendered uses of those technologies -and the need to be aware not only of what may be present online and how it may manifest (for example, through coded language 81 ), but to pay attention to absences and silences that may -ironically -be quite telling.
(b) What's there but not easily visible?
Several interviewees discussed ways to better detect online information that may be hiding in plain sight. For example, the investigator who discussed trafficking cases also explained how the means and methods of communication can vary based on whether platforms tend to prioritize text versus images. Comparing Instagram, Twitter and Facebook, she noted: [ Therefore, investigators should conduct research into, and familiarize themselves with, gendered variances in the use of and approach to all identified relevant platforms prior to investigation, and incorporate various investigative approaches accordingly. This may include both text and image searches, which can have dramatically different results. For example, searching for an alleged perpetrator or victim by name and toggling between text search results versus image results can produce substantively different leads; similarly, conducting a search based on the alleged perpetrator's or victim's image or other visual material (if available) can produce different data than text-based searching.
We asked the interviewee above to share her impressions about the cause of such gendered online patterns, which could, perhaps, give clues as to where information can be found in digital spaces -as well as what might be done to level the playing field. She responded: For me, I think it goes beyond access to technology. . . . [That's] important, but [it's too easy to] go, 'Well, the women and men have equal access, so now they should be able to report equally.' When that's not the case [and you make that assumption, your investigation's] flawed from the outset. Even when the equal access is there . . . it's not followed through in terms of women's experience and safety.
These disparate uses of technology further distort information found online. This has borne out in numerous studies, including a well-circulated report from Amnesty International that establishes that certain platforms, Twitter among them, can be hostile to women in ways that men do not experience, requiring women to use the platform in ways that differ from their male counterparts. 83 For example, Amnesty found that many women self-censor on social media in ways that men do not due to disproportionate 'violence and abuse' those women experience online. 84 This was expanded upon in a 2021 report from the Wilson Center that analyses how gender is 'weaponized' against women online.
As part of the planning process, it can also be helpful to be aware of the more 'visible' conflict-related phenomena that correlate with sexual violence. That way, the visible phenomena can act as a signal that evidence of sexual violence may be nearby. Such common indicators include village burnings, takeover and forcible transfer campaigns, the use of detention centres, the use of child soldiers, the separation of males and females, forced or slave labour, checkpoint searches, house searches and large groups of armed men involved in ground operations. 85 An analysis of the position, attitudes to, and role of women in pre-conflict societies may also provide broad indicators as to how they may have been treated during conflict. 86 Finally, NGO and media reporting can act as alarms of likely sexual violence perpetration. 87 Several interviewees underscored the relative paucity of SGBV evidence generally in international criminal cases and the need for more creative thinking on possible sources. One interviewee explained that although investigations may not surface one clear example of sexual violence, 'there are huge flags that always pop up'. In terms of planning, she noted that investigators should 'approach the investigation with understanding there's probably a gendered nature to [any] killing' and thus other crimes that have taken place. She recommended considering 'who was killed and how and was there gender separation of . . . males and females irrespective of age?' 88 She continued: For example [it is critical to understand] who was killed and when they were killed, and why were they killed. . . . Then who was not killed, and why did they survive, and how did they survive? What [was] the next atrocity experience, or gender atrocity experience? It's part of how I would frame the investigation. In addition to . . . identifying physical locations, potential witness leads, potential military evidence or evidence that relates to militias or military forces, et cetera. 89 Several also pointed out how sexual violence is notoriously frequent in prisons, suggesting detention -like killings -should trigger a closer look for evidence of such violence: 'Some armed conflicts are much more noted for detention centers. How you would investigate something at a detention center is markedly different from how you would do it under a type of armed conflict where the criminal incidences were overtaking villages and moving from village to village.' 90 Investigations planning should also include an analysis of the use of language, both in terms of how sexual violence is communicated by survivors (euphemisms, coded language, slang, etc.), how language is used online (often more informal than formal), and how it varies between platforms given different norms and structural limitations. 91 For example, as pointed out above, male survivors of sexual violence crimes may describe what they experienced as 'torture' as opposed to sexual violence in order to avoid stigmatization, and thus mentions of torture should be analysed as a potential proxy for sex crimes. 92 Investigators should also identify how language is being used by perpetrators and whether there are words and phrases, including derogatory terminologies and slang, used by various groups to describe sexual acts or specific individuals or groups (for example, 'get a key' to describe the trafficking of a woman, 93 or 'femoid' to suggest women are sub-human 94 ). A careful examination of relevant hashtags may provide useful lead information but requires caution with regard to possible mis/dis-information (for example, nefarious actors may co-opt hashtags to spread false narratives and misleading data). 95 Local investigators are often best positioned to integrate deep contextualization of many of the intersecting modes of analysis, including relevant terminology, socio-economic realities, positions of vulnerable persons, women, the LGBTQIAþ community and so on. 96 This is why it is so important to work quickly to ensure open source investigation training is widespread among conflict-affected populations. However, until digital open source training is widespread, digital investigations will often be conducted from outside the region, potentially hindering accountability. With this in mind, when international investigators are unable to access the affected territory, how can open source investigators more effectively integrate local perspectives and thereby become more aware of what they do not know and might be overlooking?
One solution for filling critical information gaps is to work with local civil society groups, including gender rights advocates, sexual violence advocates, local (and international) human rights organizations, rape crisis centres and medical groups that have extensive knowledge of the local situation, including pre-conflict realities regarding culture, politics, economics, gender, race, etc. 97 This parallels Libby McAvoy's recent call for re-centering the 'source' in open source investigation by strengthening acknowledgement of the importance of content creators and practicing greater solidarity with those closest to the 92 Investigators should be mindful that sexual violence against men and boys is not restricted to situations of detention or as a form of torture and can depend on numerous variables, including, for example, historical and cultural contexts, the nature of the conflict and the relevant parties. See 98 When asked who should participate in the process of building out an online investigation plan, one interviewee noted the importance of also making sure to include women in developing that plan, since they may have different insights into sexual and gender-based issues than their male counterparts (and vice versa), as well as civil society groups with specialized expertise. 99 Ultimately, several of the interviewees stressed the need to adopt a gendered and an intersectional lens from an investigation's outset, and to think more critically about how sexual violence occurs in a variety of contexts -including which more visible crimes sexual violence often accompanies -so that investigators better know what to look for and can carry those insights into their digital planning.

C. Courtrooms and Consent: Privacy, Trauma and Control
In this section, we dedicate more space than elsewhere for interviewees' firsthand accounts in order to better highlight the varied, nuanced and often personal approaches they have taken to address difficult issues surrounding trauma, consent and privacy. Their comments underscore the need for more clarity on numerous complex ethical issues, more engaged institutional support to help reconcile those issues, and a standardization of ethical investigative approaches to digital information generally. 100 Across the board, there was a lot of uncertainty about the legal and ethical constraints on holding and presenting digital content related to sexual violence. Interviewees were especially concerned about what happens after the investigation stage of the accountability process. Several underscored that investigators should think through how digital information will be introduced in court as well as any disclosure obligations the prosecution may have when deciding what to collect. For example, a couple of interviewees stressed that investigators should be aware of the potential need to bring witnesses to court to testify about information sourced online and the importance of being aware of that obligation when deciding what digital information to use -especially if introducing the source or those depicted in the content in court could be physically or psychologically risky for those individuals. 101 Framing this as an ethical concern, however, investigators also underscored the continued centrality and importance of witnesses to sexual violence cases. While witnesses will always be critical to case building -both from an evidentiary and an ethical perspective -and thus should remain central to any investigation and prosecution, there is also an ethical and moral obligation to minimize witness trauma, as well as a legal responsibility to support witness testimony with other sources of evidence. Ultimately, integrating open source information with witness testimony requires both care and creativity.
Interviewees also emphasized the heightened privacy concerns raised by videos, images, and other online information that includes explicit material, including the need to think through issues of potential retraumatization, and how to provide victims of sexual violence with greater control over what happens with digital materials, when possible. 102 Several described an extreme reluctance to use digital content that depicts sexual violence in ways that might violate the victim's privacy and/or retraumatize the victim or their family. As one interviewee explained: Certainly, if somebody's posted a rape scene [or even a] naked woman, a naked man, somewhere where the privacy of the person has been invaded . . . how do we deal with that? My guess is we still capture it. It still gets categorized. I'm trying to think of a scenario where we would ever use it. Would it be that we are able to identify a victim from that scene and we interview them? But I don't think we would use something [like that] ever without the authorization of or the consent of the victim. 103 She also highlighted these considerations around consent as an ethical issue: To go back to ethics. . . . Maybe you use photographic stills from the video to identify the perpetrator, but [what about] the victim? I bet you would find investigators that would give a different answer to how you deal with that. I would hope that we be protective of the victim. Back to consent, when they haven't given their consent, I would [want to] protect that victim. I wouldn't [want to] show his or her face to a witness. 104 Discussions of privacy also invoked multiple references to the need for data greater anonymization, controlled and secure storage, and team management. One investigator brought up the example of detention-related footage to illuminate these issues and the potential ethical consequences of mishandling video evidence: Another investigator similarly discussed the sharing and using of explicit videos of sexual violence as problematic: She also framed her considerations in terms of ethics: 'There's an ethical consideration there about the victims. . . . [Y]ou could be using lewd and horrible, disgraceful pictures of bodies being mutilated and sexually violated and all of that and you would want to make sure ethically that you need to do that in order to prove your case.' 108 The big takeaway was ultimately to only use what is absolutely needed to achieve justice.
She gave an example where showing a video raised heightened ethical concerns that could be similar to showing footage of sexual violence, given the sensitivity of the material: I remember the mother of a child who was killed in a sniping incident in Sarajevo. The defense team put up on the screen . . . the picture of the kid with the head cracked open and the blood all over. Not that the mother had not seen it before, okay, but they put it up and it was on the screen in front of the mother. Meanwhile, she's having heart palpitations. I'm like, 'Can you remove this from the screen?' . . . That was, I think, beyond unethical. That was like torturing the witness on the stand. You can't undo that. It's done. . . . It was at [her] expense. 109 Consent is an especially tricky issue with digital material that's collected in the context of a remote investigation. The ubiquity of atrocity-related images and videos, and the fact that many investigations are undertaken during situations of ongoing conflict, complicate the securing of consent. Further complications include that many videos and photos posted to social media are captured by non-professionals who may not be thinking about ethical issues such as securing consent, and may not be trained to capture visual information in a manner as dignified as possible -or given the circumstances, may not have the luxury to do so, for example if under threat themselves. Open source investigators need to think how they can get consent to use digital videos and photos from individuals they cannot identify or locate, or indeed whether they need to. If locating such individuals is possible, investigators should consider whether they should contact them and what the inherent dangers are in doing so both for individuals who may be implicated and for the investigation as a whole. Investigators also need to consider the use of such information without consent, including the rights of the individuals portrayed in videos and images, some of whom may be deceased, and whether there are related ethical considerations surrounding the investigator's positionality, power and privilege. 110 One longtime investigator especially stressed the intersection between gender and the need to protect privacy when dealing with sexual violence and the introduction of digital content in the courtroom.
[P]rivacy is absolutely vital. . . . I'm just really aware that there's a lot of power to make certain things visible and other things less visible. Interviewees highlighted potential complexities surrounding access to social media and 'private information' belonging to non-perpetrator parts of their network. Questions including whether the investigating entity needs to obtain permission to use information taken from social media platforms, use digital techniques to protect identity, contact the individual(s) depicted and how to keep such information secure and confidential, all comprise legal and ethical complexities, including concerning consent.
As one investigator explained with regards to using digital information, The confusion extended beyond the parties represented in the videos and photographs, to the party who posted the information. 'Whoever loaded them up -who are they? They're a witness lead to a certain extent. What did they not load up? Can they also -do we need their consent to use [the photo or video]? Or is it now on free space? Or should we go put it more into context and then need their consent for the other contextualized things that aren't there?' 114 She continued, noting that intermediaries such as non-profit organizations sometimes play an intervening role in gathering and storing digital content, then sharing it with investigators. She underscored the importance of understanding how the intermediary secured the digital content and the obligations and understandings that were part of the process: [Q]uestions arise as to . . . what information [the intermediary gave to the] victim or the primary source. What did they tell them the information is being gathered for? Are you within, you know, your rights to use this information for what you would like to use it for? How would you go back and obtain consent? Are you even . . . able to trace the information back to the primary source? Can you rely on a guarantee given by a third party, which says [the creator has] given their consent for [the information] to be used for . . . whatever purpose you want to use it for? 115 Another investigator concurred about the importance of grappling with consent, how it intersects with cultural norms around sexual violence, and how that could be frustrating and, at times, disconcerting for an open source investigator from a northern, western country: We'd look at [documentation of sexual violence online] and get very angry about it, but it's very different on the ground there [than here]. Because of that and because consent is such a huge part of what we think about here, looking at information and gathering it for investigative purposes without consent felt invasive, even though when you think about it, that's open source in general. . . . Not necessarily specific to sexual violence, but really anything, and it's all these thoughts about public property and once information goes out into the internet, does anyone own it anymore? [The way] I have come to view sexual and gender-based violence is that consent has already been taken away, you don't want to take it away any more. It's always 'what does the survivor want,' and investigating it through open source, you are getting none of that. And so it felt sort of icky, and you're like 'let's think about this a little more.' 116 Her co-investigator concurred and noted that they faced an ethical dilemma regarding whether by using digital open source content they were diminishing or eliminating the survivors' right to tell their own stories and share their experiences, especially whilst not having the consent to do so: 113 Interview by the author(s) with Interviewee 20, 15 January 2020. 114 Ibid. 115 Ibid. 116 Interview by the author(s) with Interviewee 22, 3 February 2020.
[W]hen we . . . gathered all this information, we were sort of coming to conclusions about [the information] and trying to analyze it, but then it was through our own bias and through our own lens and then [we put] that out there for the world to . . . know about what was going on. It's kind of in our minds like, are we crossing a line? Like, are we taking away survivors' and victims' right to. . . share something that's so personal when we really didn't even have consent in the first place to start taking their personal information and using it in our investigation? 117 She stressed that ultimately the investigators felt worse having the information in their possession and not doing anything with it, than sharing it with the parties that had requested their support with the investigation: Adding onto [my co-investigator's] point [what complicates matters is that] so much of the information that we got wasn't coming directly from survivors --it was blown up, multiple people reposted or tweeted the same thing [and] it was very hard to find the original content. [Therefore it wasn't] coming directly from the people that it was happening to. There were all of these concerns . . . but there was also the huge burden of like not doing anything at all: if we don't then who are we as investigators? 118 Ultimately, investigators must be clear on legal requirements and obligations with regards to data collection and use. Once legal obligations are clear, investigators must adhere to their professional code of ethics and then to emerging ethical norms related to the use of digital data. However, those norms are still being developed. In response to that, several interviewees stated a need for more ethical guidance around the proper handling of digital open source material -especially that which depicts sexual violence.

Conclusion: What Comes Next
Many of the traditional investigators we spoke with -those who rarely use digital open source information -are afraid that prosecutors and judges will become accustomed to, or over reliant on, digital evidence, threatening to resurrect gender tropes such as 'the unreliable sexual violence witness/victim', and discounting investigations that rely primarily on witness testimonyinstead expecting or weighting digital open source content more than other sources of information.
With this fear in mind, interviewees underscored the importance of ensuring meaningful participation in the justice process for witnesses and survivors: 'We have an ethical obligation to the witnesses to make sure that they have an opportunity to testify and that their evidence is properly heard . . . and that they get justice even though . . . there's no video of them being raped.' 119 One interviewee noted the inherent danger and the possible reinforcement of double standards when it comes to what evidence will be deemed sufficient for proving various crimes. '[J]udges are going potentially to see . . . [the non-sexual violence crime] as having been proved, and the [sexual violence crime as] not, even though there are evidential rules removing the requirement for corroborating evidence of sexual crimes.' 120 Another interviewee, while acknowledging witnesses' well-documented memory challenges and thus the potential corroborating value of digital information, also highlighted the danger of 'machines replacing people. . . . All of this [using digital sources] is to try to get around the fact that we don't believe the witnesses [be]cause they don't remember. It's constant efforts to avoid witness evidence. It's [going to] deny justice. [Over reliance on digital sources] will deny justice to a lot of victims in places where no technology existed.' 121 She went on to argue that a case can be prosecuted fairly, and legitimately, using witness evidence: 'I don't feel that there is any injustice done [by relying on witness testimony]. On the contrary. . . . I think the whole discussion about discarding witnesses is dangerous. Think of all the different conflicts where there won't be any justice.' 122 The concern that 'machines might replace people' emerged a few times during our interviews. However, that is relatively low risk, since digital evidence almost always requires witnesses to testify and lay a foundation for digital content to be introduced in court. Regardless, while those we interviewed varied considerably as to whether they thought the move towards digital open source investigations was more helpful or hurtful, almost all interviewees acknowledged that digital investigations complicate the already complex ethical issues that emerge when investigating conflict-related sexual violence. This strongly suggests that organizations should encourage frequent conversations around ethics, produce ethical guidance and ensure compliance by their teams. 123 If institutional guidance is lacking or investigators are working as sole practitioners, they should still spend time pre-investigation considering the myriad ethical issues that might arise and the possible implications of their actions, and then re-evaluate the ethics of their investigation throughout the duration.
One investigator underscored the importance of being intentional around one's decision-making, especially when there is a lack of formal ethical guidance: As long as [the] thought process has been made, I think that whatever decision you get to is a good decision for you. While the recently launched Berkeley Protocol provides potentially relevant ethics guidance focused on protecting the dignity of witnesses and survivors, as well as ensuring a gender balance with regards to who conducts investigations and whose harms are investigated, 125 additional guidance is urgently needed -especially as it relates to investigations of sexual violence.
Ultimately, technology has the potential to democratize the flow of information and empower those affected by conflict by enabling them to share their conflict-related narratives to bring attention to what they have suffered and potentially advance some form of accountability. 126 But the use of digital technology also presents serious issues surrounding the ownership and control of narratives through the creation, development, ownership and control of those technologies, which are often designed by westerners and thus can be westerncentric. 127 Those who create, control, own and use digital tools may become an inappropriate arbitrator of the narratives, events, crimes and histories of those using the technology to document crimes occurring against them. At its worst, this can create an extractive power dynamic concerning international criminal justice which could disempower or negate self-determination with regards to the implementation of local concepts of justice. 128 It may also create a Western-based lens of a conflict and approaches to redress that overshadow local perspectives and narrative ownership practices.
Thus, it remains critical to ask -especially in sexual violence cases -how investigators can help empower affected communities and minimize any potential damage they may cause. Flowing from that consideration, we recommend further research into how investigators might better centre the interests of populations affected by conflict and minimize approaches that perpetuate outsider-biased narratives when integrating digital open source information into investigative processes.
As several interviewees underscored, open source investigators and their investigations must not objectify conflict-affected populations in ways that serve outsiders to the detriment of those most impacted. One investigator explained: Effectively and ethically researching sexual violence online requires heightened sensitivities to those issues and an expanded set of strategies on the part of investigators. In order to continue to chip away at a legacy of impunity for sexual violence, digital open source investigators must continue to refine their understanding of the diverse ways sexual violence is communicated by perpetrators, victims and bystanders -both online and offline -and how the information those parties share can be used to corroborate or contextualize the stories of survivors. Simultaneously, gender experts must continue to expand their awareness of how online content can be used to strengthen and not diminish victim narratives. To make progress, gender experts, traditional investigators and digital open source investigators should work together to explore and refine the ways in which digital information can be respectfully and powerfully integrated into their workflows to strengthen the quest for accountability.