ASSESSING THE ROLE OF SOCIAL MEDIA AND DIGITAL TECHNOLOGY IN VIOLENCE REPORTING.

• Author: Roberts, Tony
• Position: Report

1. Capturing Violence Data Using SMDTs: A New Panacea?

   1.1 Introduction

   Our use of social media and digital technologies (SMDTs) is profoundly transforming the way violent events are reported, analysed and acted upon. Social media platforms such as Twitter, Facebook and WhatsApp allow ordinary members of the public to use their mobile phones or other digital devices to broadcast their personal reports widely, or to narrowcast them to private, shared interest groups. These new information and communication tools have been key in reducing previous information asymmetries in contexts of highly unequal confrontations, such as the Arab Spring uprisings (Zhuo et al., 2011), or the anti-globalisation demonstrations in Pittsburgh (Earl et al., 2013). As Specht (2017) has shown, in Colombia increased connectivity and new digital tools help information and knowledge spread rapidly to mobilise opinion and social protest. Social media and digital technologies (SMDTs) provide new formats of social behaviour and means of collective action (Moule, 2016) and are now used for co-ordinating violent forms of collective action by a range of actors, ranging from street gangs (Behrman, 2015), rebel groups and mafias (Carroll, 2013) to "terrorist" networks (Rudner, 2017) and nation states (MacAskill, 2015). Social media and digital technologies are now widely used by state and non-state actors as a means for the monitoring and surveillance of political unrest and armed conflict.

   Social media content and data has revealed forms and types of violence previously invisible and under-reported in traditional media due to reporting biases (Weidmann, 2015; Matthew and Zhukov, 2015). The ready availability of mobile phones with the capability to take photos and video footage and to upload them to Facebook or Instagram enables citizens to provide much more immediate, and detailed, visual and textual evidence of violence than had previously been available via mainstream media. Black Lives Matter and the Occupy Movement are examples of civil society organisations (CSOs) who have used social media successfully to circumvent mainstream media's ambivalence to police violence against African-Americans and growing social inequality, turning social media into a powerful tool for advocacy and collective action. Social media content has also been used to reveal state violence and torture in Abu Ghraib (Amnesty, 2006) and the secret drone wars being waged by the USA in Pakistan, Yemen, Somalia and elsewhere (Strasser, 2013).

   Social media use is also radically transforming the way humanitarians, policy actors and academics engage with violent events, as they offer significant advantages over traditional media, including the ability to capture more instances of more granular information about on-the-ground realities, and the potential to visualise, verify and validate it in near real-time (Sambuli et al., 2013). As information is both crucial and, often, scarce in times of crisis, the capacity to collect real time situational data using such technologies has the potential to significantly improve early warning systems and to provide enhanced information for decision making in relation to security and the targeting of interventions and humanitarian aid. It can also enhance the quantity and quality of "evaluative data," such as monitoring and evaluation systems, internal audit, and accountability procedures (Read et al., 2016).

   Enthusiasm for such technologies in the development and humanitarian sectors has been high. In their report Preventing Violence, War and State Collapse the OECD recommended member states recognise "the critical importance of adopting innovative information and communication technologies for data collection, communication, visualisation and analysis" (OECD, 2009). Similarly, the UN High Level Panel on the Post-2015 Development Agenda called for a "data revolution," which "would draw on existing and new sources of data to fully integrate statistics into decision making, promote open access to, and use of, data" (UN, 2013). With regards to violent crises specifically, one of the most significant advances has been the development of crowdsourcing platforms that enable actors to monitor violent episodes through the aggregation and mapping of social media data. The Ushahidi platform is an early example of SMDT use by local citizens to monitor unfolding electoral violence during Kenya's 2007 elections (Makinen, 2008) and, later, to monitor elections in countries including Sierra Leone, Nigeria and Ghana (Moreno et al., 2017). Crowdsourcing tools like Ushahidi have also been used to map unfolding humanitarian disasters such as earthquakes and typhoons in a practice called Crisis Mapping, again using citizen-generated data to visually map and document unfolding crisis situations (Gao et al., 2011; Meier, 2012). Although a critical review of humanitarian technologies is beyond the scope of this paper, see for example the work of Madianou, Longboan, and Ong (2015).

   Social media and digital technologies have also generated significant enthusiasm in academia, as they give researchers the potential to enhance the quantity and quality of data available for research on violent conflict. As falling costs of digital data collection devices such as tablets and smartphones have reduced barriers to carrying out surveys on violent events in conflict-affected areas. They enable researchers to capture the micro-level dynamics of violent conflict events through the collection of fine-grained, disaggregated data (Blattman and Miguel, 2010; Bruck et al., 2015). They also make it possible to capture the temporal dimension of violent events more accurately, the analysis of which was previously limited by the lack of historical data and the reliance on cross-section type data, which doesn't allow for temporal analysis.

   History cautions that emerging technologies are often accompanied by uncritical over-enthusiasm in which we overestimate the effect of technologies in the short term. Gartner (2016) characterises this phenomenon as a "technology hype cycle" in which an early flush of inflated expectations predictably precedes a later trough of disappointment (Figure 1). Read et al. (2016) have warned about the "data hubris" that has permeated the aid, humanitarian and academic spheres.

   Like any other type of data, data generated through social media and digital technologies is subject to issues of reliability and bias. Data collected through platforms such as Twitter has been shown to be fraught with potential biases resulting from the architecture of these social media platforms and restrictions of access to the data (Boyd and Crawford, 2011: 7). Ownership of these platforms also poses a range of questions with regards to political interference and potential manipulation or alteration of the data. Data on violent events presents a range of additional challenges. Unequal access to SMDTs in conflict affected areas, social norms about what constitutes violence, manipulation and falsification of content by powerful actors which are likely to be exacerbated in violent contexts are among the factors that can generate systematic biases in social media data on violent events. As a result, certain types of violent events can be systematically under-reported in social media data, which can have severe political and humanitarian consequences. Furthermore, the fact that SMDT enables data collection in real time does not make it immune to measurement error, or to inaccuracies of spatial or temporal patterns. Assessing the accuracy of social media data in capturing these patterns as well as the social, political and economic mechanisms that underpin and enable violent events is therefore crucial for its usage in academic research. Furthermore, lower costs of SMDT methods do not entail ease of implementation. In high-risk settings where data collectors can be subjected to violence, threats, repression, and manipulation, data collection comes with significant additional material, logistic, financial, and ethical challenges which have to be carefully evaluated.

   This paper analyses and assesses the relative advantages and disadvantages of the use of social media and digital technologies to collect data on violent events. More specifically, we look at whether SMDT data is reliable for reporting and analysing violent events, the logistical and ethical challenges generated by the collection of such data, and the issues posed for informing policy, humanitarian action and academic research. The report begins by situating this type of method within the growing ICT4D literature, and presents the different types of social media and digital technologies currently used to collect data on violence. In Section 2, we explore the question of reliability and accuracy of SMDT data, identifying key potential sources of biases in SMDT data with regards to geographical coverage, demographic and socio-economic factors, and biases due to the nature and configuration of violent events. We also discuss the potential of such data to accurately capture the social, economic and political processes that underpin violent events. In Section 3, we move beyond the analysis of SMDT data to understand the ways in which it can become "actionable," and identify the logistical, financial and ethical challenges rising from the implementation of SMDT data collection methods.

   1.2 Background: Digital technology in humanitarian aid and development

   The increased use of social media and digital technologies for violence reporting is part of a larger effort to harness new technologies for use in humanitarian aid and international development in what is sometimes called Information and Communication Technologies for Development (ICT4D). ICT4D is a growing field of practice and research concerned with understanding whether, to what extent, and under which circumstances, the use of information and communication...