of violent conflict. Low intergroup trust is more commonly found when groups have a history of competition or violence. Low trust in institutions is linked to a history of state governance failures, and particularly relevant to the weaponization of social media, when independent media has been restricted by the state.58 When these types of trust are low, and when alternative sources of verifiable truth are less readily available, people tend to rely upon and believe in information that is provided to them by people on social media who share friendship, kinship, ethnic or religious social ties.59 This phenomenon exacerbates the creation and implications of echo chambers, while limiting the veracity of information on social media. Inaccurate information can be highly inflammatory from a conflict perspective, particularly when tensions are running high. Social media provides a fertile breeding ground for the spread of rumors, a potent example of the conflict risks of inaccurate information spread between friends. Rumors tend to be particularly pervasive in threatening and ambiguous situations, such as periods of social unrest, riot or war. Rumors operate as a kind of collective problem solving in the face of uncertainty, but tend to overemphasize threatening information that promotes defensive or retaliatory actions against perceived villains.60 These types of information have been shown to resonate with others online, distorting and hardening potentially incendiary perceptions of threat (including who is deemed to be the cause). Speed and decentralization make weaponized information of social media difficult to police. Social media communication is a highly decentralized person-to-person technology, with billions of users continuously producing content every hour of every day. The swarm-like nature of social media allows for conversations to evolve very quickly compared with legacy media, propelled by a high number of one-tomany interactions, in which it is very hard to identify and respond to risks promptly or hold any one entity accountable for the nature or impact of the conversation. The conflict risks posed by social media discourse have proven difficult to regulate against, in part because they have evolved faster than government regulators can understand them, and also because of the complexities (both technological and contextual) involved in recognizing contextually specific harmful content, assigning responsibility for its dissemination, and determining the best pathway for addressing it. There are also inherent challenges with relying on the social media platforms to regulate themselves, and ambiguity around how they approach it. These factors combine to limit traditional mechanisms that detect, assess and seek responses to the risks of harmful information, such as watchdog organizations, media associations, public regulators, investigative branches of government and concerned citizens. 58 Pew Research Center (2017). Spring 2017 Global Attitudes Survey. http://www.pewglobal.org/2018/01/11/publics-globally-want-unbiased-newscoverage-but-are-divided-on-whether-their-news-media-deliver/ 59 NATO (2017) Digital Hydra:Security Implications Of False Information Online. Available: https://www.stratcomcoe.org/digital-hydra-securityimplications-false-information-online 60 Difonzo, N. (2010) Ferreting Facts or Fashioning Fallacies? Factors in Rumor Accuracy. Social and Personality Psychology Compass. 4(11):1124 1137. MERCY CORPS The Weaponization of Social Media: How social media can spark violence and what can be done about it 19