6 The International Journal of Press/Politics 00(0) were widely trusted sources of news when the government censored domestic media, lost credibility.9 Thus, Facebook gained influence at a time when government publications seemed to condone extreme speech and when trust in foreign media was declining. Therefore, Facebook became popular when conditions in Myanmar were ripe for online extreme speech to occur and for disinformation to remain unchallenged. There are several reasons why people are vulnerable to disinformation and extreme speech. McDermott (2019) argues that beliefs in disinformation can be explained in part by people’s psychological propensity toward simple answers. For example, as cognitive misers, people rely on their “intuition” of what constitutes correct information by believing things that they hear repeated often and that circulate in their immediate networks. Moreover, people (in the West and elsewhere) tend to believe information that corresponds to their existing beliefs, thereby eliminating the discomfort of doubt (see Borges and Gambarato 2019). People who are dogmatic or very religious, however, are even more vulnerable to believing false information than people whose cognitive style is more analytic (Bronstein et al. 2019). Given Myanmar’s history of authoritarian rule, people in the country may be even more susceptible to disinformation. Research has shown that rumors play a big role in Myanmar (Tosa 2005). What effect does Internet diffusion have on emerging democracies? In looking at the role of the Internet in Myanmar’s 2007 uprising, Abbott (2012) takes a positive view of the Internet suggesting that it will level hierarchies and facilitate democratization. Bailard (2014), however, offers a pessimistic view arguing that Internet use in low-performing democracies can negatively affect democratic values and impede democratization. Ganious et al. (2015) find that Internet use in low-performing Asian democracies is associated with a decrease in traditional politics. The researchers suggest that increased exposure to dissident or insurgent voices on the Internet lowers people’s allegiant political behavior (e.g., voting) and increases their use of protest behavior, including the use of force or violence. Thus, Internet use in low-performing democracies is—at least statistically—associated with an increase in behavior that can be problematic if it results in violence toward marginalized social groups. One reason Internet use is associated with more extreme opinions and behaviors is because it allows users to avoid information that would challenge their views. A substantial body of research in the United States suggests that individuals engage in selective exposure to information (Garrett 2009; Stroud 2008) when confronted with a “high choice” media environment like the Internet (Prior 2007) potentially opting into information “echo chambers” (Sunstein 2001). Scholars in the United States have warned about negative consequences of this phenomenon for democracy: affective polarization (e.g., more anger toward the political outgroup) (Iyengar et al. 2012; Lau et al., 2017), attitudinal polarization (Sunstein 2001), and diminished democratic deliberation (Garrett 2009). Because these echo chambers can include fabricated information or extreme speech, selective exposure could be one reason for increased ethnic violence.10 But even if people themselves do not engage in one-sided information seeking behavior, the algorithms used by social networking sites are designed to do just that: to present content that mirrors the user’s implicit preferences, including political