Authors: Sebastian Schutte (PRIO), Daniel Karell (Yale), Ryan Barrett (Yale)
Abstract:
How does speech on social media affect communal conflict? While a growing literature has examined the link between online speech and certain kinds of offline hostility, such as hate crimes and protests, relatively little attention has been paid to hostility between social groups, particularly in divided societies. Furthermore, past research has overwhelmingly focused on disinformation and explicit calls for violence, overlooking speech that is seemingly innocuous to outsiders while powerfully evocative in specific cultural contexts. We advance the scholarship on online speech and offline conflict by analyzing 5.1 million user records with an associated 31 million posts and comments from Koo, a social media network popular among India's Hindu nationalists. We combine this data with information on 420 attacks on religious minorities in India for the 2020-2022 period. We find that the frequency of hashtags with a Hindu-supremacist connotation are predictive of greater attacks on Muslims and Christians. We also find that online speech alluring to the overcoming of religious boundaries predicts fewer attacks. These results are statistically robust. Additionally, the observed effects disappear during internet outages, offering further evidence that the effect is mediated through online speech. Our findings shed new light on the role of social media in fostering offline strife by examining communal conflict and a setting from the Global South, as well as presenting evidence that the message of social media content matters for promoting or militating offline harms. Since the content we study conveys primarily values rather than factual claims, it does not classify as hate speech, mis- or disinformation. We therefore demonstrate that censorship and fact-checking can fall short of addressing online speech's negative consequences, while showing that culturally-specific appeals to unity seem to reduce offline violence.