Research

As social media usage reaches record highs, personalization algorithms risk radicalizing users by reinforcing existing beliefs. However, evidence on how algorithms and user behavior jointly shape harmful online engagement is limited. In this paper, I conduct an individually randomized experiment with 8 million users of a prominent TikTok-like platform in India, replacing the feed-ranking algorithm with random content delivery. I focus on hateful content targeting minority groups, given its prominence on Indian social media and establish a trade-off: random post recommendation lowers exposure to anti-minority ("toxic") content by 27%, but at a substantial cost to the platform as overall platform usage falls by 35%. Strikingly, treated users share a larger proportion of the toxic posts they view, mitigating the decline in the number of toxic posts shared from the platform. Users with a higher interest in toxic content at baseline drive this result as they seek out posts the algorithm does not show them. I rationalize these findings with a model of a revenue-driven algorithm that faces heterogeneous users choosing which posts to consume. Counterfactual simulations evaluate alternative interventions that target toxicity in the algorithm's recommendations. Finally, I collect survey evidence to trace users' behavior beyond the platform and show that the most affected users substitute away to other platforms. These results underscore the limits of piecemeal algorithmic regulation intended to moderate harmful content online.

Inter-group inequality has serious ramifications for economic growth. This paper investigates how ethnic violence and subsequent residential segregation shape children's lives across social groups, thus affecting economic growth. Using variation in communal violence due to a Hindu nationalist campaign tour across India, I show that violence displaces Muslims to segregated neighbourhoods. I exploit exogenous differences in the planned and actual route of the campaign trail to show that communal violence is associated with an increase in residential segregation of communities threatened by violence. Surprisingly, I find that post-event, Muslim primary education levels are higher in cities that were more susceptible to violence. For cohorts enrolling after the riots, the probability of attaining primary education decreases by 2.3% every 100 kilometres away from the campaign route. I interrogate the role of neighborhood effects in driving primary education outcomes across social groups in India.

This paper examines the efficacy of non-invasive interventions in mitigating engagement with politically divisive content on social media platforms. Leveraging a large-scale experiment in collaboration with a major social media platform in India, I introduce "viewpoint-blind" content, nudging users with politically neutral, localized news stories, as an alternative to contentious content moderation practices and biased mainstream media sources. The experiment identifies the causal effect of these nudges on user engagement patterns, particularly focusing on interactions with harmful anti-minority content. Results indicate that increasing the visibility of neutral content significantly reduces engagement with divisive narratives, with a 3% decrease in interactions with polarizing content. This study offers policy-relevant insights for addressing online misinformation while balancing concerns over regulations infringing upon free speech and discouraging overall platform usage. These findings have important implications for platform governance and the design of algorithms that shape information consumption in digital spaces.

The internet, once hailed as a global space for free expression, is increasingly being restricted through government-imposed shutdowns. Despite the internet’s critical role in facilitating independent business operations, social connections, and diverse expressions of opinion, little is known about the nature and extent of these shutdowns. In this study, we focus on India, the world's largest democracy and the country with the most frequent internet shutdowns. Using a high-resolution, novel dataset, we systematically document the geographical distribution, timing, and justifications for these restrictions. Our analysis reveals that while internet shutdowns impact more than 38% of India’s population, they tend to be highly targeted. Notably, these shutdowns are disproportionately concentrated in poorer areas with larger Muslim populations. We also find that shutdown often follow instances of civil unrest, suggesting that the government uses shutdown to supress protest or prevent violence.

How does the identity of local political leaders change public and private expression of political opinions? Using data from a very popular Indian content generation platform, we analyze if social media users are more likely to engage with hateful content when the local leader belongs to a vulnerable minority group. We consider engagement activity of 19 million users with one million Hindi political posts around the village council elections of April, 2021 in Uttar Pradesh, India. This is done to test the existence of a backlash effect when a member of the minority group gains political power. We employ a regression discontinuity design to compare anti-minority hate speech in villages where a Muslim candidate won an election with villages where Muslim candidates lost by a small margin. We find no change in engagement with toxic content in villages where the elected leader was Muslim after the election result was announced. A potential explanation is that engagement with hate is driven by electoral competition and campaigning efforts, and not the announcement of results. We also find evidence that engagement on the platform is generated by national news trends, and platform algorithms that do not take into account user location while generating customized content recommendations. This can make highly localized political conditions less salient for political engagement on social media platforms. This paper provides new evidence on political factors, like local election campaigns and identity of local political leaders, that are expected to change norms of engagement with political and toxic content on social media.