Trust and Outrage

Trust is the residue of repeated, reliable interactions. It builds slowly, through classrooms, neighborhoods, workplaces, and shared institutions. Outrage, by contrast, is fast. A single divisive post can travel farther in an hour than a carefully reported story travels in a week. In India, where hundreds of millions of people now get news, entertainment, and political identity from the same algorithmic feeds, the speed asymmetry between trust and outrage has become a structural problem.

This article looks at how platform design tilts that balance toward outrage, what the public evidence says about the consequences for social cohesion in India, and why the usual responses—media literacy, fact-checking, and goodwill—are necessary but not sufficient.

Point C1 On major social platforms, engagement-based ranking systematically amplifies emotionally charged and divisive content because anger and moral outrage generate more likes, shares, and comments than neutral or constructive material.

The Outrage-Reward Loop

The mechanism is not mysterious. Platforms that rank content by engagement reward whatever keeps people interacting. A 2021 Yale study in Science Advances analyzed 12.7 million tweets and found that users who received more likes and shares for expressing moral outrage went on to express more outrage over time. The effect was strongest among users in politically moderate networks—precisely the people whose tone is most shaped by social feedback. Point C2 The researchers described it as a learning process: the platform teaches users, through visible rewards, that outrage is the language that travels.

A later large-scale audit by Milli et al., published in PNAS Nexus, compared engagement-based and chronological timelines on Twitter/X. The engagement-based algorithm amplified emotionally charged, out-group hostile content that users themselves said made them feel worse. Among political tweets, 62% in the engagement-based feed expressed anger, compared with 52% in the chronological feed. Point C3 The gap is not marginal; it is the predictable output of an optimization target that treats attention as the only success metric.

Source: Milli et al., PNAS Nexus; Brady et al., Science Advances. Figures are experimental observations on Twitter/X, not India-specific platform data.

India in the Feed

India is not a passive recipient of a global pattern. A 2022 study by researchers at Microsoft Research Bengaluru, published at the International AAAI Conference on Web and Social Media, analyzed 6,000 Indian influencers and 26,000 politicians during political crises. They found that 84% of influencers had a higher median retweet rate for tweets related to polarizing events than for their other tweets. Point C4 In other words, the outrage-reward loop operates domestically, and it rewards public figures who frame politics as conflict.

The consequences show up in harder measures. The World Economic Forum’s Global Risks Report 2024 ranked India highest among surveyed countries for misinformation and disinformation risk over the next two years. Point C5 Separately, the India Hate Lab documented 1,165 in-person hate-speech events targeting religious minorities in 2024, a 74.4% increase from 2023. The report notes that a large majority of these events were amplified through social media platforms including Facebook, X, and YouTube. Point C6

These numbers do not prove that platforms cause hate speech. Offline polarization, electoral incentives, and historical grievances all matter. But the design of the feed determines which speeches find an audience, which grievances get nightly reinforcement, and which communities wake up to targeted abuse in their notifications.

Misinformation and the Trust Tax

Outrage and misinformation are not the same problem, but they travel well together. False or misleading claims that confirm a group fear or insult an out-group spread faster than dry corrections. A cross-country study of viral WhatsApp content by Garimella et al. found that among 158 instances of misinformation identified in India, only one was actively corrected inside the group chat. Point C7 The study was limited to a sample of private groups that consented to research, but the ratio is striking: correction is rare, and misinformation circulates largely unchallenged in the spaces where Indians actually talk.

The result is a trust tax. Every forwarded rumor that turns out to be false, every outrage cycle built on a clipped video, every community-targeting post left standing makes the next public conversation harder. Trust is not just an emotion; it is a form of social infrastructure. When it erodes, coordination becomes expensive. People spend more energy verifying, defending, and withdrawing, and less energy building.

The Accountability Gap

If the problem were only bad actors, clearer rules and faster takedowns might be enough. But the deeper issue is accountability misaligned with scale. India is Meta’s largest market by users, yet the Meta Oversight Board’s 2023 annual report shows that Central and South Asia accounted for only 5% of the cases it selected for review. Point C8 The figure is a proxy, not a precise measure of safety spending, but it captures a familiar asymmetry: the region generates enormous engagement and revenue, while its grievances receive disproportionately little independent scrutiny.

India has its own regulatory tools—the Information Technology Rules 2021, the Digital Personal Data Protection Act 2023, and proposed amendments that would shorten takedown windows. These are necessary floors. But rules written for a broadcast era struggle with a system that optimizes in real time for emotional reaction. Takedowns address symptoms. They do not change the ranking objective that produces the symptoms.

Can Platforms Reward Cohesion?

The diagnosis is not that people are naturally hateful or that technology is irredeemable. It is that the current metric—time and engagement at almost any cost—selects for content that fractures. Alternative designs exist in principle: chronological defaults, user-chosen algorithms, friction before sharing, downranking of outrage-bait, and metrics that reward whether users leave a conversation better informed rather than more agitated.

Some of these ideas are explored later in this series, particularly in Designing for Substance. The point here is narrower: trust and cohesion are not luxuries that platforms can ignore until regulation forces action. They are preconditions for the public problem-solving—education, health, infrastructure, climate—that India and every large democracy depends on.

Sources and Method

This article draws on peer-reviewed studies of platform behavior (Brady et al. in Science Advances, 2021; Milli et al. in PNAS Nexus, 2025; Dash et al. at ICWSM, 2022), international risk assessments (World Economic Forum Global Risks Report 2024), watchdog reporting (India Hate Lab, 2024), and research on private messaging (Garimella et al., ICWSM 2025, arXiv:2407.08172). It also uses Meta Oversight Board transparency data for 2023. Where findings come from non-India samples or limited samples, the text says so.

Open Questions

• Would chronological feeds or user-chosen algorithms reduce outrage amplification in India without increasing exposure to harassment or abuse?
• How much of the rise in hate-speech events is driven by platform design versus offline political organization?
• What metrics could platforms use to reward constructive participation without becoming paternalistic?
• Can public-service broadcasting or community-led moderation fill the trust gap where global platforms underinvest?
• How should Indian regulators balance free expression with duty-of-care obligations in algorithmic systems?

Related in This Series

• The Attention Extraction — the overview and thesis of the diagnosis.
• By the Numbers: What Indians Actually Do Online — the digital time budget that feeds the outrage economy.
• The Generational Bet — why the attention economy is an AI-era stakes question.
• The Substance Builder — individual practice as a counterweight to extraction.
• Designing for Substance — systemic alternatives to engagement-first design.
• Attention, Substance, and the AI Moment — the full series guide.