A look at Twitter's harassment problem in India, where local language abuse is common and offensive hashtags can appear in the top five trends for hours
Pranav Dixit / BuzzFeed : Tweets: @johnpaczkowski and @daveyalba . Thanks: @pranavdixit See also Mediagazer Tweets: John Paczkowski / @johnpaczkowski : Crucial point from @PranavDixit's latest on Twitter's harassment problem in India, https://www.buzzfeed.com/... pic.twitter.com/ukWq5lT9Y1 Davey Alba / @daveyalba : Abuse-filtering algorithms are more important to Twitter than hiring more humans to check multilingual harassment https://www.buzzfeed.com/... Thanks: @pranavdixit See also Mediagazer
Context & Ripple Effects
Pranav Dixit's reporting documents how Twitter's India problem works at the product level: local-language abuse goes unfiltered because the company bets on abuse-filtering algorithms rather than hiring multilingual human moderators, and the trending algorithm itself can push an offensive hashtag into the top five for hours before anyone intervenes.
The piece became a reference point for everything that followed: experts warning that Facebook, Google, and Twitter weren't doing enough to prevent abuse and political manipulation ahead of key elections in India (election-integrity coverage), Jack Dorsey confronting caste-based harassment on a visit to India (Dorsey's Dalit-harassment encounter), and later reporting that Elon Musk's moderation cuts hit large non-US markets like India hardest (Musk-era moderation cuts).
First-order effects
- Indian users face sustained local-language harassment that Twitter's algorithms fail to catch, while offensive hashtags sit in the top-five trends for hours because the trending system has no effective gatekeeping.
- Twitter's stated choice to prioritize abuse-filtering algorithms over hiring human moderators means non-English markets like India get weaker enforcement than English-language ones.
Second-order effects
- Coordinated troll campaigns learn they can weaponize Twitter's own rules — as when trolls got a reporter's account locked over an old joke — exploiting enforcement gaps while high-profile users receive better support.
- Political actors gain room to run abuse and manipulation campaigns in India, pushing Facebook and Google alongside Twitter into the same election-integrity criticism from experts.
Third-order effects
- If algorithm-only moderation persists, non-US markets become structurally second-class on global platforms, inviting government intervention — visible later in Twitter applying its Synthetic and Manipulated Media policy to content tweeted by BJP's social-media head in India.
- The gap between US-centric moderation investment and multilingual markets becomes a recurring accountability question for platform leadership, from Dorsey's India reckoning to Musk's cuts amplifying hate speech there.
The trend: Global platforms' reliance on English-trained, algorithm-first moderation keeps leaving large non-US markets like India underprotected, turning local-language abuse into a recurring trigger for platform-accountability scrutiny.