A Researcher Says Virality Is Lying to You About What's Actually Trending

Key points

• Ruby Thelot, a cyber-ethnographer at New York University, says virality has become a poor guide to real cultural trends because the internet has fragmented into separate, isolated communities.
• Thelot's analysis of roughly 1,000 social media videos with over 1 million views found that only about 25 percent of dating-related content was genuinely negative, suggesting "dating burnout" is overstated.
• He argues that young people on prediction markets like Kalshi and Polymarket are seeking social closeness, not financial gain.
• Thelot's forthcoming book, In Defense of Being-On-Line, publishes in autumn 2025.
• He warns that AI development risks discarding forms of human intelligence, emotional, artistic, physical, that cannot be easily measured.

Ruby Thelot has spent years studying how people behave online, and his message for anyone reading the cultural temperature from a viral post is blunt: stop trusting the numbers.

Thelot, a cyber-ethnographer (someone who studies communities and behaviour on the internet) and media theory researcher at NYU, spoke with Wired about why the metric we've relied on for a decade is now actively misleading us.

Why does virality keep fooling us?

It was gamed. Thelot invokes Goodhart's Law, a principle from British economist Charles Goodhart stating that when a measure becomes the goal, it stops working as a measure. Once "go viral" became the target, creators built content to hit that number, stripping out whatever made the thing culturally meaningful.

A video can reach ten million people and leave no lasting mark. NFTs did exactly that. So did Clubhouse, the audio chat app briefly pitched as the future of social media before vanishing from most people's lives within a year. The view count was real; the cultural shift never arrived.

Thelot adds a structural explanation. The internet has "balkanized," his word for how it has broken into sealed-off communities, each living on its own "digital island." A clip that feels inescapable to one group may be invisible to another. Virality today travels between those islands, not across the whole population.

What does the data say about "dating burnout"?

The narrative is overblown, Thelot says. He examined roughly 1,000 videos on Instagram and TikTok, each with more than 1 million views, looking for genuinely negative takes on dating. About 25 percent qualified. That's a real signal, but hardly the epidemic media coverage implies. Platform design matters too: he found that X (formerly Twitter) surfaces more negative content than other apps, skewing the overall impression. We covered AI's growing role in dating apps in our 6 August story on Ditto's text-message matchmaking service, which now has 150,000 sign-ups, and Thelot's data adds important context to that shift.

What are prediction markets actually about?

Platforms like Kalshi and Polymarket let users bet real money on future events, from election results to a pop star's Super Bowl appearance. Thelot studied young men trading speculative digital currencies called meme coins between 2020 and 2024 and found a near-identical pattern: the gambling was almost always social, conducted in group chats called "the trenches," where friends shared tips and absorbed losses together.

His read on prediction markets follows the same logic. Financial reward is secondary. What people want is a way to feel close to events they can only watch.

"Consumers no longer want to just purely consume," he told Wired. "They want to participate in culture."

For anyone tempted to join, that's worth keeping front of mind: treat money risked the way you'd treat the cost of a night out, something you can afford to lose.

Will AI end the world?

Thelot says no, and his reasoning is pointed. Most human problems aren't held back by a lack of intelligence. They stall because large groups of people can't agree and coordinate. Solving that requires emotional intelligence, trust, bodily skill, none of which current AI systems measure or prioritise.

The people building those systems, he argues, are projecting their own values onto the idea of superintelligence, an AI that surpasses all human thinking. French philosopher Gilbert Simondon warned that each new technology leaves certain human capacities behind. Whole categories of intelligence, artistic, bodily, emotional, may be discarded simply because they're hard to put into a benchmark. That's the part of the AI debate that rarely makes it into the breathless coverage, and it's where Thelot is worth taking seriously.

Common questions

If virality is unreliable, how can I tell what is a real trend?

Thelot suggests looking for adoption outside the community where something first appeared. A genuine trend crosses between different social groups and changes behaviour over time, not just viewing habits.

Does this mean I should ignore social media entirely when trying to understand culture?

Not entirely. Treat high view counts as a starting point, not a conclusion. Ask who is watching, where they live online, and whether the interest is converting into anything lasting: a purchase, a habit, a conversation in the physical world.

What should I make of prediction markets if I want to try them?

Thelot's research suggests the social element drives most participation. Be clear-eyed that the appeal is community and closeness to events, and treat any money involved the same way you'd treat the cost of a night out: something you can afford to lose.