Angry by Design

ANGER travels fast, and has done for thousands of years. Long before algorithms existed, news of a betrayal would move through a village faster than news of a birth, and a single insult could gather a crowd faster than a call to reason ever could. Psychologists call the basis of this the negativity bias: the tendency to rank threats above rewards, because missing a threat costs far more than missing a reward. Going online did not make the negativity bias disappear. It simply gave it a new way to travel.

The modern method of delivery even has a name: "rage bait", Oxford's Word of the Year for 2025. Rage bait is content created specifically to make people angry so that they will interact with it. Most people encounter some form of it every day without realising what it is. Three things tend to confuse people when they talk about it: what it actually is, how it became a business, and how it changes the way people see the world.

More Than Bad Behaviour

Most people think of rage bait as simply people being obnoxious online. The evidence suggests something more systematic. Researchers at Beihang University in Beijing analysed some 70 million posts on Weibo, China's equivalent of Twitter, sorting them by emotion: anger, joy, sadness and disgust. Anger spread between users more widely than any other emotion, rippling out even three steps from the original post, while sadness and disgust barely spread at all.

The title of their 2014 paper sums it up: "Anger is more influential than joy". Whether or not the people posting realise it, anger is the signal that travels furthest, and both platforms and content creators have become very skilled at producing it.

There is also a belief that only clever, media-literate people can resist rage bait's pull. The psychology suggests otherwise. Piotr Winkielman, a psychology professor at the University of California, San Diego, has researched how negativity bias operates beneath conscious evaluation: the emotional response can begin before the reasoning part of the brain has assessed the information at all. Intelligence does not switch off a reflex.

That is exactly why rage bait makes such an effective business model: a trigger that works without having to convince anyone is far easier to pull consistently than an argument that depends on evidence.

How Anger Became a Business Model

None of this was accidental. Internal documents reported by The Washington Post in 2021 showed that from 2017 Facebook's ranking algorithm treated emoji reactions, including the angry face, as worth five times as much as a simple "like". Posts that provoked strong reactions, anger among them, were therefore pushed into more people's feeds than posts that people merely enjoyed. Facebook later cut the weight of the angry reaction, eventually to zero, but content creators had already learned the lesson: manufacturing controversy and staging arguments worked.

The pattern is not unique to Facebook. In 2018, Soroush Vosoughi, Deb Roy and Sinan Aral of the Massachusetts Institute of Technology published a study in Science that tracked around 126,000 true and false news stories as they spread across Twitter between 2006 and 2017. False stories spread further, faster and to larger audiences than true ones, and bots were not mainly responsible: people were. The false stories tended to be more novel and to provoke stronger emotional reactions, such as surprise and disgust. Facebook's design choice, in other words, matched something fundamental about how attention economies work once emotion can be measured and sold.

One consequence follows directly. A platform that rewards strong reactions cannot tell the difference between someone genuinely enraged by an injustice and someone faking outrage for views; both produce the same angry clicks. Once creators began being paid directly for the attention they attract, producing outrage stopped being a side effect and became a profession in its own right. Picking fights, stirring up comment sections and posting deliberately provocative "hot takes" are all far cheaper to produce than genuine insight, and they often pay better.

The Same Mechanism, Different Countries

Because the incentive to go viral is built into the platforms themselves, the mechanism cannot stay confined to one country or one political system. What varies from place to place is what people do with it, and how much resistance they meet.

In India, from 2017 onwards, false rumours about child kidnappers spread through WhatsApp, designed to provoke fear and anger and to be forwarded quickly. The messages were linked to a series of mob attacks in which dozens of people were killed. In July 2018 WhatsApp limited how many chats a message could be forwarded to, in an attempt to slow the spread. Part of the problem was the design of the app itself: because WhatsApp messages are private and encrypted, the moderation tools built for public platforms such as Facebook or Twitter could not be applied to them.

In Myanmar, the story was darker still. A United Nations fact-finding mission reported in 2018 that Facebook had been a useful instrument for those spreading hatred against the Rohingya, a Muslim minority who suffered mass killings and expulsion at the hands of Myanmar's military, and Amnesty International later argued that the platform's algorithms had actively amplified such content.

Facebook itself acknowledged in 2018 that it had not done enough to stop its platform being used to incite violence. It had entered a country with little independent media and very few staff who could moderate posts in Burmese, so the same mechanisms that are partly checked elsewhere ran almost unopposed.

The Philippines shows another version of the same story, this time by design. In a 2018 study, the researchers Jonathan Corpus Ong and Jason Cabañes documented a professional industry of political trolling, in which advertising and public relations strategists, working for political clients, directed networks of paid influencers and fake accounts to shape online debate.

Rather than treating outrage as a by-product of chasing revenue, these networks set out to manufacture it as a political strategy and to drown out opponents. The algorithms, as before, did not distinguish between manufactured and genuine anger. They were never designed to.

Three countries, three very different outcomes, yet underneath each lies the same mechanism: anger is rewarded, and it spreads faster than calm. What changes is who exploits it, and how much pushback they meet.

How Outrage Reshapes What We Believe

The damage goes beyond individual stress or the rarer cases of real-world violence. Over time, rage bait changes how people decide what is true and what matters. Because platforms reward outrage above calm, what rises to prominence often has more to do with its power to provoke a reaction than with its actual importance. Debate, journalism and civic life all depend on persuasion through evidence, which now has to compete with messages that need no evidence at all, just a strong enough hook.

None of this means that real anger at real injustice has disappeared or become less valid. It means that the systems spreading anger cannot tell manufactured outrage from the genuine kind, and that makes it harder for users to tell them apart too.

There are knock-on effects as well. Researchers studying affective polarisation, the tendency to see political opponents as untrustworthy or immoral rather than simply wrong, have found that heavy social media use and hostility towards the other side go together. The direction of cause is still debated: a 2023 study of Dutch adults by Maria Nordbrandt found that people who were already more hostile to opposing parties went on to use social media more, rather than the other way round.

Either way, the result suits rage bait. Angry users seek out the feed, and the feed rewards the content that keeps them angry. Rage bait does not need to change what anyone believes in order to do harm; making people angry enough to dislike those who think differently is enough. Feelings, unlike arguments, do not need evidence to survive.