Swift's line about falsehood flying was a complaint. In 2018 it became a measurement. Three researchers at MIT, Soroush Vosoughi, Deb Roy and Sinan Aral, published a study in Science that tracked about 126,000 stories shared on Twitter between 2006 and 2017, each one checked by independent fact-checking organizations. Together the stories had been tweeted more than 4.5 million times by about 3 million people. It remains one of the largest studies ever made of how true and false news spread.

What the MIT study found

The results were stark. False stories were about 70 percent more likely to be retweeted than true ones. True stories took about six times longer than false ones to reach 1,500 people. Falsehoods reached deeper into networks, passing through longer chains of sharers, and false political news spread further and faster than false news about science, business or natural disasters.

The researchers then asked why. One answer was novelty, since false stories tended to be surprising, unlike what people had seen before. New information is valuable to share because it makes the sharer look in the know. Replies to false stories expressed more surprise and disgust; replies to true ones expressed more sadness, anticipation and trust.

The study also tested a popular suspect: automated bot accounts. Bots did speed up sharing, but they spread true and false stories at similar rates. The difference came from people. That finding is uncomfortable and useful. If humans drive the spread, then human habits can slow it.

Fair criticism: the study covered stories that fact-checkers chose to check, which tilts toward claims that were already contested and viral. It describes one platform in one decade. Later work has found that most people share very little false news and that a small group of heavy sharers accounts for much of it. The finding is best read as a pattern about what travels, not proof that everyone is gullible.

The pull of moral emotion

In 2017 a team led by the psychologist William Brady studied more than half a million tweets about gun control, same-sex marriage and climate change. They found that each moral-emotional word, such as "shame," "evil" or "attack," raised the retweet rate by about 20 percent. The boost came mostly from sharing within like-minded groups, not across them. Outrage travels well among people who already agree, and the content most likely to spread is often the content most likely to inflame. Since true claims are often duller and more qualified than false ones, a feed that rewards emotion quietly tilts against accuracy.

Repetition feels like truth

A second force works inside each mind. In 1977 the psychologists Lynn Hasher, David Goldstein and Thomas Toppino found that people rated statements as more likely to be true when they had seen them before. This is the illusory truth effect: familiar claims from repetition feel more likely true, because the ease of processing something familiar is mistaken for evidence. Later studies found something more worrying. In 2015 Lisa Fazio and colleagues showed that repetition can sway belief even against what people know, for instance that the short pleated skirt worn by Scots is called a sari rather than a kilt. Knowledge helps, but it does not make anyone immune.

This is one reason that repeating a false claim, even to mock it, can help it along. It also explains why a rumor seen in several places from several friends feels settled, even when every copy traces back to one unreliable source.

The cost of cleaning up

An Italian software developer, Alberto Brandolini, offered a rule of thumb in 2013 that became known as Brandolini's law: refuting nonsense takes far more effort than making it, by his estimate about ten times more. Making up a statistic takes seconds. Showing that it is false means finding the real figure, explaining it and getting the correction in front of the same people. Brandolini's law is not a scientific finding, but anyone who has tried to correct a viral claim recognizes it. It is the economic reason that prevention, through better habits before sharing, matters more than cleanup afterward.

Worked example

During a heat wave, a post claims that the city has banned air conditioners in apartments. It uses the words "outrageous" and "tyranny," names no source, and is shared thousands of times in an afternoon. Run it through the level. Novelty: a ban like that would be startling, which is part of why it spreads. Moral emotion: two charged words, aimed at an obvious villain. Repetition: you see it from three friends, which makes it feel confirmed, yet all three link to the same unsourced post. Bots may be involved, but the friends are real people. A check of the city's website shows the actual rule: a voluntary request to raise thermostats by two degrees during peak hours. The correction is accurate and boring, and Brandolini would predict it reaches a fraction of the original audience. The lesson is to treat the features that make a claim spread as reasons to slow down, not reasons to believe.