Twitter is especially efficient in a crisis. News can spread quickly from people on the ground to the wider world, whether it’s of protests, a natural disaster, or something else. Unfortunately, the information that spreads via Twitter isn’t always correct. Think of the conflicting reports that arise as mass shootings unfold: There are often rumors of a second shooter even if the gunman is alone, and estimated death and injury tolls tend to go up and down several times long before the actual numbers are released.
But a new study [PDF] finds that there’s a bright spot inside Twitter’s rumor mill. Official accounts actually can work to slow the spread of misinformation, according to work by researchers at the University of Washington.
The study (titled “Keeping Up with the Tweet-dashians,” because academia is fun) examines two rumored crises as they unfolded on Twitter. One was a rumored plane hijacking, another a raid on homes in a Muslim neighborhood during a Sydney hostage situation in 2014. Both rumors were false. The researchers traced back the first tweets to appear online propagating these rumors, then looked at how mentions of the rumored event spread over the day.
In January 2015, a rumor of a WestJet flight being hijacked on its way to Canada from Mexico popped up on Twitter after being reported by a flight-tracking website. Within 20 minutes of the rumor showing up on Twitter, there were 400 tweets per minute being published spreading the false news. At that time, only 10 tweets per minute were denying it. But within ten minutes, denials began to gain more traction, peaking at 500 tweets per minute at 4:55 p.m., less than an hour after the initial rumor was first tweeted. These denials mostly consisted of people retweeting two different messages from the official WestJet Airlines account, both confirming that everything was totally normal on the flight. After these tweets appeared, the news accounts that had been spreading the rumor also posted corrections to their Twitter feeds.
In the case of the Sydney raids, tweets affirming the rumor outstripped those denying it for the first few hours of the rumor’s lifespan, until the Australian Federal Police account tweeted out a firm denial of the event. After that, almost all the tweets about the situation were retweets of the AFP account or denials of the rumor, signaling that the official account successfully quashed the spread of misinformation (at least on Twitter). Almost three-quarters of the tweets denying the rumor came from just four accounts, while most of the original rumor can be attributed to only nine tweets, showing the outsized role of a relatively small number of Twitter users.
The main takeaway from the study, besides that the truth can eventually win out on the Internet, is that when it comes to squashing rumors, the sooner a tweet goes out, the better. Information—correct or not—spreads very quickly on Twitter, and organizations need to be able to react fast to correct false statements, because just a few minutes of waiting could mean thousands of tweets featuring false information spreading across the Twittersphere.