A Lie Detector Test For Twitter

By Joelle Renstrom | Published

This article is more than 2 years old

twitter hoaxI use Twitter all the time, mostly for news and tips for stories I might want to write up for GFR. I require my students to use it for possible leads on sources for their research papers. On the day of the Boston Marathon, I happened to be on Twitter just as the first explosion occurred — bystanders tweeted about it a good 45 minutes before news outlets reported on it. But then, reports of other explosive devices emerged on Twitter, and other detonations, and all kinds of stuff that simply wasn’t true. It’s tough to draw the line between believing tweets from firsthand witnesses and being skeptical about tweets from people jumping to conclusions. As with all social media, responsible use of Twitter requires a carefully honed bullshit detector, and sometimes not even that is enough. Enter science, and a new system that can tell fact from fiction in our 140-character tweets.

Engineers at universities in England, Germany, and Vienna, as well as companies in Spain, Kenya, Bulgaria, and Switzerland, have created a system called Pheme (in Greek mythology, Pheme used to delight in spreading all kinds of outrageous rumors) that can gauge the veracity of the information in a tweet and trace it back to its potentially false start. All over Europe, researchers are using the program to test real-time statements made on social media with the hopes of preventing rumors.


England has particular incentive to develop such a system — during the 2011 London Riots all kinds of rumors appeared on Twitter, including that animals had been set free from the London Zoo and that a number of famous landmarks were ablaze, causing widespread panic. This is precisely the type of situation Pheme’s creators are hoping to prevent, although I wonder how exactly it would work. Even if Pheme can identify misinformation and the people spreading it, police, emergency services, the government, and other media outlets would have to take to social media to straighten things out, at which point damage may have already been done. Still, spreading the truth could curtail the spread of misinformation and lessen its consequences.

Pheme categorizes rumors as speculation, controversy, misinformation, or disinformation (disinformation is the intentional and malicious spread of inaccurate information, while misinformation is the spread of inaccurate information when the user doesn’t know the information is wrong). Pheme can verify information, and also evaluate sources — it knows that new sources (some of them, anyway) are more reliable sources of information, so it might not vet those nearly as much. On the flip side, it can identify spambots and Twitter accounts with a history of spreading false information.

The engineers think it’ll be 18 months before the final version of Pheme is ready to go. However, they also think they’ll have working prototypes ready before then. The nearly $6 million project was funded by the European Union, who seem to think the truth is actually worth paying for.

Twitter lies

Subscribe for Science News
Get More Real But Weird

Science News

Expect a confirmation email if you Subscribe.