What makes a tweet believable?
- 7 July 2015
- From the section Magazine
During major news events, how easy is it to sort the tweets from people who are caught up in it from the frauds? Look for tell-tale words and punctuation, writes Alfred Hermida.
At times of natural disasters, terror attacks or unrest, Twitter lights up with first-hand experiences, emotion, rumours and speculation. The bigger the news, the more sensational a story, the more noise there is, the further information travels and the harder it becomes to detect the truth. Yet this is when it is also most important to sort out fact from fiction from the frenzied maelstrom of social media.
That's when swearing comes in. Cussing is one of the clues to figuring out whether a tweet is coming from someone caught up in a major news event rather than a fraud. Letting off a string of expletives seems a natural reaction to a life-or-death scenario.
It is what American software engineer Mike Wilson did after the Continental Airlines 737 he was travelling in skidded off the runway in Denver in 2008. Wilson used his Twitter account, @2drinkafter, to document the experience, becoming what is believed to be the first person to tweet a plane crash.
The tweet announcing the news to the world started off with the word "Holy", followed by the f-word and the s-word, ending "I wasbjust in a plane crash!" (sic)". Wilson now describes himself on Twitter as "Oh who am I kidding... I'm the guy that tweeted from the plane crash."
The f-word turns out to be one of the ingredients in the magic formula sought by scientists studying how to automatically rank the credibility of individual messages. At times of stressful events, such as a plane crash or natural disaster, swear words tend to suggest a message comes from someone in the middle of it all.
Scientists trying to detect the language of truth are less concerned about the actual content of a message. For them, the clues to truth lie in the wording and punctuation of a message.
Spelling mistakes and grammatical errors tend to spark the ire of readers of news sites. But it's a different story when it comes to social media, as Canadian baseball player Brett Lawrie found out during a Saturday in June.
Back in 2012, Lawrie was unwinding with a friend at Toronto's downtown shopping mall, the Eaton Centre, after a game against the Boston Red Sox. As he headed out of the mall, there were a series of bangs. Lawrie didn't know what was going on but joined the shoppers rushing to get out of the place.
On his Twitter @blawrie13 account came some of the first accounts of a shooting spree at the mall that killed two and injured five others. "Pretty sure someone just let off a round bullets in eaton center mall .. Wow just sprinted out of the mall ... Through traffic ..." read his first tweet from the scene.
A minute later, he wrote: "People sprinting up the stairs right from where we just were ... Wow wow wow."
With more than 125,000 followers at the time, Lawrie's quick-fire messages spread quickly. On that Saturday afternoon, the baseball player became a credible source of news, with his tweets cited as proof of the mall shooting both by media outlets and the public. As a public figure, Lawrie would have much to lose if it turned out he was incorrect.
But his tweets demonstrate how language signals authenticity. Lawrie's punctuation and grammar was all over the place so he sounded like someone trying to make sense of the alarming events unfolding around him.
The same mistakes by a journalist would undermine authority. People in the middle of a dramatic breaking news situation tend to focus on specific details of what they are seeing and experiencing.
More hints at the veracity of the tweets come from what's missing. There are no question or exclamation marks. Researchers have found that credible messages tend not to use these. The news is dramatic enough.
Also absent are phrases such as "breaking news" or "confirmed" - the journalese adopted by the media. Messages that tend to mimic what journalists say deserve a degree of extra caution.
One of the most infamous sources of bogus information when Superstorm Sandy tore through the US eastern seaboard in October 2012 was Shashank Tripathi.
Using the alias, @ComfortablySmug, he sent out fake tweets such as: "BREAKING: Con Edison has begun shutting down ALL power in Manhattan." Another read: "BREAKING: Confirmed flooding on NYSE. The trading floor is flooded under more than 3 feet of water."
Tripathi was not a journalist but a New York City hedge fund analyst. The use of journalese should have been enough to cast doubt on his tweets. Why would someone in the centre of a storm use the abstract tone of journalism?
Humans' natural tendency to talk about themselves is a bonus when it comes to personal stories shared on social media. Most people who share their experience of a storm or other dramatic event will use a personal pronoun. There will be a "me," "my" or "I" in the message. It is a way of broadcasting - look at what happened to me.
Essentially, if it sounds like an authentic personal experience, then it is more likely than not to be authentic.
Put together, understanding the language of truth on social media can help to separate fact from fiction. Never mind what a post actually says. Bad grammar and punctuation, journalistic cliches and the expletives are vital signs to evaluate how far something may be true.
But a word of caution. What may be pointers of truth today may not hold in the future, given how fast social media is evolving.
Alfred Hermida is the author of Tell Everyone: Why We Share and Why It Matters, out now. A former BBC journalist, he is director of the University of British Columbia Graduate School of Journalism.
Subscribe to the BBC News Magazine's email newsletter to get articles sent to your inbox.