I spent much of yesterday afternoon glued to Twitter, constantly clicking the 20 new tweets link, trying to find credible information about the events in Boston. Let me tell you--between celebrity thoughts and prayers simperings and breathless NY Post reports of double-digit death tolls, it was not easy going. Since the Arab Spring, "Twitter is great for communication during dangerous events" has been a popular refrain. It's true, but what's being communicated during these events sometimes isn't.
The spread of misinformation on Twitter drew a lot of notice from the mainstream media during Hurricane Sandy, when a user named @comfortablysmug started disseminating reports of the Stock Exchange being flooded and Governor Cuomo being trapped and Con Ed shutting down all power in lower Manhattan. Smug's tweets caught the eye of Buzzfeed reporter Andrew Kaczynski who, after confirming their baselessness, outed their author as one Shashank Tripathi, "the campaign manager of Christopher R. Wight, this year’s Republican candidate for the U.S. House from New York’s 12th Congressional District."
Kaczynski's admirable reporting resulted in deserved pillorying and, more importantly, ensured Smug's silence thereafter. But here are the rubs: 1) misinformation, especially of the hyperbolic kind, spreads faster than its corrections, and 2) there are countless Smugs, many of them with a good deal more followers. By the time Kaczynski managed to take a screenshot of Smug's tweet about the MTA shutting down all service for the remainder of the week, it had been retweeted 540 times. Yesterday, the New York Post's tweet reporting 12 dead at the marathon had been retweeted 1722 times as of 5:15 PM. The Boston Police Department's report at that same time: 2 dead (it has since been updated to 3). The police department also brushed aside the Post's report of the suspect being a "Saudi national who suffered shrapnel wounds in today's blast [and] is currently being guarded in a Boston hospital," saying "Honestly, I don't know where they're getting their information from, but it didn't come from us."
I know the Post loves blood, but this is disgusting. More than that, it's harmful, as evinced by Anti-Arab reactions to the "breaking" Saudi tweet. I'm obviously no journalist, but back in high school, I did do a stint at my town newspaper. The very first thing I was told: check your sources. The second: cite them. The third: no confirmation? No story. Such commandments are clearly not in use at the Post, or at Before It's News, or Fox 11, where "Breaking" seems to be the newest euphemism for "spurious allegations."
The thing is, of course, that there were plenty of credible, on-the-ground sources providing information via Twitter yesterday (side note: someone should give about-to-graduate NU senior Taylor Dobbs a job; his reports were and continue to be clear-eyed and comprehensive). So why, then, did people retweet aggregators like Before It's News or truth-be-damned hype machines like the Post?
The answer, in short, is: they saw the rumor tweet first. In "Why Rumors Spread Fast in Social Networks," Doerr et al find that on a social graph, given n modes with a density of more than 1, "after a surprisingly short time a news [story] spreads to all nodes." Generally, a rumor starts with user with a small number of followers. One reason for this is that users with a small number of followers pick up a rumor from one of their followers and quickly pass it on, acting as "an automatic link between neighbors." The second reason is that, once a popular user picks up the rumor, after a few rounds "all popular nodes are informed." Interestingly, a rumor that starts with a small degree node spreads to popular nodes faster, after which the remaining small degree nodes all become informed.
If you skipped all that, the summary is:
Rumor spreading is extremely fast in social networks.
Why rumors, though? In Rumor Psychology, DiFonzo & Bordia says they function "to help people make sense and manage risk," but I'd say the same can be said for any updating news, rumor or true. During a disaster, or any event where facts are presently few, people will scrabble for any information they can find, and spread it so the burden of knowledge isn't theirs alone.
During the UK Riots, the Guardian published a fascinating visualization of the flow--and ebb--of rumors on Twitter. A story like "Police beat 16 year old girl" gets tweeted, generally by someone with a respectable amount of followers, then is picked up by followers and followers' followers and so on. Sometimes, it is then picked up by shame shoddy publication like the Daily Mail; but sometimes, it is questioned, after which point it gradually dies out.
That last bit is a pleasant take away, and it's not unique to the rumors the Guardian studied. Indeed, a recent study of rumor tweets during the Japanese earthquake found that if you call out a rumor tweet as such, you can help kill it. When receivers see a criticism before the original tweet, the likelihood of their spreading the rumor decreases, and the likelihood of their stopping the rumor increases by 150%
We can't predict what we'll see on Twitter at any given point, but we can make sure to follow or keep a list of reputable journalists, and we can try to stay skeptical, especially during high anxiety events.
So, to meld a bunch of highfalutin' advice: in times of trouble, stay classy, readers, and beware the irrational, however seductive.