On Wednesday, Feb. 25, a Scottish 20-something posted a picture of her friend’s mother’s dress on the social media platform Tumblr. The blue-and-black dress, which appeared white and gold (to some) in the now-infamous photo, would become a hot topic on Buzzfeed, Twitter, even morning talk shows. Weirdly, for a day — maybe a bit more — the dress was the topic in pop culture. It was inescapable.
Turns out that the epicenter of that digital earthquake was a Tumblr user called mushroombeast. “That [expletive] white/gold/blue/black dress on Tumblr has ruined my life,” she tweeted.
Devin Gaffney, a PhD student in network science at Northeastern University, thinks that was the tweet that started it all.
Gaffney, 27, tracks viral cascades, a term used to describe “the phenomenon of content spreading quickly and widely through a human social network via its digital shadows.”
In other words, the life and death of a meme. By tapping into hard-to-access Twitter streams, Gaffney found what he believes is the tweet that started it all, and graphed the quick rise and sudden fall of the godforsaken garment. This isn’t the first time he’s dug deeply into a viral cascade; he similarly tracked the way the news of Osama bin Laden’s death spread on Twitter, making headlines in the business press.
Overall, the dress followed the typical tidal pattern of digital waves: one crest right as the dress began to get buzz, another once the blogs started cranking out listicles and contextualizations.
“The media, in particular media that relies on Web traffic and click through, has a tendency to cash in on viral cascades,” Gaffney said. “It pays to be first, it pays to write early and often about what looks like it may take off on the Internet. In many ways, they get to determine what the story is going to be.”
A clearer understanding of viral cascades is in high demand. Companies like SocialFlow — Gaffney’s former employer — are capitalizing on corporations hungry for the secret of virality. People like Gaffney could figure it out
Gaffney tracks those patterns: the way information surfs conversations between groups, cuts across oceans through retweets and replies. With the help of a few pre-written programs, Gaffney can collect global conversations in searchable data sets while he goes for a coffee run.
“I have probably 30 to 50 programs running right now, collecting things, cleaning things, transforming things, counting things. . .” Gaffney sat at a table in the corner of Somerville’s Diesel Cafe, iced coffee ignored. “I was thinking I might want to play video games this weekend, but I probably won’t because I have to let [the computer] sit there and work.”
Gaffney is an Internet native. He grew up exploring chatrooms and Internet radio stations, playing games and making friends online.
“I derive great pleasure in finding a quiet corner of the Internet to call home,” Gaffney said. “I’ve always been endlessly fascinated with the Internet.”
That hasn’t changed; now, however, he has the vantage point of a sociologist. With a master’s degree from the Oxford Internet Institute, a department within Oxford University, and work in several publications, Gaffney can study online behavior from the comfort of home.
“[Twitter] is problematic, it’s messy, but it’s a mirror of some form of human communication,” Gaffney explained.
“I derive great pleasure in finding a quiet corner of the Internet to call home. I’ve always been endlessly fascinated with the Internet,” said Devin Gaffney. (above)
Gaffney has spent years playing detective in the backlogs of Twitter. With the help of prominent computer scientist Gilad Lotan, he created a timeline of how Osama bin Laden’s death in 2011 became common knowledge on the Internet. Within 38 minutes of a tweet by former secretary of defense Donald Rumsfeld’s secretary about bin Laden, Twitter was ablaze — even before President Obama had spoken to the public.
So I'm told by a reputable person they have killed Osama Bin Laden. Hot damn.— Keith Urbahn (@keithurbahn) May 2, 2011
Gaffney remembers being amazed by the speed. “It only takes 30 minutes to tell everyone on the planet something? That’s cool,” Gaffney said.
But no programmer forgets his first data set. A data set can be any collection of information, but in this case we’re talking tweets. When he was a senior at Bennington College, Gaffney was fascinated by Iranian politics. Inspired by a political science professor and an interest in computer science, he began an e-mail exchange about the Iranian election with his friend and fellow programmer Ian Pearce.
It was the summer of 2009, and the college students were at their respective homes (Oregon for Gaffney; Pennsylvania for Pearce). The two came up with an idea: create a data set that included all (or most) of the tweets about the Iranian election.
“If we collect all this data, we can seriously have a chance at making some real progress on whether or not citizen journalism actually exists or is real or not,” Gaffney said in an e-mail to Pearce.
Gaffney and Pearce’s Iranian election data set became the second-largest outside of Twitter the two had encountered, and a talking point in Internet circles. It was also Gaffney’s ticket into the Northeastern program.
The network science PhD program is an interdisciplinary take on the science of viral behavior, be it that of a disease or of Internet outrage. It is small and new: Gaffney is one of just four students, including a biologist, an applied mathematician, and a physicist.
“Devin is very technically capable and very creative in terms of his thinking around networks,” said David Lazer, who is one of the leaders of the program, in a phone interview.
Gaffney’s current project: Internet outrage. From #GamerGate to #CancelColbert, Gaffney is exploring the difference between viral cascades — which rise and fall quickly and rarely last — and the other stuff, the incandescent conversations, the stuff that leaves the Internet and forces users to act.
“I want to know what distinguishes viral cascades like The Dress from situations like Gamergate,” Gaffney said. “How much of it looks the same, how much of it looks different, what distinguishes the cases of Internet outrage where it explodes and just stays, like this tumor on the side of social networks, this malignant, awful thing that we all live with and revisit as it becomes darker and scarier and more upsetting.”
But not all Internet outrage is violent. Gaffney has also been working with Northeastern communications professor Brooke Foucault Welles to investigate online activist movements like #BlackLivesMatter and #YesAllWomen. The professor plans to co-write a book on the subject.
“It’s a book about contemporary activism that focuses on ‘counterpublics,’ an academic term to describe marginalized groups who often don’t have a voice in places like mainstream media, like people of color and women, to name a few,” Foucault Welles said in a phone interview.
Gaffney’s quiet corner is getting louder and more crowded. For now, that’s a benefit: More voices means more stories to hear, more information to spread. But he’s not interested in placing bets on viral potential — he’s interested in human beings.
“I don’t care about predicting,” he said. “I want to know how humans work. An important part of achieving social justice and equality and reducing violence . . . is a nuanced understanding of how we get to the point where we become part of something and what that means.”
That’s a harder question to answer than the color of a dress.
Gaffney’s #TheDress Twitter graphs:
Gaffney tracked the use of the hashtag at various times on Feb. 26.
Related coverage:firstname.lastname@example.org. Follow her on Twitter @bjackgli.