Take a look at the pictures from Tahrir Square -- that improvised amphitheater of 2011’s Egyptian Revolution. Take a look at its form, its sway of center tents, its hubbub of banners and clustered people. The crowds appear in one body with a hole cut in its middle and throngs that make tails through Cairo’s streets. What stories do the images call up? What relationships and connections does it depict? Tales of democracy? The fall of a regime? Nationalism? Unity? Separation? Or is it a combination of a sort?
Unearthing the story behind the human crowd, in Cairo and elsewhere, has always been an ambition for Sociologist Marc Smith, but now, he’s taken his fascination and made it digital, mapping Twitter conversations with diagrams that appear not too dissimilar from an aerial crowd shot.
“If I show you the pictures of Maiden Square in the Ukraine or of Tahrir Square in Egypt, the current picture is newsworthy whatever it is,” Smith said. “And if you accept that whenever several hundreds of thousands of people gather it’s newsworthy, then it’s worth noting that several hundreds of thousands of people are gathering this second [through Twitter], around almost every topic imaginable.”
Smith, the director at the Social Media Research Foundation (SMRF), is a co-author to a new and potentially ground breaking study administered by the Pew Research Center and SMRF that’s taken Twitter conversations and plotted them into a set of six structures — also called “conversational archetypes.” Each and every conversation ever had within Twitter, the study asserts, can likely fit into one of these six network structures, with the big news being that these maps can identify all key influencers and the innate design behind the dialogue.
Sweetening the pot, the six structures were discovered by a collaboratively created open source tool called NodeXL, an acronym standing for Network Overview Discovery and Exploration for Excel. The tool is free and can be used by anyone using Microsoft Excel to plot and chart networks from Twitter handles, hashtags and posted URL addresses.
“It makes it much easier to do network analysis," Smith said. "You no longer need to have a Ph.D. in computer science."
To find the six structures, Smith, the study’s team of authors and their research collaborators used the tool to analyze thousands of Twitter networks based around hashtags and Twitter accounts for four years. Currently there are about seven NodeXL servers running around the planet and mapping 500 to 700 maps per day. The tool, Smith says, has been downloaded more than 250,000 times.
“NodeXL is like a digital snapshot camera for cyberspace, it’s simple enough that we set it up and then pointed it at any topic that crossed our minds.”
A map rendering by NodeXL of Government Technology's Twitter network and the connections between users.
What emerged from the excavation of tweets manifested itself into the form of the six network structures each with its own personality and way of acting.
There was the Polarized Crowd, which held two dense groups of Twitter users with little connection to each other. Topics here being hot button political disputes or highly divisive issues. Polarized Crowds on Twitter don’t argue, the report said, but prefer ignoring one another while pointing to different Web sites and hashtags to support arguments.
In the Tight Crowds structure, discussions were marked by close knit users, few were isolated, and many delved into topics around conferences, professional issues, hobbies and learning communities.
Brand Clusters — a structure for products like soft drinks and smart phones and other breeds of goods and services — is a network populated with a host users that came en masse but generally never talk with one another. In fact, the larger the population of Tweeters in a brand cluster, the less likely they are to tweet each other. Another takeaway, Smith discovered, is that usually most are satisfied to simply mention the product without ever contributing a dialogue about it.
Community Clusters could be a cousin to Tight Crowds, only they typically spring from the news and popular topics that push the commentary away from the news outlets and into hubs, each with its own audience, influencers and sources of information. The study describes them as “bazaars with multiple centers of activity.” Here, some are connected through dialogue while others are not.
The Broadcast Network is mapped like a hub with sprawling spokes. This is your CNN, Fox News, New York Times and San Francisco Chronicle, all the news outlets and their own community of Twitter followers and associated readers, watchers and listeners. The posts are primarily reposts or “retweets” of the news. Most don’t talk with each other. However, sometimes a few small sub-groups do, and the report dubs these as “subject groupies.”
And finally, there’s the Support Network, a network housing the gnaws and bites of customer complaints of big business. Opposite to the Broadcast Network that has tweets pointing back at it, the Support Network has most its tweets exiting and managed by a large staff of customer service representatives. Most customers don’t interact with each other but have a back and forth with the provider.
Without a map, it’s hard to know where to head. The six network structures are the beginning point, a foundation, for what Smith sees as the next chapter in human interaction.
“More than 2 billion humans have moved their culture and more than half of their interactions from the face-to-face realm and into the computer mediated realm. And whether that’s good or bad it’s certainly a real thing,” Smith said, referencing statistics of world Internet users.
Within Twitter alone, the company reports 241 million monthly active users, 500 million Tweets sent per day, and 77 percent of its accounts are outside the U.S.
For Smith, the growing numbers translate into a need for continued research into the virtual landscape. He stresses that the use of the six network structures isn’t just an isolated pursuit for academia, but one that will likely be ubiquitous, with businesses, governments and organizations all using analytics and mapping to be more competitive.
“This is for anybody who’s attempting to communicate strategically,” Smith said.
Academic research has been the initial impetus for NodeXL, however, Smith said it’s highly likely — if not already underway — that the tool will be used by brands to understand their followers, competitors’ networks, key conversation influencers and they’re own Twitter use. The analysis, he said, can be directly linked with decision-making, social media strategies, and customer and citizen engagement initiatives.
“It’s not just the number of people (or followers), it’s the pattern they create collectively that really matters, the pattern in how they interact," Smith said. "I would say we’re very much at the beginning, but we’ve documented a path for other researchers to follow."
Jason Shueh is a former staff writer for Government Technology magazine.