This is just a test: Emergency responders tap the Twitterverse

Thinkstock

Simulation examines how social media can aid response to a terrorist attack.

Fairfax County Fire and Rescue Battalion Chief Anthony Barrero ran his hands over a large Google Earth image of the George Mason University campus Tuesday afternoon and assessed the situation. The map was covered with dots representing students, some of them sick and injured, sheltering after what appeared to be a dirty bomb attack on campus. Improvised explosive devices placed along nearby roads had delayed an immediate evacuation and there were few rescue workers on campus.

Most reports were coming in from the students themselves, using a Twitter look-alike called Chirp. Some students chirped they were injured or feeling sick. Others were reporting strange backpacks around campus. In the makeshift emergency operations center spread out behind Barrero, state, local and federal emergency management officials were monitoring students' chirps, chirping back to gather details about the suspect backpacks and asking students to chirp about their injuries and locations with the hash tag #gmuhelpme.

This simulated calamity was organized by MITRE, a nonprofit that manages federally funded research projects. Its conclusions could lead federal, state and local officials to look more closely at how they can use social media to improve disaster response.

“We’re examining the hypothesis that public participation in crisis response can improve the outcome of the crisis,” said James Dear, project leader. “We’re using the public not just as sensors but as participants in the decision-making process.”

Dear and his team at MITRE spent about four months putting together the GMU simulation and will spend another month reviewing data gathered during the weeklong exercise before issuing a final report.  

The operations center at MITRE’s headquarters in suburban Virginia was filled with more than a dozen representatives from Fairfax County, the Commonwealth of Virginia, the National Guard, U.S. Northern Command, the FBI and the Federal Emergency Management Agency on Tuesday morning.

The officials essentially played themselves, acting as they would in a genuine emergency. The only difference was the chirps -- coming in from about 200 student volunteers on the George Mason campus who received $95 toward on-campus purchases in exchange for their participation -- and a system called the Citizens’ Emergency Response Portal System, or CERPS, that emergency managers used to mine student chirps, geographically locate reports and repost verified information.

The nonprofit kept CERPS and the student chirps behind a firewall to avoid any War of the Worlds-like confusion by outsiders who might have concluded GMU was actually under attack, Dear said.

While the emergency operators did their work, his team collected data. Observers wandered through the simulation taking notes, but software collected a much larger share of the data through programs that tracked emergency operators’ eye movements and captured every page and image that scrolled across their computer screens.

MITRE has managed about 40 simulations for government agencies since 2002, mostly for the Defense Department. This was the company’s first unclassified simulation, according to Dear.

The emergency operators reran the same basic scenario throughout the week, introducing new plot twists to see how chirp and CERP work in a variety of circumstances. On Tuesday, for instance, one chirper who may or may not have been allied with the bombers spread misinformation about the explosions and threatened more were to come. The simulation’s on-the-ground manager Tobin Bergen-Hill, who Dear says plays something like a director’s role to his producer, could change the scenario midstream to maximize data collection and learning.

The emergency operators didn’t know any of the plot twists in advance.

Early Monday, the first simulation day, emergency managers largely ignored student chirps and went about business as usual, Bergen-Hill said. By the end of the day, they were actively engaging with student chirpers, organizing a chirping strategy, and relying on student chirpers to be their “eyes and ears,” he said.

The student participants weren’t physically gathered on campus but were looking at a computer interface that showed a Google Earth image of their location in the simulation. Those images included graphics of what was happening at that point in the exercise, such as explosions or other students running away, and pop-up icons describing what the students were smelling and hearing.

MITRE had trained the students to use the technology and instructed them to act their parts fully, but once the simulation began, the exercise coordinators allowed the students to act on their own without further guidance, Dear said.

Social media in emergency management isn’t a new concept. Emergency managers have mined social media for information and tried to communicate with victims through Facebook, Twitter and other social media in the past.

CERPS is actually based on Ushahidi, a crowdsourced mapping application that was used prominently after the 2010 earthquake in Haiti by aggregating Haitians’ text message reports and sending them back to relief workers. That effort was led by volunteers who were proving their programs’ value, not by the emergency responders themselves. If the GMU simulation proves successful, Dear hopes government or industry will develop technology that incorporates social media more effectively into the fabric of emergency response.

“We’re actually trying to engage the American public in crisis response,” he said. “It’s like e-democracy at the tactical level.”