This article was adapted from a presentation by Jeannette Sutton at the 2009 World Conference of Disaster Management.
We know a lot about how people react in disasters, and emergency managers can draw on that knowledge and background. For example, we know how people communicate in disasters, and we know how to build effective warning systems so failures are designed out. We know how to develop messages that most directly impact people so they will take protective action. And we know how to craft preparedness campaigns to help people prepare for disasters. But can we take what we know and apply it to the use of social media in disasters? To answer this, I draw from research on three cases: the Virginia Tech shooting, Southern California wildfires of 2007 and Democratic National Convention of 2008.
In the immediate aftermath of Virginia Tech School shooting that occurred on April 16, 2007, my colleague Leysia Palen, who is an assistant professor of computer science at the University of Colorado, and I sent two graduate students to Blacksburg, Va., to investigate how people were sharing information, what tools they were using, and from what sources they were receiving their information. When the graduate students returned from the field after talking with students, faculty and community members, we started gathering data online to learn how that community was using social media.
We focused on Facebook. We archived a number of different groups on Facebook that emerged around issues of information sharing and condolences related to the school shooting, and ones that had the largest numbers of participants. And then we looked at the conversations between people.
What was most interesting about the ways that students and others were using Facebook during this event was how coordinated the information sharing was and how people were so instrumental in bringing together knowledge from different parts of the United States to answer a particular question: who were the deceased.
Late in the morning of the Virginia Tech shooting, the university announced that there were a number of deaths. It initially didn't identify how many there were. But students immediately began identifying people that they knew had been killed, and posted that information to various Facebook groups. By the late afternoon, the university announced there were 32 deaths. At that point, we observed students and others on Facebook ramp up their investigation and sharing of information to the point that by the next day, within 24 hours, they had identified all 32 victims before the university officially released the names of those who had been killed.
This is an amazing feat. They used their informal networks to identify the names of all of those students who had been killed. This does raise the question about whether or not this information should be shared on an open platform like Facebook. But the power of collective intelligence that we observed through Facebook showed that there is a great deal of power in these distributed networks and that wisdom can rise up through the crowd.
What's also really important about this is that in the lists that we looked at, we never saw a wrong name listed. And there was turn taking: Every time someone listed a death, they had to verify how they knew the information. This certainly lends some credibility to the ways that people are sharing information.
The Southern California wildfires of 2007 led to the evacuation of much of San Diego. During this disaster, I was working with Palen and Irina Shklovski, a postdoctoral researcher at the University of California at Irvine. During the disaster, Shklovski went to evacuation sites to ask people how they were sharing information and what information was most valuable to them.
After the evacuation orders were lifted, we quickly developed an online survey to draw a broader sample of respondents who were affected by the wildfires. To distribute the survey we used social networks and community forums that we knew technology users were already using. Although the survey we conducted wasn't representative of the entire Southern California population, it was instructive about how tech-savvy users were utilizing technology during the wildfire.
Our survey respondents said they first turned to major media. That wasn't a surprise. I don't think that is changing yet. We haven't reached a tipping point -- we don't even know what the tipping point is -- in terms of people turning first to social media.
But in disasters information seeking is intensified. People will look for information, and if it's not coming fast enough or isn't perceived as being accurate they look somewhere else. We discovered that many people who went online for information found that there was misinformation coming from the public authorities and from major media. In contrast, many of these information seekers had access to information that was accurate at the local level, so they provided that information as a corrective. That was a very interesting thing to find out: that there was a perception that the locals were more accurate than the authorities.
In another example, the Democratic National Convention of 2008 in Denver attracted international attention, so there was significant concern about a public disruption. It was actually a National Special Security Event, which meant the Secret Service was in charge. But the city and county of Denver had responsibility largely over the emergency operations and the Joint Information Center, in coordination with federal authorities.
During the convention I had a number of researchers embedded in the Joint Information Center. Our research purpose was to observe the ways that public officials, namely public information officers, were looking at and using information that was flowing among the public who were using social media. We looked at how members of the public were using these channels, and also how public officials were making use of them.
There were two key pieces of information: The first was that the Democratic National Convention Committee validated the importance of social media for this convention. The committee credentialed 180 blogs at the beginning of the event. That means every blogger who had a credential had the same legitimacy as major media in the eyes of the convention. The bloggers had access to all the different press events, and they were doing interviews. The second point was Denver's IT department lifted all its restrictions on social media. Normally government agencies are restricted from accessing social media sites.
During the convention, government workers could log on to Twitter, Facebook and MySpace, and use streaming video and social networking applications. But what we found largely was that there was no plan in place. Many people didn't know how to use social media. They didn't have awareness of the value of the information that was flowing, nor resources that could be put toward monitoring these sources of information. But there was monitoring of blogs, and that has the potential to lead to some very important information. We didn't actually see many public disruptions during the convention, so that may have been a reason why there was less attention to breaking news through social media channels -- and there wasn't a lot of information posted on the credentialed blogs about what was going on, outside of the political commentary.
Many emergency managers I've talked with have expressed concerns about social media as a channel, and that the public at-large is going to be sharing misinformation. They also say there isn't a sense of organization within the online communications.
But this isn't an entirely accurate perception of how social media is being used online. It's very organized. It just isn't organized through a central point. And, it's self-correcting. Those who participate on sites like Wikipedia or are invested in a particular conversation have some sort of stake in making sure the information is correct. So they put out information to correct misinformation. Largely it's accurate, as were the people who were posting information during the Virginia Tech School shooting. They were verifying their sources and saying how they knew it was correct.
Emergency managers have also expressed concerns that rumors and misinformation are flowing through social media, this is especially so because people can post anonymously. This is a valid concern, particularly because these channels enable information sharing to flow quickly in what's considered a "viral" manner. But as we saw in our wildfire research, information that was misinformed or inaccurate didn't come only from the public. It also came from the major media and public officials. And social media provides one point across a vast number of points of information.
From a sociological perspective, rumors are not pathological. They aren't intentionally malicious forms of information sharing. Instead, informal networks of people are improvising in situations of intensified information seeking, which is what occurs in disasters. If officials don't release information to the public quickly enough, the public might conjecture as it tries to figure out what's happening. This is an important reason for emergency managers to be integrated into these channels of communication, so they are pushing out accurate information as quickly as possible, preventing the start and continuation of rumors.
Jeannette Sutton is a sociologist at the Natural Hazards Center at the University of Colorado at Boulder.