How Twitter Helps in a Disaster

The evidence continues to pile up that Twitter is a news service, not a social network. Of course, Twitter only works as a news service because its news is routed according to social connections–and that’s the secret to the service’s ability to endlessly issue, digest and re-synthesize news into actionable 140-character memoranda. This is true even–or perhaps especially–in an emergency.

A recent study on the use of Twitter during natural disasters neatly illustrates the paradox of conversational micro-blogging. The majority of information that was retweeted during the 2009 record flooding of the Red River in North Dakota was news – as in, information that did not already exist on Twitter or even the web. But a great deal of the utility of the service is demonstrated not by this new information, which constituted less than 10% of tweets culled from a representative sample of Twitter accounts during the disaster, but the derivative and synthetic tweets that followed in the wake of these original tweets.

Many of these tweets were issued by local and national news media, but a surprising number originated with disaster-specific Twitter accounts that arose for the purpose of updating others with useful information.

Local and national news organizations, especially, engaged in synthetic and derivative tweeting during the disaster, while 80% of the original, “citizen-reported” tweets came from locals who were living the disaster.

This complicated interplay between original reporting by locals and synthesis by both traditional news media outlets and flood-specific twitterers led the researchers to condlude:

“Our data indicate that Twitter activity cannot be defined completely in terms of generative and synthetic information production. Twitter is not simply a platform for broadcasting information, but one of informational interaction. […] navigation of this unwieldy space is difficult. Many of these conventions have evolved to aid this navigation, directing other users to valuable information, placing virtual signposts within a complex information space.”

Retweeting, the researchers argue, is what Twitter has instead of a formal recommendation system. It’s not simply that information is syndicated: a retweet is a signal that a particular piece of information is important. Retweeting, synthesis and the occasional addition of local knowledge of even out-the-kitchen-window type reporting by locals generated even more noise in the stream of millions of tweets on the disaster that, in turn, the twittersphere condensed into actionable information:

“Through these activities, Twitterers both self-organize and create the need for more self-organization, as they generate even more noise that gives rise to the need for more directing and focusing behaviors. Derivative information production is therefore a user-driven cycle of shaping and re-shaping a shared interaction and information space.”

A natural question in disaster tweeting is whether or not the information pouring out of Twitter on a particular event can be processed quickly enough - either by Twitter users themselves or some outside body - to allow decision-makers to act. A second paper by the same researchers takes a first step toward answering that question. Sarah Vieweg and colleagues at the University of Colorado, Boulder had to manually code thousands of tweets to come up with a classification scheme that could be applied to future disasters, but without automation their procedure is too slow to be used in its current state.

Both papers illustrate that while the synthesis and analysis tweets carried out by both journalists and citizens provides the most value for users, it is original reporting that ultimately makes all the secondary analysis possible.

“When individuals were the original source for the retweets, two-thirds of the time they were local or peripheral. The interpretation of this is that locals and peripheral Twitterers (individuals, media or flood-specific services) are the locus of retweeted information.”