THIS PROTEST WILL BE TWEETED

Abstract

This article examines the use of Twitter at protests surrounding the G20 meetings held in Pittsburgh, PA in September 2009. Based on work on information communication technologies and protest, and on more recent work on Twitter usage at protests, we develop several hypotheses about the content of tweets during protests. Most significantly, we argue that Twitter is a widely available mobile social networking tool that can be used to reduce information asymmetries between protesters and police. Examining the content of 30,296 tweets over a nine-day period, we find that protesters frequently used Twitter to share information, including information about protest locations, as well as the location and actions of police, which is information that was formerly monopolized by the police. Twitter use may be creating a new dynamic in protester and police interaction toward information symmetries. We conclude by identifying implications for policing practices and for protesters.

Keywords:

• Twitter
• protest
• protest policing
• repression
• information asymmetry

Acknowledgements

We would like to thank Alex Halavais, who created the archives of tweets studied here, and Maria Garrido, Dana Fisher, Rachel Einwohner, and Kelly Garrett for their comments and feedback. We would also like to thank the National Science Foundation for generous support of the content coding of the tweets and subsequent analysis through an NSF CAREER Award (SES-0547990). Finally, we thank the many research assistants who worked on this project (see http://jearl.faculty.arizona.edu/internet_and_activism_lab for a full listing) and Heidi Reynolds-Stenson for her research assistance.

Notes

We focus on a single, prominent hash tag related to the G20 event to mimic Twitter user practices and to contribute to existing Twitter research that analyzes one prominent hash tag per protest event (see Heverin & Zach 2010 and Gaffney 2010 for a similar approach).

We do not report an inter-coder reliability score because our coding procedure ensured that there was complete agreement on the final coding (i.e. everything was double-coded and disagreements were discussed until consensus was reached).

In Figures 2 –4 in the findings section, we dichotomize these variables, while in Figure 5 we report on only information, opinions, and questions that included police.

Note that the total number of tweets on the 25th was a little over 9,000, and approximately 4,000 (with many of those coming before 3 am) on the 26th. On the 27th and 28th, there were only about 1,000 tweets each day. Because all figures (except Figure 2) represent proportions, a much smaller number of actual tweets about police were exchanged following the protests even though the proportion of police-themed tweets was greater.