Automation, Mobilization, and Coordination in Political Astroturfing on Twitter

• In Event: China's Propaganda Apparatus

Abstract

The past years have witnessed political chaos caused by large-scale information operations. Foreign governments are often claimed to be actively astroturfing on social media platforms including Twitter and Facebook in order to expand their influence, shape national image or manipulate public opinion in the United States. In the last two years, Twitter has announced that it has identified state-backed information operations and has therefore suspended many troll accounts. The accounts, as well as their tweets, provide a great opportunity for researchers to look into how political astroturfing is organized and operated.

In this study, the astroturfing accounts and tweets from China, Iran, Russia, Saudi Arabia, and Turkey were analyzed to explore the pattern of their organization and operation. While previous research using the same sets of data released by Twitter mainly focus on content analysis to see the topic or sentiment, or their activity patterns, or interaction with targeted audiences or within themselves, we would like to ask a meaningful but often neglected question: why did their posts come from? By looking into the source of their tweets, we hope to provide some evidence regarding automation, mobilization, and coordination of political astroturfing.

More specifically, we focused on the URLs used in astroturfing tweets. After analyzing over 10 million tweets, we found some significant evidence where propagandists used automation services to generate and distribute content on Twitter. We also observed that a significant portion of astroturfing tweets was shared synchronically from state-owned media outlets, indicating potential mobilization organized by state actors. Moreover, we also analyzed the co-sharing network from the same source to see the dynamics within the astroturfing community and explored cross-platform sharing from other major platforms including Facebook and Youtube. Both co-sharing networks and cross-platform sharing networks validated the existence of coordination of astroturfing.

Operationally, we traced back to the domain level of URLs embedded in astroturfing tweets and analyzed the categories of each domain. Apart from domains for automation and social platforms, we also focused on domains for news websites as well as domains related to the foreign country. In order to have a better understanding of the source information, we also collected the description and title for all websites related to news and the foreign country itself. Then topic modeling was conducted to understand the main theme of the website. Special attention was paid to exploring the extent to which these foreign states propagandize themselves and promote their national images.

Overall, the study provides a comprehensive overview of the source of political astroturfing operations. Evidence was found to unveil the patterns of automation, mobilization, and coordination in astroturfing, which helps us to have a deeper understanding of how the public sphere is interfered by political astroturfing.

Author

• Mengqing Zhang