Adapting Social Spam Infrastructure for Political Censorship

Chris Grier
Vern Paxson
Proceedings of the 5th USENIX Workshop on Large-Scale Exploits and Emergent Threats(2012)


As social networks emerge as an important tool for political engagement and dissent, services including Twitter and Facebook have become regular targets of censorship. In the past, nation states have exerted their control over Internet access to outright block connections to social media during times of political upheaval. Parties without such capabilities may however still desire to control political expression. A striking example of such manipulation recently occurred on Twitter when an unknown attacker leveraged 25,860 fraudulent accounts to send 440,793 tweets in an attempt to disrupt political conversations following the announcement of Russia’s parliamentary election results. In this paper, we undertake an in-depth analysis of the infrastructure and accounts that facilitated the attack. We find that miscreants leveraged the spam-as-a-service market to acquire thousands of fraudulent accounts which they used in conjunction with compromised hosts located around the globe to flood out political messages. Our findings demonstrate how malicious parties can adapt the services and techniques traditionally used by spammers to other forms of attack, including censorship. Despite the complexity of the attack, we show how Twitter’s relevance-based search helped mitigate the attack’s impact on users searching for information regarding the Russian election.

Research Areas