Socialbots in politischen Online-Konversationen: Eine (uberschatzte) Gefahr fur demokratische Willensbildung im Internet? [Socialbots in online political conversations: An (overestimated) danger for democratic opinion-formation on the Internet?]

The manipulative use of so-called social bots on the Internet is increasingly regarded as a problem and danger to the public and to democracy. At the same time, however, research regarding the political use of social bots is nascent . Up to now, only few empirical studies on the topic exist, which primarily make use of automated big data analyses to identify social bots and their political agenda . In this article we argue that existing research on the role of social bots in the 2016 US presidential election campaign has significant shortcomings, which may limit the validity of results on the influence of social bots in political conversations on Twitter . While existing research suggests that social bots (identified using automated methods) were able to influence online conversations on Twitter and demonstrated bias towards the Republican presidential candidate Donald Trump, we contend that these findings were in large part the result of methodological choices that are inherent to conducting empirical research at scale using big data . In this paper, using Twitter data collected during the 2016 first US presidential debate, we combine network analysis methods with qualitative sociological approaches and find in fact that social bots tweeted rather negatively about Trump and did not exert significant influence on online conversations.