The inner (net-)workings of authoritarian regimes...

17th Central Committee | Social Networks | Chinese Elites | Hu Jintao

My research is situated in the growing subfield of studies on authoritarian regimes. I try to redirect the field's attention from irregular leadership change and formal institutions - like parties, elections, and parliaments - to the role of informal institutions in facilitating stable turnover among the leader's entourage. I propose that social network analysis offers a principled, but nevertheless nuanced, method to conceptualize and measure some informal institutions.

 

I elaborated on that idea in my dissertation, part of which was published in the Journal of East Asian Studies, and have further explored the more complex effect of having friends in high places in a study on networks formed in the Chinese Communist Party School published in Political Studies. In ongoing research projects (see below), I use expert survey data instead of publicly available CVs as a source for the influence and networks of Chinese political elites.

 

... and how do we learn about them?

 

My published research is based mainly on publicly available information, which has become harder to obtain and less reliable in China and other authoritarian regimes such as Russia in the last few years. In the past, assessments of expert working in journalism, think tanks and academia - often called "China Watchers" or "Kremlinologists" in the case of the Soviet Union - had often filled such gaps. My current research projects thus deal with two alternative sources of information on elite politics:  the first examines how one could automate the extraction of network information from unstructured text, such as newspaper articles and reports based on expert research and knowledge. In it, I use natural language processing and machine learning on a manually labeled training dataset of almost 5000 sentences from American elite politics, mostly from the Senate Intelligence Committee Report on Russian active measures during the 2016 US Elections and from newspaper articles on elite politics in the Trump White House. The other related research project surveys China experts to ask for the network connections among and the level of influence of Chinese political elites.

 

In both cases, I am also interested in the influence of the network among the different sources of information: How does the social network among China experts influence their assessment of Chinese politics and elites? Do different newspapers report on different parts or portray a different image of political elite networks because of their different sources?

... and the ties to the outside world...

 

In other working papers I examine these sources of power and instability outside the elite. In Degrees of Separation, I explore how the elite network is connected to the wider citizenry in Kazakhstan, by asking respondents to forward a message to government representatives through a chain of acquaintances. Individuals with more relatives working in the government express a higher level of vague satisfaction with the state and are more likely to vote – so are respondents closer to the regime along the transmission chains.

 

Networks among ordinary citizens are also the focus of How to Spot a Successful Revolution in Advance.

It develops an agent-based model of how potentially dangerous protest activity spreads along social ties within a country. Contrary to common expectation, I find that increased segregation and fractionalization can make the emergence of large-scale protests more likely.

 

Finally, in (Why) Do Revolutions Spread? I create a data base of large-scale popular protests and use a spatial model to examine why and how revolutions spread. In particular citizens in closed regimes appear to take protest in countries with similar regime type as a reason to update their beliefs about the true popularity of their ruler.

Revolution | Simulation | Agent-based Modeling | Protest

... and on the internet

In recent years, I have become interested in hidden disinformation campaigns  (so-called political AstroTurfing) on social media. I argue that we can discover such campaigns by searching for networks of message coordination (i.e. accounts that tweet or retweet the same or similar content within a short time window), which are an almost inherent feature of such campaigns, because principal-agent problems make it difficult or very expensive to hide these tell-tale patterns. I have documented those patterns in the case of one of the most earliest known AstroTurfing campaign in South Korea 2012 (ICWSM and Political Communication) and have shown that they are present also in the most recent campaigns revealed by Twitter .