Internet Shutdowns in Authoritarian Regimes

In my dissertation, “Internet Shutdowns in Authoritarian Regimes”, I examine how governments use their oppressive techniques to control the internet. Despite growing research interest in media control, we know little about a crude means of internet manipulation to which dictators often resort – wholesale internet shutdown. I develop a theory explaining why weak states’ dictators containing anti-regime protests often lack means beyond complete internet blackouts to prevent coordination by the opposition.

On June 13, 2019, Ethiopian government blocked the internet to allegedly prevent cheating during school exams.

The central problem with assessing theories about internet shutdowns is that they are quite difficult to measure and detect. Tackling this challenge, I create a new cross-national data set on internet shutdowns since 2009. I compile a data set on internet activity for 210 countries that contains history of national-level hourly metrics on Google activity and daily users of Tor – a VPN tool allowing people to circumvent internet blockages. I apply robust ARIMA models to these time series data in order to reliably detect anomalies in the data. In addition, I develop a method of indicating whether a blackout is a government-made or a “naturally occurring” event. I do so by comparing people’s activity on Google and Tor. This is a major improvement over existing internet shutdowns datasets as it does not rely on self-reported cases of blackouts and hence it results into a more comprehensive view. I integrate these newly developed data on national internet shutdowns as well as the data on the number of internet Service Providers in the country and country’s connectivity with other daily event data sets.

On June 22, 2019, government shut down the internet following an alleged coup attempt. Note the increasing number of Tor users from Ethiopia after both incidents.

The new data enable me to test my theoretical predictions. Periods of mass unrest and protests are often particularly perilous to dictator’s survival. In some well-studied authoritarian regimes, such as China and Russia, the regime has developed sophisticated techniques to monitor dissidents on social media as well as to flood the internet with pro-regime or other misleading information. However, these authoritarian regimes are exceptional for their high state capacity, and many rulers lack the technological infrastructure to carefully manage what their citizens see on the internet. Instead, faced with threatening situations, lower-capacity autocrats may simply choose to cut off the internet to prevent citizens from communicating. Thus, I expect internet shutdowns to occur primarily in low-capacity authoritarian regimes when facing mass protests. I expect to demonstrate that protests in countries with lower GDP and fewer ISPs are followed by internet blackouts more often.