In How Sudden Censorship Can Increase Access to Information, Hobbs and Roberts (2018) ask under what circumstances increases censorship indeed expand a subset of a country population's access to such information. Their question is important because it challenges the conventional wisdom that increases censorship will suppress and reduce the probability that citizens will access to such information. To substantiate their argument, Hobbs and Roberts use a combination of data sources. The authors collected geolocated Instagram posts from mainland China before and after a sudden Instagram block by the Chinese government on September 29, 2014, to assess the discontinuity after the censorship. As expected, they found that a little less than half of the sampled Instagram users evaded the censorship and continued using the application.
Second, the authors also gathered the download statistics of censorship-evasion applications, namely numerous VPN service providers, from App Annie, an application analytics service, to measure the number of people who sought to continue Instagram usage. They found that VPN service applications suddenly became popular, four of them making it to the top-ten productivity most downloaded applications on September 29, the day of the block. None of them were present in the top ten prior to the censorship.
Third, they collected new users statistics from websites such as the New York Times Chinese edition and Apple Daily, a Hong Kong newspaper, to show that the popularity of these pages also surged after the censorship occurred. Following the Instagram block, they detected a sharp increase in new users who signed up for the aforementioned websites. Hobbs and Roberts interpreted that users who were affected by an Instagram block, once gained access to it via VPN services, also visited other websites that were blocked in the mainland.
Finally, Hobbs and Roberts attempted to show that the effect of sudden censorship like the Instagram block may have a longer-term effect on the population. They conducted a simple test by looking at a subset of users whose signup dates on Twitter were between August and September 2014 who mentioned Liu Xiaobo in their tweets after his death in 2017. They found a sharp increase in the number of users who signed up for Twitter after the Instagram block in September 2014.
They concluded their study that the unintended effect of censorship which increases the population's access to the censored information may occur in three circumstances:
when the censorship is sudden and disruptive of citizens' habits
when the censored information cannot be easily substituted
when the cost of evading the censorship is not high
This paper is illuminating, but I think the project can be extended in two ways. First, expanding the timeframe and types of social media to evaluate the long term effect of sudden censorship should validate or invalidate a simple test included in the paper by Hobbs and Roberts. Since Instagram was blocked in 2014, there may not be long enough time to be certain about a long term effect. Twitter which was blocked in 2009 may present a better opportunity to assess the long term effect (2009-2019). Additionally, the limitation of the empirical evidence of users who signed up for Twitter in 2014 and tweeted about Liu Xiaobo is that there could be many users who went undetected if they did not tweet. Some users may be politically active by retweeting or being online to check the news but may not actively tweet. I propose to collect data on users of Instagram and Twitter who signed up during a period immediately after the block occurred until the present. We can do a couple of things with this data. First, I can count how many users who signed up for Instagram after September 2014 and are still active. Being active means the users are still posting photos, liking, or commenting on others' photos. Second, once we have the number of active users over time, we can calculate the drop out rate, meaning when the users became inactive and stopped using the application. This could be telling because Hobbs and Roberts mentioned in their paper that the Chinese government changed a censorship strategy from a blunt blocking of social media to make it more difficult for Internet users to use VPN services. If we see a sharp drop out rate coinciding with the VPNs crackdown, that would not be surprising. However, if we observe the opposite, this could mean that the effect of blunt censorship which incentivizes people to use censorship-evasion technology may last.
Secondly, the project can be extended by analyzing users' networks in order to show that users remain politically active and opt for VPN services to access information that is censored by the government. For example, on Twitter, we may be able to identify thought leaders or dissidents who criticize the Chinese government on these platforms. We then find whom these new users have in common in terms of "following" after the Twitter block. I think that the paper demonstrates strong evidence that people did evade censorship, but the evidence is weak when they tried to link it to politics since I suspect that Instagram users may not necessarily be the politically active type.
Reference
Hobbes, William R., and Margaret E. Roberts. 2018. "How Sudden Censorship Can Increase Access to Information." American Political Science Review 112 (3). Cambridge University Press: 621–36. doi:10.1017/S0003055418000084.
In How Sudden Censorship Can Increase Access to Information, Hobbs and Roberts (2018) ask under what circumstances increases censorship indeed expand a subset of a country population's access to such information. Their question is important because it challenges the conventional wisdom that increases censorship will suppress and reduce the probability that citizens will access to such information. To substantiate their argument, Hobbs and Roberts use a combination of data sources. The authors collected geolocated Instagram posts from mainland China before and after a sudden Instagram block by the Chinese government on September 29, 2014, to assess the discontinuity after the censorship. As expected, they found that a little less than half of the sampled Instagram users evaded the censorship and continued using the application.
Second, the authors also gathered the download statistics of censorship-evasion applications, namely numerous VPN service providers, from App Annie, an application analytics service, to measure the number of people who sought to continue Instagram usage. They found that VPN service applications suddenly became popular, four of them making it to the top-ten productivity most downloaded applications on September 29, the day of the block. None of them were present in the top ten prior to the censorship.
Third, they collected new users statistics from websites such as the New York Times Chinese edition and Apple Daily, a Hong Kong newspaper, to show that the popularity of these pages also surged after the censorship occurred. Following the Instagram block, they detected a sharp increase in new users who signed up for the aforementioned websites. Hobbs and Roberts interpreted that users who were affected by an Instagram block, once gained access to it via VPN services, also visited other websites that were blocked in the mainland.
Finally, Hobbs and Roberts attempted to show that the effect of sudden censorship like the Instagram block may have a longer-term effect on the population. They conducted a simple test by looking at a subset of users whose signup dates on Twitter were between August and September 2014 who mentioned Liu Xiaobo in their tweets after his death in 2017. They found a sharp increase in the number of users who signed up for Twitter after the Instagram block in September 2014.
They concluded their study that the unintended effect of censorship which increases the population's access to the censored information may occur in three circumstances:
This paper is illuminating, but I think the project can be extended in two ways. First, expanding the timeframe and types of social media to evaluate the long term effect of sudden censorship should validate or invalidate a simple test included in the paper by Hobbs and Roberts. Since Instagram was blocked in 2014, there may not be long enough time to be certain about a long term effect. Twitter which was blocked in 2009 may present a better opportunity to assess the long term effect (2009-2019). Additionally, the limitation of the empirical evidence of users who signed up for Twitter in 2014 and tweeted about Liu Xiaobo is that there could be many users who went undetected if they did not tweet. Some users may be politically active by retweeting or being online to check the news but may not actively tweet. I propose to collect data on users of Instagram and Twitter who signed up during a period immediately after the block occurred until the present. We can do a couple of things with this data. First, I can count how many users who signed up for Instagram after September 2014 and are still active. Being active means the users are still posting photos, liking, or commenting on others' photos. Second, once we have the number of active users over time, we can calculate the drop out rate, meaning when the users became inactive and stopped using the application. This could be telling because Hobbs and Roberts mentioned in their paper that the Chinese government changed a censorship strategy from a blunt blocking of social media to make it more difficult for Internet users to use VPN services. If we see a sharp drop out rate coinciding with the VPNs crackdown, that would not be surprising. However, if we observe the opposite, this could mean that the effect of blunt censorship which incentivizes people to use censorship-evasion technology may last.
Secondly, the project can be extended by analyzing users' networks in order to show that users remain politically active and opt for VPN services to access information that is censored by the government. For example, on Twitter, we may be able to identify thought leaders or dissidents who criticize the Chinese government on these platforms. We then find whom these new users have in common in terms of "following" after the Twitter block. I think that the paper demonstrates strong evidence that people did evade censorship, but the evidence is weak when they tried to link it to politics since I suspect that Instagram users may not necessarily be the politically active type.
Reference
Hobbes, William R., and Margaret E. Roberts. 2018. "How Sudden Censorship Can Increase Access to Information." American Political Science Review 112 (3). Cambridge University Press: 621–36. doi:10.1017/S0003055418000084.