A new study published in The Journal of Politics examines this question by exploring the effect of a 2017 Ukrainian ban on online activity among Russian social media users with close affiliations to Ukraine and Russia.

Authoritarian and nonauthoritarian states alike use censorship to police cyberspace, notes author Yevgeniy Golovchenko in “Fighting Propaganda with Censorship: A Study of the Ukrainian Ban on Russian Social Media.” In recent years, an increasing number of European states and tech firms have used it to combat digital disinformation and foreign interference.

Even before the Russian attacks in 2022, Ukraine offered some of the most extreme examples of censorship used to respond to information war and online propaganda from abroad. In 2017, the Ukrainian government issued an executive order that forced internet service providers to block access to major Russian websites, including VKontakte, the second most visited social media platform in Ukraine. The Kremlin’s control over Russian social media was one of the reasons why the Ukrainian government viewed the VKontakte ban as a national security measure against Russian propaganda and surveillance.

Censorship may successfully limit overall access to information, or it can backfire and draw attention to the “forbidden” or political outrage. “Even if a government succeeds in partially reducing the overall online activity on forbidden media, the ban may backfire if the supporters of the regime become less active on the censored platform than the opposition,” Golovchenko writes. “The government would risk making the opposition more prevalent on the platform than the supporters of the regime.”

Golovchenko uses publicly available data from VKontakte and a natural experiment research design to estimate the causal effect of the ban on online activity among different user groups. The findings indicate that a vast majority of Ukrainians on VKontakte could circumvent censorship by logging on through tools like VPN. However, the Ukrainian government still succeeded in reducing the overall online activity among Ukrainians on the Russian platform. “Government attempts at curbing Russian influence reduced the wall posting activity on VKontakte among users with pro-Russian attitudes at least as much as among pro-Ukrainian users,” notes Golovchenko, who found the same pattern when comparing citizens in Ukraine with few social ties to citizens within Russia versus those embedded in the Russian social network.

Even without legal repercussions for circumventing the ban, the increased access time and effort is enough to disrupt online activity among pro-Russian (and pro-Ukrainian) users, who would instead shift to cheaper and more accessible alternatives. “In other words, the accessibility of the media appears to play a much more important role in the decision to use censored social media than do politics or social ties with citizens in the hostile state,” Golovchenko writes.

“The results are favorable from the perspective of the censor, who wishes to combat foreign propaganda and disinformation by using one of the most drastic countermeasures available,” he writes. He went on to speculate, “If Russia were to use its newly upgraded censorship infrastructure to ban Facebook to prevent foreign influence, one would expect the ban to be successful from the point of view of the government if Russians were to respond in a similar manner as Ukrainians have.” Now that such a ban has come to pass, the effects of its harsher censorship remain to be seen.

Many states have become concerned with Russian cyberattacks and online propaganda. The Ukrainian government responded to the information threat in 2017 by blocking access to several Russian websites, including VKontakte, one of the most popular social media websites in Ukraine. By exploiting a natural experiment in Ukraine, I find that the sudden censorship policy reduced activity on VKontakte, despite the fact that a vast majority of the users were legally and technically able to bypass the ban. Users with strong political and social affiliations to Russia were at least as likely to be affected by the ban as those with weak affiliations. I argue that the ease of access to online media—not political attitudes toward the state—was the main mechanism behind the users’ response to the ban. These findings suggest that this pragmatic view on the effects of censorship holds, even in the highly politicized military conflict between Russia and Ukraine, which later evolved into a full-scale Russian invasion.

To what extent does censorship reduce activity on banned media among different societal groups in the context of international conflict? The answer to this question is essential if one is to understand the factors enabling or limiting the capacity of the state to control the flow of online information.

Authoritarian and nonauthoritarian states alike use censorship to police cyberspace (Edwards 2009; Howard, Agarwal, and Hussain 2011). In recent years, an increasing number of European states and tech firms have used it to combat digital disinformation and foreign interference. These information threats are often attributed to the Russian government and portrayed by the authorities as a threat to national security and election integrity (BBC 2017; Fiorentino 2018). In this context, Ukraine offers some of the most extreme examples of censorship use as a response to information war and online propaganda from abroad.

In 2017, the Ukrainian government issued an executive order, which forced internet service providers to block access to major Russian websites, including VKontakte, the second most visited social media platform in Ukraine (following YouTube) at the time (Roth 2017). The Kremlin’s control over Russian social media was one of the reasons why the Ukrainian government viewed the VKontakte ban as a national security measure against Russian propaganda and surveillance.

How effective are such interventions? Scholarly literature, ranging from media studies to political science, suggests that censorship may successfully limit the overall access to information (Kalathil and Boas 2010; Morozov 2012; Roberts 2018; Zhang et al. 2017). Censorship can also backfire via the so-called Streisand effect, by drawing attention to the forbidden information, causing political outrage against the censor, and increasing public awareness or interest in the forbidden content (Jansen and Martin 2015). In addition to this, bans may incentivize users to master circumvention through virtual private network (VPN) services, thereby gaining access to even more forbidden platforms (Hobbs and Roberts 2018). Conversely, the decision not to censor ideologically “undesired” content or the inability to fully implement a ban can in some cases help stabilize authoritarian regimes by providing citizens with highly demanded entertainment and news about societal problems in liberal democracies (Kern and Hainmueller 2009).

In sum, the existing studies document diverging instances when censorship backfires in some scenarios and successfully limits information in other cases. Similarly, the research on censorship presents examples when censorship is met by public outrage or discontent (Jansen and Martin 2015; Roberts 2018) by some groups and political support by others (Esberg 2020). Despite this, the existing literature offers little empirical research regarding the extent to which the same censorship policy may affect opposing societal groups differently (i.e., backfiring in one group while being effective in others). This question is crucial when evaluating the consequences of government censorship. Even if a government succeeds in partially reducing the overall online activity on forbidden media, the ban may backfire if the supporters of the regime become less active on the censored platform than the opposition. In other words, the government would risk making the opposition more prevalent on the platform than the supporters of the regime.

The Ukrainian context highlights the importance of this question. If the VKontakte ban reduces social activity among pro-Ukrainian users more than among pro-Russian users, the anti-Kremlin policy would make profiles with close ties to Russia more prevalent in relative terms. From a surveillance point of view, this would give Russian authorities less access to data on pro-Ukrainian users (e.g., activists, Ukrainian volunteers, or soldiers). From a propaganda point of view, however, the censorship would backfire by leaving pro-Kremlin propaganda even less contested on VKontakte. Drawing on the Ukraine case, this article seeks to answer the following question: What is the effect of the Ukrainian ban on online activity among VKontakte users with close affiliations to Ukraine and Russia?

I approach the question empirically by using publicly available data from VKontakte and a natural experiment research design to estimate the causal effect of the ban on online activity among different user groups. It is important to distinguish online activity—measured as number of public posts uploaded by the users on their own respective walls—from mere access to the VKontakte profiles (i.e., the ability to log on the platform).

The findings indicate that a vast majority of Ukrainians on VKontakte were able to circumvent censorship by logging back on to the forbidden website—likely through tools like VPN. Nevertheless, the Ukrainian government succeeded in reducing the overall online activity among Ukrainians on the Russian platform. Government attempts at curbing Russian influence have reduced the wall posting activity on VKontakte among users with pro-Russian attitudes at least as much as among pro-Ukrainian users. I find the same pattern when comparing citizens in Ukraine with few social ties to citizens within Russia versus those who are strongly embedded in the Russian social network. The increasing costs of using the banned platform (in terms of time and effort) explain the response to censorship much more than do the social ties or political attitudes toward the states involved in the armed conflict.

It is important to note that the ban has increased the cost of going online to a relatively small extent when considering the accessibility of free VPN tools in Ukraine and the fact that users are not legally persecuted for circumventing the ban. Even a small increase in the cost of accessing the Russian platform is enough to disrupt online activity among pro-Russian (and pro-Ukrainian) users, who would rather shift to cheaper and more accessible alternatives. In other words, the accessibility of the media appears to play a much more important role in the decision to use censored social media than do politics or social ties with citizens in the hostile state. The findings are in line with the “accessibility view” on censorship. The theory emphasizes the costs of accessing forbidden content—and not political attitudes—as the main mechanism behind the effect of censorship on online behavior. This view has been previously used to explain the effects censorship among impatient and relatively apolitical users in China through subtle “friction” in online access (e.g., by slowing down the connection; Hobbs and Roberts 2018; Roberts 2018). This study suggests that the relatively pragmatic view on censorship holds even in the highly politicized context of the hybrid Russia-Ukraine war (Lanoszka 2016; Reisinger and Gol’c 2014), which later turned into a Russian full-scale invasion of Ukraine.

This article contributes to the literature on censorship by addressing the debates on the driving mechanisms behind large-scale bans (Bunn 2015; Lorentzen 2014; Roberts 2018). Furthermore, this research adds to the burgeoning literature on misinformation (Nyhan and Reifler 2015; Tucker et al. 2018) and propaganda (Peisakhin and Rozenas 2018; Slutsky and Gavra 2017; Stukal et al. 2017), by empirically examining the effects of one of the most radical, large-scale policies to combat online manipulation.

Background

Russian authorities view the internet as a strategically important domain closely tied to national security (Vendil Pallin 2017). Following a series of legal restrictions in 2014, VKontakte and other Russian websites have been under increasing pressure to hand over private information about its users to FSB, the Russian Federal Security Service. This includes information about administrators behind Euromaidan-related pages (Sanovich 2017, 12). VKontakte’s founder, Pavel Durov, initially attempted to resist the pressure from authorities (Pan 2017). Eventually, he was pushed from the firm by major shareholders and left Russia (Economic Times2014). Today, VKontakte is predominantly compliant with Russian authorities, who have used the data in multiple criminal cases against individuals, some being prosecuted for antigovernment social media posts (Interfaks 2018; Robinson 2018).

Ukrainian authorities responded to the Russian government’s increasing influence over Russian social media in 2014 by advising citizens to delete their accounts on Russian-owned social media websites (Boichak and Jackson 2020, 270) before proceeding to a censorship policy in 2017. The VKontakte ban is part of a decree that imposed sanctions on 468 organizations and 1,228 individuals, including the Odnoklassniki social media platform and the Yandex search engine. VKontakte, however, was by far the most popular social media platform in Ukraine among all of the sites on the list.

The goal behind the decree is “to protect the national security and territorial integrity of Ukraine” (Dek, Kononova, and Marchenko 2019, 58). According to Oleksandr Turchynov, secretary of the National Security and Defense Council of Ukraine at the time of the ban, the forbidden websites were used for Russian propaganda, state-driven surveillance, and cyberattacks against Ukrainians (Andrusieczko 2017). The Ukrainian authorities argue that the ban is an important countermeasure against these information threats following the Russian annexation of Crimea in 2014 and the information war on Ukraine (Kiselyova and Prentice 2017; RFE/RL 2017). As a result of the annexation, the Ukrainian government could only fully implement the ban in territories under its control and not in Crimea. For the purpose of this article, I refer to the affected region north of the Crimean peninsula as “Mainland.”

To this day, Ukraine remains divided by political tension between pro-Ukrainian citizens (consisting of both ethnic Ukrainians and Russians), who support Ukrainian sovereignty, and the pro-Russian minority, who praise and support the Russian Federation or Kremlin-backed separatists in southeast Ukraine (Laruelle 2014, 2016). VKontakte is a popular source of entertainment in the post-Soviet space, and it has a history of being an important source of pirated music (Kiriya and Sherstoboeva 2015; Popkova 2019). Following the Kremlin’s turn toward strict internet regulations, VKontakte has fallen under great pressure and control from Russian authorities (Economic Times2014; Pan 2017). While most of the content on the platform is unrelated to politics, researchers, journalists, and authorities have argued that VKontakte is also a platform for pro-Russian propaganda (Dek et al. 2019; Van der Vet 2019; Volchek and Sindelar 2015), disinformation, Russian surveillance, cyberattacks (Andrusieczko 2017), and recruitment for the separatist movement (Deutsche Welle 2017). In this sense, the VKontakte users are potentially exposed to pro-Kremlin content, through either their pro-Russian friends and family or their newsfeed.

The Russian surveillance “threat” is relevant for both civilian and military targets. Shklovski and Wulf (2018) find that, despite military regulations, Ukrainian soldiers in the war zone use social media (including VKontakte) to search for information and to maintain personal contacts, despite army regulations. This even applies to soldiers who are aware that the enemy may potentially use media surveillance to geolocate Ukrainian positions and to select targets for artillery strikes (7, 10).

The VKontakte ban took place in a time of war but simultaneously also in a relatively democratic context, where citizens still enjoy legal access to a wide range of media outlets and platforms. The ban was not intended to reduce the overall access to social media but rather to push users away from the Russian VKontakte and toward other media alternatives, such as Facebook, Twitter, and Instagram, which are controlled by neither the Russian nor the Ukrainian authorities.

Theory: Mechanisms Behind Censorship

The literature on censorship describes multiple factors that may cause people to abandon or desist from consuming, sharing, or producing forbidden content, as well as instances when censorship “backfires.” Overall, this literature can be divided into at least two strands, each with its own theoretical view on the mechanisms behind censorship. On one hand, censorship may work by increasing the “costs” of accessing forbidden content in terms of time, money, or effort. I refer to this as the “accessibility view” on censorship. On the other hand, censorship can be seen as a signal that calls the citizens to fall in line with government policy, creating fears of reprisal or social exclusion. For the sake of this article, I will refer to this line of thought as the “political-signaling view.” As I argue below, the two strands explain not only “successful” censorship but also unintended outcomes that go against the censorship goals or undermine the censors.

The accessibility view on censorship

The accessibility approach emphasizes the accessibility of the forbidden media and the apolitical evaluation of whether accessing content is difficult. Instead of banning undesired content entirely, the government may effectively limit user engagement with the content by creating what Roberts (2018) refers to as friction: slowing down connections or blocking content in ways that can still be accessed through circumvention tools, such as VPN. By putting up minor barriers to censored content, the government increases the incentives to use more easily available, noncensored alternatives, while simultaneously avoiding the full-scale persecution of the masses and the potential political backlash that may follow (Dickson 2016; Roberts 2018).

Numerous studies indicate that citizens in authoritarian states respond to censorship by accessing available content instead (Chen and Yang 2019; Stockmann 2013). Interestingly, the literature based on the accessibility view emphasizes that the users’ impatience and apolitical use of social media may also backfire—especially if the censored platform cannot be replaced by a similar and freely available alternative to fill in the gap (Hobbs and Roberts 2018). For instance, Hobbs and Roberts describe how the abrupt Chinese government ban on the relatively apolitical Instagram increased the online traffic toward prohibited and more political platforms such as Twitter and Facebook. They argue that this pattern is driven by a “gateway effect.” The concept refers to a mechanism whereby the motivation to access a newly banned platform opens up for skills (i.e., how to use VPN) that give access to other websites that have long been censored (623–24).

It is important to stress that this perspective does not explain the response to censorship by referring to the users’ political attitudes toward censorship or the government. On the contrary, this strand of literature is in line with the so-called cute cat theory, which emphasizes that the search for entertainment—not political attitudes—is the main driving force behind the citizens’ consumption of online content (Hobbs and Roberts 2018; Zuckerman 2014, 624).

There are few studies that test the accessibility view on censorship in a democratic context, likely because of the literature’s focus on authoritarian or hybrid regimes. The view, however, is highly relevant in democratic or semidemocratic societies. This is the case, precisely because liberal states are ideologically bound to prioritize “softer” tools—such as adjustments to the cost of going online (in terms of time and effort)—over large-scale persecutions and fear of reprisal.

The political-signaling view on censorship

Whereas the accessibility view emphasizes the more practical and apolitical aspects of censorship, the political-signaling view shifts the focus toward the political signal behind censorship as well as the role of political attitudes in the public response toward such policies. The political-signaling approach sees state-driven censorship and propaganda as a means of demarcating forbidden and undesired conduct. Governments can therefore use censorship as a political signal to socialize citizens into the desired behavior—through implicit political messages, fear, or uncertainty (Huang 2015; Stern and Hassid 2012).

For instance, Stern and Hassid (2012) reveal how the Chinese government uses an atmosphere of uncertainty to govern journalists, editors, and lawyers. Not only is the exact line between forbidden and allowed information unclear, it is also constantly changing. This leads to politically engaged individuals creating their own bottom-up explanations for which content is forbidden (and why) and internalizing a practice of self-censorship (1240–41).

This literature strand explains not only successful censorship but also instances when such interventions backfire by provoking a political response. The political signals inherent in censorship stand as a possible driving mechanism behind the previously mentioned Streisand effect. Here, the apparent censorship generates more traffic toward the forbidden content by making people (1) more aware of the forbidden content, (2) more curious, and (3) outraged (Jansen and Martin 2015). While the first two aspects of the Streisand effect are not political per se, the third mechanism generates a backfire effect due to a political response against the censorship policy itself. Using a more theoretical approach, Shadmehr and Bernhardt (2015) argue that censorship may cause the population to evaluate the regime more negatively in some instances, because the lack of information may promote a belief that “there might have been bad news that was censored” (280). In their empirical study, based on news media content data and approval surveys, Gläßel and Paula (2020) find that the German Democratic Republic’s censorship of information about the emigration crisis in 1989 backfired by causing outrage among people who detected state misinformation through access to Western television.

Hypotheses

In this article, I use a multidimensional approach to analyze user affiliations with Russia and Ukraine. First, I examine the political aspect of the affiliations by distinguishing between users with pro-Russian and pro-Ukrainian attitudes. The former refers to praise and support for Russia; the latter, to support of the Ukrainian national state. Second, I capture the social aspect by examining social ties to individuals living in Russia.

My hypotheses and initial theoretical expectations are grounded in a political-signaling view for several reasons. Unlike the more covert or subtle friction described in the accessibility literature (using China as a case), the Ukrainian ban against VKontakte was officially announced by the censors and widely discussed by the public. The censorship policy itself was described as a political move against Russia and a matter of national security. The government, various journalists, and civil society groups framed the use of VKontakte as a political choice, a lack of patriotism, or even a part of the Russian information war against Ukraine. Using VKontakte in this heavily politicized context can be interpreted in the Ukrainian public as indifference toward patriotic ideals in times of war, whereas compliance with the ban can be a sign of political loyalty toward Ukraine as a nation-state.

If one sees censorship as a way for the Ukrainian government to signal that using VKontakte is both unpatriotic and undesired by the state, one would expect pro-Ukrainian users to become less active on the Russian platform out of political support for Ukraine and resentment toward the Russian “aggressor state” or due to a fear of being stigmatized as “unpatriotic” by their Ukrainian peers. Following this line of thought, I expect the ban to succeed overall, because the majority of the Ukrainian population is relatively supportive toward Ukraine’s sovereignty, while the pro-Russian individuals remain a minority (Arel 2018). I therefore begin with the following hypothesis:

H1.

The censorship reduces online activity among VKontakte users in Mainland.

It is important to note, however, that the accessibility view would predict a similar outcome for different reasons: the ban would reduce the online activity by forcing users to install a VPN, regardless of their political orientation, and therefore slow down the log-in procedure or the connection. However, there is an important difference between the two views when it comes to the role of political attitudes. According to the logic in the political-signaling view, users with strong Russian affiliations would be less affected by the ban for at least two reasons. First, they may be less affected by the social pressure to reduce their activity on the Russian platform because they are already stigmatized as “pro-Russian” in the context of the armed Russian-Ukrainian conflict. Second, the Russian-affiliated users would be more politically outraged by the “anti-Russian” policy and therefore more likely to resist a ban that is heavily embedded in anti-Kremlin sentiment. This theoretical expectation contrasts to the more pragmatic accessibility view on censorship, which would predict an equal decline in both groups due to increased costs of going online, regardless of the users’ political attitudes toward Russia and Ukraine. Using this view as a point of departure, I formulate the next hypothesis:

H1a.

Pro-Ukrainian users are more affected by the censorship than pro-Russian users.

As mentioned earlier, the Ukrainian government sees VKontakte as not only platform for Russian surveillance but also a pathway of pro-Kremlin propaganda. From a theoretical point of view, VKontakte can be seen as a platform for “participatory propaganda” in which the Ukraine-Russia conflict is socialized by ordinary users (Asmolov 2019) who actively take part in producing and sharing propaganda with their friends in Ukraine and abroad. The cross-national ties are key in this context. A Ukrainian user with many social ties to individuals in Russia may be heavily embedded in Russian society, either through friends and family or by having previously lived in Russia. The user likely has high potential exposure (Hjorth and Adler-Nissen 2019) to pro-Kremlin news, political narratives, and worldviews that are widespread among contacts in Russia (Toal and O’Loughlin 2018). One would therefore expect these users to be more pro-Russian and more likely to resist the anti-Russian ban, which may essentially jeopardize their online ties to peers in Russia. In line with this, I supplement the hypothesis above with the following hypothesis:

H1b.

Users with fewer social ties to individuals living in Russia are more affected by the censorship than users with many social links to Russia.

According to a report published by NATO StratCom, the overall number of wall posts in Ukraine did fall substantively following the ban (Dek et al. 2019, 58). It is worth noting that pro-Russian posts increased, which, according to the authors, indicates “users moving to the pro-Russian infosphere” (51). This alone is in line with my theoretical expectations based on the political-signaling view above. However, the report does not use an explicit natural experiment design, nor does it compare the effect of the ban among users with strong and weak affiliations to Russia. The report presents little information on the sampling strategy used to select users or posts, and it does not document the clustering algorithms used to infer the ideology of posts. While the report may offer useful and indicative results, it is difficult to draw causal conclusions on the effect of censorship among different societal groups because of the above-mentioned issues. Further systematic inquiry is needed if one is to understand how the effect is mitigated by individual-level factors.

Censorship as an Exogenous Shock

Before proceeding further, I first present arguments for why the ban can be seen as an exogenous shock to the online media ecology (i.e., the timing of the ban being largely unrelated to online activity on VKontakte). The executive order to ban a list of Russian websites, including VKontakte, was signed by Ukraine’s president at that time Petro Poroshenko on May 15, 2017, and publicly disclosed the next day, with no prior announcement. Before the ban, there had been only a few waves of interest in the topic in February, when a Ukrainian public official suggested that the government should ban VKontakte, with no indication of whether this would occur or when (see fig. A1).

In this sense, the sudden ban can be seen as an exogenous shock to the online environment, with the exact timing being largely unanticipated by the public and therefore “near random” from the perspective of the user. This is important, because knowledge of the ban date would give users the opportunity to prepare by decreasing their activity in advance (e.g., migrating to other platforms) or increasing activity before the ban in defiance against the policy. This could bias the estimated effect size of the actual ban. Because the exact timing of the actual censorship in 2017 is unexpected by the users, the period immediately before the ban may approximate the counterfactual scenario in which the “treatment,” in the form of a ban, did not occur. From a research perspective, this enables a natural experiment setting with a relatively clear delimitation between the period before (control) and after (treatment) the ban separated by a narrow window of up to 2 weeks for the policy to be fully enacted.

According to journalists, some users began reporting the effects of the ban as early as the evening of May 17 (UNIAN 2017), although the ban was not fully implemented among all internet service providers at this stage. Because of technical constraints and privacy concerns, it is not possible for me to pinpoint the exact hour of the ban for each individual user. For pragmatic reasons, I treat 00:01 a.m. on May 18 as the beginning of the full-scale implementation. As I show in the analysis, posting activity fell drastically on this day.

The Ukrainian government cannot implement censorship in all of its regions, because parts of its territory has been annexed by Russia in 2014. I use this spatial variation to distinguish between users who have not been exposed to the Ukrainian ban, as they are located in Russian-occupied Crimea, and those who were likely exposed to the ban, as they are located in Mainland (i.e., on the nonoccupied side of the Crimean border in Kherson Oblast).

I use Crimea for comparison—and not neighboring regions countries like Russia, Belarus, or Poland—because of its relative similarity to adjacent Ukrainian regions as well as a clear border separating censored and noncensored territories, unlike the frontline in the war-torn Donbass region (see app. G). It is important to note that the geographical border separating the censored and noncensored users is not random. For this reason, the spatial delimitation does not in itself enable a natural experiment research design, unlike the as-if-random, temporal “line” separating the censored and noncensored time period. However, the comparison of online activity across geographic space adds both valuable nuance and validation to the comparison of online activity over time.

Data

To compare the change in VKontakte before and after the ban, I collect publicly available data on VKontakte from the company’s own application programming interface (API).1 All of the user-level data were collected during August 2018, while the collection of wall posts took place in August–September of the same year (see app. T for an overview of the collected data). The data collection consists of six steps.

First, I identify all of the cities in the VKontakte database that are located in administrative regions (“oblast”) adjacent to the border separating Crimea and Mainland. In the second step, I use Google Maps API to automatically locate coordinates for each city. I test the reliability of the automated approach by manually examining the results for each city. Of the 370 cities, only six have been misplaced in the automated process, which I have then corrected manually.

Third, in order to filter out cities located more than 50 kilometers from the border, I use a geographic information systems approach to compute the distance between each city and the nearest point at the administrative land border separating Crimea from Mainland. This relatively narrow delimitation is intended to maximize the regional similarity between the towns exposed to censorship in the treatment region and towns in the control region that have not been exposed, while simultaneously keeping the distance bandwidth broad enough to capture a sufficient number of public profiles from different political groups.

In the fourth step, I sample the individuals by using the VKontakte API to return data on up to 1,000 users from each city within 50 kilometers of the border. I refer to this initial group of users as the “seed.”

In the fifth step, I develop a crawler to initiate a snowball sampling with the seed group as the initial starting point. In this crawl, I first download the metadata for the friends of the seed group, only keeping those that live in cities within the 50 kilometer distance from the border according to the user’s self-registered data. I then download the metadata for the friends’ friends, continuing the same procedure six times. Because of the highly interconnected nature of both online and offline networks, this snowballing strategy yields a nearly complete sample of 109,191 users within 50 kilometers of the border (as well as their 3.2 million friends from around the world with self-registered locations). Approximately 65,213 of the users within the 50 kilometer border logged on to the website for the last time at least 30 days before the ban or during the period after the ban. Like the rest of the data, the last log-in dates for the public accounts are collected using VKontakte’s API. This information is also a highly visible feature of the public profile and easily accessible by all VKontakte users. In order to ensure that the data are not biased toward users with high posting activity after the ban, this sampling step is carried out independently of the users’ posting behavior (the data therefore also include accounts with no public posts). The distribution of these users on both sides of the border between Mainland and Crimea is illustrated in figure 1.

Figure 1. 
Figure 1. VKontakte users within 50 kilometers of the Crimean border. All of the users have logged in within 30 days of the ban.
View Large ImageDownload PowerPoint

In the final step, I select a smaller subsample of 3,553 active users and all 2,127,398 public posts from their respective walls to capture individual-level online activity before and after the ban. I select the individuals using a stratified sampling, to ensure that I have users in the different political groups from both sides of the border (approximately 590 users per group in Mainland and 660 in Crimea) as well as randomly selected users from the pool of 65,213 active profiles. I then remove 51 profiles that have posted on average more than three posts daily (26,456 posts in total) in the 90 days before the ban (treatment), in order to filter out extreme outlier accounts that are likely automated (i.e., inauthentic bot accounts; Varol et al. 2017). These profiles account for less than 1.5% of the original subsample. The findings in this article remain robust when also including the hyperactive users. While VKontakte API provides data on an account’s last log-in time, the website does not disclose when the account was created. In order to ensure that the profiles existed before the ban, I further delimit the data to the 3,024 users who have uploaded at least one public post on their own walls before the May 18, 2017, implementation date. Descriptive statistics for the final subsample used in the analysis are available in appendix A. The findings remain robust when including users without a single public post before the ban (see app. Q). In total, the subsample in the analysis section includes 1,067 pro-Russians, 1,112 pro-Ukrainians, and 845 users from both sides of the border. While the random users are not necessarily politically neutral, they are likely less pro-Russian or pro-Ukrainian than those who publicly expose their political affiliation. I use the information about the users last log-in date to examine whether users accessed their VKontakte account after the ban.

The data from VKontakte are used to operationalize user activity, geographic location, social ties to Russia, and political attitudes toward Russia and Ukraine. These variables are described a follows.

User Activity is operationalized as the number of public posts written by the users on their own profile walls. In this sense, user activity is different from mere access to an account. Users may, for instance, circumvent the ban through VPN tools to passively access their accounts without generating any public posts on their walls. Geographic Location is operationalized using the profiles’ self-reported information on “current city” in August 2018. Social Ties to Russia is measured as the proportion of users’ friends who have publicly indicated that they live in Russia.

Political Attitude toward Russia and Ukraine is operationalized as follows: pro-Russian users as those who only follow pro-Russian VKontakte community pages, while pro-Ukrainian users are those who only follow pro-Ukrainian communities (see apps. B–D for details on sampling and coding of the communities). While the military conflict is often discussed in Russia in terms of an ethnic divide (Shklovski and Wulf 2018), speaking Russian does note equate with Russian identity, and Russian identity among Ukrainian citizens does not necessarily imply pro-Russian attitudes in the current Russia-Ukraine conflict (Arel 2018).

Difference-in-Differences Specification

In this section, I describe the empirical setup for measuring the effect of the ban on online activity. All of the models in this article are based on individual-level data, where each user-time observation represents a number of wall posts uploaded by the individual users on a given day.

I use a difference-in-differences (DD) approach (Card and Krueger 1993) to estimate the average effect of the ban on user activity among the individuals in the subsample living in Mainland within 50 kilometers of the border. Minor fluctuations in posting activity are common on both sides of the border, both before and after the ban (see fig. A2). One could reasonably assume that online activity in Mainland would have experienced a small and gradual decline—even if the ban would not have occurred. This means that a simple comparison of posting activity before and after the ban only in areas exposed to censorship (treatment) runs the risk of overestimating the effect size. For these reasons, I introduce users from noncensored Crimea as an additional control.

The advantage of using DD design in this context is that it allows treatment and control groups to be different, since the method is not based on the assumption that the treatment (the ban) is randomly assigned. The key argument for using this design is the similarity in time trends between the two groups—that the Crimean trend approximates the counterfactual scenario in Mainland where the ban did not occur.

I specify the DD regression using the following ordinary least squares model:

yit=α+β1Banit+β2Mainlandit+β3Banit×Mainlandit+δX+εit.(1)

In this equation, yit indicates the mean number of posts uploaded by user i during day t, where the earliest implementation date (May 18, 2017) is standardized as 0. The binary variable Banit indicates whether the posts are uploaded by the user before or after the ban, and Mainland reflects whether the user lives in Mainland (Mainlandit=1) or Crimea (Mainlandit=0). The interaction term, β3Banit×Mainlandit indicates the main effect of the ban when using Crimea as a control. In the other words, the term reflects the difference between the change in the mean number of posts per day in Mainland (treatment group) and Crimea (control group). The main hypothesis predicts that the ban has a negative effect on posting activity (β3Banit×Mainlandit<0). Additional control variables include total number of VKontakte friends per August 2018 and a dichotomous variable indicating whether the post was written during the weekend as well as the users’ self-reported gender.

To test whether the effect is mitigated by political affiliations and social ties to Russia, I create two additional DD models only for pro-Russian and pro-Ukrainian users from Mainland. Here, I replace Mainland with (1) a binary variable indicating whether the user has an above-median or up-to-median proportion of VKontakte friends and (2) political affiliation toward Russia and Ukraine, respectively. If the effect of the ban is mitigated by these factors, the respective interaction terms will be significantly different from zero.

In the final step, I examine the difference between the effect size of the ban among the various groups, by introducing Crimea as a control. The literature refers to this approach as the difference-in-difference-in-differences, or triple differences (DDD), method (Yelowitz 1995). More specifically, I expand the equation above by introducing a three-way interaction between Ban, Mainland, and a dichotomous variable (Pro-Russian) indicating whether or not the user is pro-Russian:

yit=α+β1Banit+β2Mainlandit+β3Pro‐Russianit+β4Banit×Mainlandit+β5Banit×Pro‐Russianit+β6Banit×Mainlandit×Pro‐Russianit+δX+εit.(2)

This triple difference is indicated by the interaction term Banit×Mainlandit×Pro‐Russianit

, which can be broken down into several parts. The interaction Banit×Pro‐Russianit shows the difference between the change in mean posting activity among pro-Russian and pro-Ukrainian profiles in Crimea (which is in the reference category). This provides an estimate of how much more pro-Russian users are affected by the ban than the pro-Ukrainian ones. By introducing Mainlandit to the interaction term, the estimate shows the difference between the estimate for Banit×Pro‐Russianit

in Mainland (treatment) and Crimea (control). If, for instance, the difference between the pro-Russian and pro-Ukrainian users is identical on both sides of the border, the interaction term will be close to zero. I use the same approach by replacing Pro-Russianit with a binary variable, Friends in Russia, which indicates whether the users have an above-median (Friends in Russia = 1) or up-to-median proportion of VKontakte friends in Russia (Friends in Russia = 0). Because users in Crimea have on average a higher proportion of VKontakte friends in Russia, I compute two separate median values for users in Mainland and Crimea, respectively.2 I use cluster robust standard errors for all of the DD and DDD models, in order to take into account the structure of the panel data, where there are multiple user-time observations for the same individuals across time.

Last Log-Ins

Before proceeding with the analysis of daily posting activity, I first examine whether users log in to their accounts after the ban or whether they last did so during the ban period. “Last log-in,” in this case, refers to the last time a nondeleted account logged in as of August 2018, the data collection period. Of the 109,191 accounts from both sides of the border in this study, 65,213 (59.7%) last logged in during the 30 days before the ban and the data collection date. The remaining 40.3% logged in for the last time 1 month before the ban or earlier.

The findings suggest that the vast majority of the nondeleted accounts continued to log back in after the ban. Approximately 94.1% of the 23,506 active users in Mainland logged in at least once after the implementation date, and 90.8% continued logging in more than 30 days after the ban. For users in Crimea, the figures are 99.0% and 98.3%, respectively (see app. H for an in-depth comparison between Mainland and Crimea). Of all the 1,433 users in the subsample from Mainland who have posted at least once before the ban, 93.4% continued logging in more than 30 days after the ban, even though the policy remains both legally and technically in place to this day. For pro-Russian, pro-Ukrainian, and random users in Mainland the proportions are 96.4%, 91.8%, and 91.4%, respectively. The pattern is illustrated in figures 2A and A3. The difference between the proportions for the pro-Russian and pro-Ukrainian users is statistically significant (p<.05

) albeit not substantively large. This indicates that a large percentage of users have found a way of avoiding the censorship, some through freely available VPN services. However, these numbers do not include the change in posting activity among the users who have bypassed censorship. I therefore turn to an analysis of a smaller subsample of users from both sides of the border in order to estimate the impact of the ban on their daily posting activity.

Figure 2. 
Figure 2. Last log-ins and posting activity. A, Proportion of users who have access to their VKontakte profile. The users are considered to have access to VKontakte until their very last log-in date. All of the 3,024 accounts are assumed to have existed throughout the entire period before the censorship. This can be confirmed for at least 98.84% of the accounts (see the description in fig. A3 for details). Note that the decline begins before the ban. This is because the data are based on last log-ins; that is, some users log in for the last time, not knowing that they will be locked out by the sudden ban. B, Posting activity for all 3,024 accounts. Horizontal lines reflect observed means for the respective periods and groups.
View Large ImageDownload PowerPoint

The Overall Effect of the Ban

How did the Mainland posting activity change compared to northern Crimea within just 50 kilometers, where there was no ban? To answer this question, I use DD to estimate the mean effect of the ban on posting activity in Mainland. In this part of the analysis, I use a subsample of 3,024 profiles from pro-Russian, pro-Ukrainian, and random users from both sides of the border. Each of the 547,344 user-day observations reflects the number of posts uploaded by the respective individual on a given day within the bandwidth of 90 days before and after the cutoff date (181 days in total), standardized as 0. Figure 2B shows the observed mean number of posts per user for all of the 3,024 users in the subsample. The figure illustrates the drastic reduction in the daily posting activity for all Mainland users. In comparison, the ban is accompanied by a relatively small, continuous decline in Crimea.

The effect size of the sudden censorship policy in the DD model is reflected by the interaction term Ban × Mainland in figure 3 and table I1. The ban reduced the posting activity for the average user by 0.8 posts per week or 0.114 posts per day (t=−12.4075

, p<.01

) in Mainland, when using a 30 day bandwidth before and after the ban (61 days in total). In other words, Mainland online activity (treatment group) declined by an additional 0.114 daily mean number of posts for the average user, compared to the change in Crimea (control group).

Figure 3. 
Figure 3. Change in mean number of posts per day after the ban, 95% confidence intervals. Reflects models 1 and 2 in table I1, which are based on data from all 3,024 users (including pro-Russian, pro-Ukrainian, and random users).
View Large ImageDownload PowerPoint

This effect size is substantively large. The ban decreased the daily mean number of posts per user by 45.42%, according to the most conservative estimate (based on 30 day bandwidth). When using the 90 day bandwidth, the decline is equivalent to 47.01% of the preban daily mean number of posts.

The Difference in Effect Size

The findings above confirm the first overall hypothesis: the ban reduces the online activity on VKontakte in Mainland, also when examining posting activity. This leads to the following question: To what extent is the censorship effect mitigated by different factors?

Figure 4 gives a descriptive overview of the observed mean number of posts per user for each day for the different groups and all of the users combined. The change in posting activity appears strikingly similar when visually comparing user groups with different (political and social) affiliations to Russia and Ukraine. In this section, I use DD and DDD models to test whether the difference is statistically significant and substantively large when including control variables.

Figure 4. 
Figure 4. Change in mean number of posts per day after the ban, for all users and the split sample. Posting activity for all users, as well as the split samples: pro-Russian, pro-Ukrainian users, as well as those with up-to-median and above-median proportions of friends in Russia. Horizontal lines reflect overall means for the respective periods and user groups.
View Large ImageDownload PowerPoint

I turn to this question by testing hypothesis 1a, which predicts that the anti-Kremlin ban affects pro-Ukrainian users more than pro-Russian users. This part of the analysis is limited to pro-Russian and pro-Ukrainian users (without random profiles), in order to enable the comparison between the two groups. The variables of interest for the DD and DDD models are illustrated in figure 5.

Figure 5. 
Figure 5. Change in mean number of posts per day after the ban, with 95% confidence intervals. Reflects models in tables I2 and I3, which are based on data from users with political affiliations (not random users). The order of the interaction terms is based on the model numbers, so that the first interaction term (top) is based on model 1 and the fifth interaction term (bottom) is based on model 5.
View Large ImageDownload PowerPoint

I find no statistically significant difference between the two groups when using DD and a significance level of 5% to examine the data only from Mainland. The difference for the DD models is indicated by the coefficient for the interaction term Ban × Pro-Russian in the output in figure 5 and table I2 (model 3). While the difference is statistically significant at a 10% level for the 90 day bandwidth, these estimates indicate that pro-Russian users are more affected by the ban than pro-Ukrainian users. Here, the ban reduced the average daily activity among pro-Russian users by 0.017 posts more than their pro-Ukrainian counterparts (t=−1.8229

, p<.1

), contrary to the hypothesis.

Similarly, the results show little evidence supporting the hypothesis when adding Crimea as a control to the DDD model. Here, the difference between pro-Russian and pro-Ukrainian users is reflected by the interaction term Ban × Pro-Russian × Mainland in figure 5 and table I2 (model 5). According to the most conservative estimates—based on a 90 day bandwidth—the effect size of the ban is 0.004 posts greater among pro-Russian users than pro-Ukrainian ones (t=−0.3202

, p=.7488

), contrary to the hypothesis. While this difference appears even greater, when narrowing the bandwidth down to 30 days, it remains statistically insignificant. The data therefore provide little evidence in support of hypothesis 1a.

I find similar results when testing hypothesis 1b, which predicts that individuals with fewer social ties to users in Russia are more affected by the ban than those with extensive social ties linking to the country. I examine the effect by comparing users with an above-median percentage of VKontakte friends living in Russian to those with an up-to-median proportion. There is no statistically significant difference in how much the two groups are affected when only using data from Mainland. This is reflected by the Ban × Friends in Russia interaction term in figure 5, when using a bandwidth of 30 and 90 days. The pattern is consistent when introducing Crimea as an additional control in the DDD model, by including the following interaction term: Ban × Friends in Russia × Mainland. According to the most conservative DDD estimate, the ban has reduced online activity by 0.0185 posts (t=0.8212

, p=.4115

) more among users with above-median proportion than those with fewer friends in Russia—contrary to the theoretical expectation. However, this difference remains statistically insignificant. These results provide little evidence supporting hypothesis 1b.

I find similar results when rerunning the models behind figure 5 with user-level fixed effects (see fig. I1) or negative binomial regressions instead of ordinary least squares. Furthermore, the findings remain robust when testing for long-term effects by expanding the time period, using regression discontinuity in time models instead of DD, using different independent variables (including strength of ties to Russian users), and reiterating the analysis on users from Kyiv. The results based on DD and DDD remain the same when delimiting the analysis to users who managed to technically circumvent the censorship by logging in at least 30 days after the ban. In other words, the censorship reduced VKontakte activity even among users who were capable of circumventing the ban. I validate the political affiliation variable by examining whether users labeled “Pro-Russian” are more connected to Russia than “Pro-Ukrainian” users through friendship networks, reposts, and use of Russian websites, as one would expect. I find that this is the case. I conducted a placebo test by reiterating the analysis on the same dates one year before the ban. By doing so, I show that the results are not driven by reoccurring seasonal events. Appendix J provides an in-depth overview of the robustness and validity tests.

A Pragmatic Response to Censorship

The vast majority of pro-Russian and pro-Ukrainian users continue logging onto the forbidden website despite the ban. Pro-Russian users are slightly more likely to maintain passive access to the platform than their pro-Ukrainian counterparts. However, the difference is not substantively large, and they are at least as affected when it comes to online activity. Why do users, who can legally access VKontakte and have the technical know-how to do so, still choose to shift their attention away from the Russian platform?

From the political-signaling point of view, the decision to either contest or comply with the government ban is driven by political attitudes toward the regime. From this perspective, the ban is effective because it actively signals that using the online services provided by the “aggressor state” is unpatriotic and a potential threat to national security. In other words, the underlying reason for reducing activity on the platform is more political than practical. One would therefore expect pro-Ukrainian individuals and those with fewer social ties to Russia to be more compliant with the anti-Kremlin ban on Russian social media by reducing their online activity. The findings present little evidence for this view; on the contrary, if there is a difference between the two groups, pro-Russian users and those with relatively many social ties to Russia are more likely to comply with the anti-Russian ban by reducing their online activity on VKontakte.

The potential difference between the two groups cannot be explained by fears of legal reprisals or social stigma alone. More than 90% of individuals in the subsample continued to log in to their public accounts at least one month after the ban without hiding their last log-in date. This means that their friends, strangers, and authorities alike can see that the respective users continued using VKontakte at least one month after the ban. This information is visible even if the users do not write any posts.

Implications

I argue that the behavioral response to censorship in Ukraine is largely in line with the competing accessibility view on censorship, represented by scholars such as Hobbs and Roberts (2018). Seen from this theoretical perspective, the response to censorship is driven by consumer impatience and the increasing costs of going online and to a lesser extent political affiliations with Russia or Ukraine. This view explains why pro-Russian users and those with potential exposure to Russian information networks are no more likely to resist the anti-Russian ban than pro-Ukrainian users or those with few social ties to Russia.

The Ukrainian ban offers a hard case for the practical accessibility view on censorship. Not only does the government actively politicize Russian media as a means of information warfare against Ukraine, the users are also encouraged to demonstrate loyalty to their country by ceasing to use services from the aggressor state. Yet the findings suggest that the mechanisms described by Hobbs and Roberts (2018) and Roberts (2018) in their study of China offer a powerful explanation even in a heavily politicized context of war, where the forbidden media itself is portrayed as a weapon.

Whereas Hobbs and Roberts (2018) discuss the incentive to acquire the skills to access forbidden websites via VPN, the Ukrainian case shows that the mechanisms behind the accessibility view may hold even when the majority of the politically interested users find a legal way to circumvent censorship free of government persecution. The practically oriented and pragmatic user has an incentive to log on to VKontakte occasionally, either to respond to unanswered messages or to check for new ones. However, free VPN services often add friction by slowing down the connection or providing secure browsing only for a limited traffic volume. For example, at least three out of the five VPN services that the VKontakte team has publicly recommended for circumventing the Ukrainian ban (VKontakte 2017) require individuals to pay a fee for a fast connection or unlimited browsing. One of the free options requires individuals to use the Opera browser. This introduces additional obstacles for those who are used to Google Chrome, Firefox, or other widespread browsers. Even in this case, some bloggers have complained that their connection began to lag more when using the Opera VPN—right after users from Ukraine began to switch to the service (help-wifi.com 2018).

The cost of logging in increases dramatically, if the user chooses to move away from the free options in order to acquire a faster and more efficient service. A paid VPN service may cost an additional US$3–US$7 per month—a sizable amount in Ukraine. However, VPN clients may potentially complicate the log-in procedure even if they succeed to provide a fast connection. A pro-Ukrainian resident in Kyiv explained to me in an interview why he no longer uses VKontakte on a daily basis: the VPN on his PC required more clicks and slowed down the connection by a few seconds. Interestingly, the same person could enter VKontakte automatically on his smartphone without extra clicks, likely due to a preinstalled VPN. Despite this, he still decided to stop using VKontakte as his main platform because he felt it was too troublesome to manage his account only from a smartphone and not being able to do so unhindered through a PC. This evidence is anecdotal and calls for more (qualitative and quantitative) research on the user experience of banned platforms. Nevertheless, the story emphasizes the potential complexities of managing forbidden accounts across multiple devices and operating systems.

Because of the friction created by the censorship policy, the impatient user has an incentive to switch to a cheaper alternative that is unaffected by the ban: Facebook. The American website has many of the same functions as its Russian competitor but can be accessed without VPN. While the data in this article do not reveal the extent to which the users in the sample have migrated to other social media platforms, aggregated data from online traffic monitors suggest that this may have been the case. According to data from SimilarWeb (an analytics firm), used in NATO StratCom‘s report, VKontakte’s ranking fell from the first most visited website in Ukraine on March 18, 2017, to the fifth most visited site on August 14, 2018 (Dek et al. 2019, 42). Facebook, however, jumped from the eighth to fourth place in the same period. This can be viewed as a great success, seen from the perspective of the Ukrainian government. As mentioned earlier, the purpose behind the ban is not necessarily to reduce social media behavior as a whole but to push users toward other media alternatives—away from a social media platform that provides Russian authorities with surveillance opportunities while showing no commitment to curb pro-Kremlin disinformation campaigns.

Because of these factors, even users with access to the “freedom technology” needed to circumvent online bans may still be highly vulnerable to government censorship. This is important, because it emphasizes that the effects of censorship cannot be evaluated by examining the technical ability of the population to bypass censorship alone. The findings also serve as a reminder to political scientists that practical and relatively pragmatic considerations may play a more important role in terms of how the users respond to online censorship than their political affiliations toward the censoring state or to its foreign “enemy”—even if the opposing political groups are involved in war against each other.

How generalizable are these findings? As mentioned previously, the results are in line with the accessibility view on information restrictions, which has been used to describe online censorship in China. The findings in this article suggest the pattern can be generalized to a more democratic and politicized context, such as the war in Ukraine. Following the logic of this accessibility view, one could expect a similar pattern in other countries if at least three conditions are met.

First, the ban must be technically implemented in a way that significantly complicates and slows down access to the forbidden source. This is not always the case. For example, the Russian authorities arguably failed to do so in their ban on the Telegram app in 2018 (see app. P). Because Telegram moved its service to Google and Amazon domains (i.e., through “domain fronting”), users could continue accessing the website with relative ease during prolonged periods. The service was officially unblocked in 2020.

Second, users must have unhindered access to similar services that can partly replace the banned product. In the case of Ukraine, VKontakte can be replaced with highly similar alternative: Facebook. Conversely, if Russian authorities were to ban Facebook, as they have previously threatened to do (Doffman 2019), the service could be replaced with VKontakte. In the example above, Telegram was likely difficult to replace in Russia with a more “cooperative” platform, because its core selling point at the time was precisely its unwillingness to share user data with the authorities. If there is no popular replacement for the banned platform the impatient users, some of whom are addicted to the product, would have had a much higher incentive to circumvent the ban (Hobbs and Roberts 2018).

Finally, I theorize that similar bans may be more effective, if the forbidden platform is already widely known. As pointed out by the literature on the Streisand effect, a ban may backfire by making the general public more curious and aware of the forbidden content. This was less likely during the Ukrainian ban in 2017, precisely because VKontakte, unlike Telegram in Russia during 2018, was already a mundane household name and a part of everyday life for the majority of online users. While these theoretical perspectives may offer a guideline for future inquiries, more cross-national research is needed to empirically examine when and how censorship succeeds in one context and not the other.

Limitations

For ethical reasons, I measure online activity by using only publicly available posts from public accounts and not private posts or messages. This article does not argue that politics do not matter in the citizens’ response to censorship or that politically engaged users are less affected by the ban than those who are disinterested in politics. Instead, the study is focused on the difference between users with strong and weak (political and social) affiliations with Russia. This question is pivotal to understanding the extent to which the ban has backfired by affecting one affiliation group more than the other.

The findings suggest that users with either pro-Russian or pro-Ukrainian affiliations are on average less affected by the ban than random users from the same region. One possible explanation is that users with political affiliations are on average more active online (see app. A) and therefore more invested in the platform. More research is needed to validate this view. Although Ukraine’s measure against the “unpatriotic” Russian platform may have made political users on both sides more prevalent in relative terms, it affected the pro-Russian users at least as much as the pro-Ukrainian ones. It is important to note that the pro-Russian users in the data set publicly follow pro-Russian communities in a time of war and are therefore likely to be relatively extreme in their political attitudes. More research is needed to test whether the findings hold among individuals with only slightly pro-Russian views or individuals who do not have any visible affiliation with Russia in the online realm. Similarly, this research does not exclude that the effects of the ban may vary depending on whether the users access VKontakte through a static connection or a mobile connection on a smartphone.

Furthermore, the study does not shed light on whether the ban has affected the consumption of pro-Kremlin content among Ukrainian users offline or whether it changed the popularity of different political topics (see Dek et al. 2019). In line with this, the findings do not reveal the extent to which the ban has affected political attitudes among ordinary citizens.

Instead, the results show that the Ukrainian authorities succeeded in achieving an important goal in their antipropaganda policy: to reduce online activity on Russian social media. From the perspective of the Ukrainian government, this is an important achievement because it sees Russian social media as a platform for foreign propaganda, surveillance, disinformation, and a national security threat.

The findings in this study are limited to instances when users respond to sudden censorship on social media. Although political and social affiliations have not played a major role in how the users have responded to the ban, it is still possible that these factors influence the consumption of Russian services in the long run, gradually and independent of sudden shocks to the media ecology.

Conclusion

This article shows that online censorship may work even when users have the technical and legal means to circumvent it. The Ukrainian government succeeded in reducing the overall online activity on VKontakte, which it perceives as a platform for foreign propaganda and a tool for surveillance used against Ukraine in times of war. Even a minor increase in the costs of entering on the banned website is enough to significantly shrink online activity, as long as there are more accessible alternatives.

The data support the accessibility view on censorship, which argues that the users respond to censorship primarily on the basis of the costs of going online (in terms of time and effort), not their political attitudes toward the regime. The anti-Russian ban, meant as a response to Ukraine’s conflict with Russia, affected pro-Russian users as much as pro-Ukrainian ones in terms of online activity. Similarly, the ban is at least as likely to reduce the activity among users with high levels of potential exposure to pro-Russian content, measured as social ties to Russia, as those with low levels. In other words, the ban did not make VKontakte more pro-Russian in relative terms.

The results are favorable from the perspective of the censor, who wishes to combat foreign propaganda and disinformation by using one of the most drastic countermeasures available. However, this article paints a more concerning picture for those who see censorship as a threat to democracy. This study contributes to the growing evidence suggesting that states can greatly regulate online consumption by setting up minor obstacles between the impatient user and the banned service. The Ukrainian case indicates that this may be true even in a relatively democratic context and among users who already have the technical means to bypass censorship and can do so free of persecution.

Given these findings, one would expect censorship in other countries to lead to a similar outcome. A sudden, large-scale ban would not make the opposition more predominant on the forbidden platform compared to the proregime movement. Similarly, if Russia were to use its newly upgraded censorship infrastructure to ban Facebook to prevent foreign influence, one would expect the ban to be successful from the point of view of the government if Russians were to respond in a similar manner as Ukrainians have. Understanding the mechanisms behind such interventions is important if one is to shed light on how states police cyberspace. This calls for more research on the individual-level factors that may mitigate the effects of censorship, together with cross-national comparative studies of how and when large-scale social media shutdowns are met with political resistance among citizens rather than user compliance.

AcknowledgmentsI would like to thank the JOP editors and anonymous reviewers for their detailed comments and invaluable work. I am very grateful to Rebecca Adler-Nissen, David Dreyer Lassen, Sune Lehmann, Margaret E. Roberts, Sarah Oates, Anders Wivel, Mikhail Suslov, Asmus Leth Olsen, Valentina Shapovalova, Frederik Hjorth, Benjamin Carl Krag Egerod, David Szakonyi, Joshua A. Tucker, James Pamment, Alicia Fjällhed, Lene Hansen, Kasper Møller Hansen, Barbara Maria Piotrowska, Arjen van Dalen, Marie Heldgaard Seestedt, Jeppe Vierø, Frederik Kristoffer Kjøller Larsen, Gregory Eady, and Simon Polichinel von der Maase for their thorough and helpful feedback. I would also like to thank everyone who discussed this article during events at the 2019 American Political Science Association Annual Meeting, Social Media and Political Participation Lab, Copenhagen Center for Social Data Science, and the PhD seminar as well as the workshops at the University of Copenhagen.

Notes

Important disclaimer: This article was written prior to the Russian Federation’s full-scale invasion of Ukraine on February 24, 2022, which is why it mainly reflects events prior to this. Edits have been made to make the reader aware of the invasion. However, the changes are minimal in order to stay true to the accepted version.

Yevgeniy Golovchenko (yg@ifs.ku.dk) is a postdoc in the Department of Political Science, University of Copenhagen.

1. For this purpose, I have used and modified code from the vkR package in R developed by Dmitriy Sorokin: https://cran.r-project.org/web/packages/vkR/index.html.

2. The findings remain robust when using a single median value for profiles on both sides of the border.

References

https://www.journals.uchicago.edu/doi/epdf/10.1086/716949

The post Fighting Propaganda with Censorship: A Study of the Ukrainian Ban on Russian Social Media appeared on BIDD.

Original Article

(Visited 1 times, 1 visits today)