Putin Fans or Kremlin Bots?

The Russian public reaction to Putin’s announcement of ‘partial mobilisation’ was far from that desired by Russia’s authoritarian regime. Up to 700,000 citizens left the country in order to dodge the draft over the course of two weeks, mass protests erupted across the country, and the approval ratings of the president and key political institutions slid as a result. In order to counter this reaction, the Kremlin mobilised both broadcast and online media. State media received a new set of guidelines [temniki] instructing them to emphasise the limited nature of ‘partial mobilisation’ and that it would affect only 1% of reservists – younger men with military experience. In addition, the regime mobilised bloggers to promote the hashtag #nopanic [#безпаники] and to compare the number of mobilised men to 1% of everyday objects such as a woman’s makeup bag, a portion of fries, or a pack of candies in order to downplay the scale of the event.

Over the past decade, the Kremlin has built a formidable digital army and infrastructure to shape public opinion online, including automated bots and paid authors on Russia’s multiple social media platforms. After the start of Russia’s invasion of Ukraine, state-organised digital “astroturfing” has been actively used to shape online discussions surrounding the war. While pro-war narratives on Russian social media have been investigated in recent months, these studies tend to focus exclusively on one platform: Telegram (TG). When compared to other Russian social media platforms, TG is more diverse and hosts both pro-regime and anti-regime channels. In addition, the increased attention it has recently received is due to the presence of influential pro-war bloggers – ‘military reporters’ (voenkors) and radical nationalists. As they have consistently criticised the implementation of the ‘special military operation’, they are often viewed as potential challenges for the Putin’s regime, and thus attract the attention of experts and journalists. However, TG represents only one pocket of Russia’s vast online social media space. Two other Russian social media platforms – VKontakte (‘InContact’, VK) and Odnoklassniiki (‘Classmates’, OK) have audiences of a similar or bigger size. As political discussions can vary significantly across social media platforms, we adopted a comparative approach to examine these platforms, focusing on possible signs of online astroturfing. Our analysis is based on a total of 1,544,918 messages (over 352,000,000 words) regarding the Russia-Ukraine war published from July to September 2022 on TG (3% of all messages), VK (60%), and OK (37%). We employed Brand Analytics to trawl social media content. The distribution of our data from the three platforms demonstrates that the system struggled to extract data from TG. This distribution likely reflects differences in the platforms’ data protection policies. Given unequal representation of the platforms in our data, the comparative results should be interpreted with caution.

OK is often considered to be the most pro-regime platform and has a much older audience than both VK and TG. According to multiple, extensive surveys, there is a positive correlation between age and support for the regime, the invasion, and mobilisation. Our findings provide additional evidence for the link between socio-demographic profiles and political preferences. We determined that OK is the most pro-war platform, followed by VK, while TG is the most polarised platform, including both pro-war and anti-war discussions. Given the peculiarities of the different platforms, one would expect VK and TG to be the primary targets of regime-controlled accounts and networks. However, we found that the activities of regime-controlled accounts are more visible on OK than on VK and TG. These findings point to possible mechanisms behind the Kremlin’s digital war propaganda. Instead of attempting to reach opponents of the war or users without any clear preferences, the regime’s astroturf communication seems to flourish in a predominantly pro-war environment. In line with both classical research on media effects and contemporary research on the effects of propaganda in authoritarian Russia, these findings suggest that the main strategy of the regime’s online astroturfing may be similar to the role of authoritarian propaganda: to reinforce the beliefs of those who are already pro-regime, rather than to win over new supporters.

The Kremlin Going Digital

Social media platforms play a crucial role in contemporary propaganda. Once praised as a ‘liberation technology’, these platforms are increasingly instrumentalised by authoritarian governments to spread falsehoods, to control political narratives, and to build various versions of ‘networked authoritarianism’. In the past decade, social media platforms have become an essential part of the Kremlin’s propaganda machine. While television is still used as the main conduit for propaganda narratives, this is complemented by extensive networks of bots, trolls, and paid influencers used to swing public opinion in the government’s favour and crowd out alternative and critical voices.

As a part of ‘third generation controls’ – the techniques which are used to actively shape rather than restrict the online sphere — Putin’s regime has actively used both trolls and bots to shape online discussions over the past 10 years. Using paid human commentators, the infamous Kremlin-linked Internet Research Agency (IRA) has conducted multiple disinformation campaigns attempting to sow discord in the U.S. Trolls are also actively used to shape domestic political discussions. What do trolls do? Researchers studying IRA-linked accounts have suggested that ‘foreign’ trolls mostly target existing divisions within societies in order to increase polarisation and to spread fear. For instance, American IRA-linked accounts assume the role of citizens with far right or far left views to amplify polarisation and sow discord, or simply use disinformation about fabricated crises to spread fear. It is much more difficult to assess the impact of trolls because they appear identical to real users. However, studies based on leaked accounts of trolls in Russia show that they might be not that successful at imposing a pro-regime agenda. Instead, they serve to distract citizens from political discussions to decrease critical sentiment.

Bot accounts occur much more frequently than troll accounts operated by humans. For instance, scholars have shown that the number of bot accounts tweeting about politics in the Russian Twittersphere during key events, such as the annexation of Crimea in 2014, might be as high as 80%. Unlike trolls who engage in discussions to influence beliefs, bots perform a variety of functions. For instance, they might create the appearance of popularity for officials by simply following the accounts of ministers, governors, and other officials. In addition, bots simply create information noise during key political events, such as protests, in order to make it more difficult for citizens to find relevant political information. Finally, bots post headlines and links to news stories to manipulate the result rankings of search engines and make pro-regime stories more visible in search queries. Bots are used by different political actors, not only the regime. For example, in 2015-17, there were a similar number of pro-Kremlin, anti-Kremlin, and neutral bots on Twitter. Similarly, at this current moment, Twitter discussions on the Russia-Ukraine war are also influenced by both pro-Russian and pro-Ukrainian bots.

Russian Social Media Ecology

Social media platforms drastically differ in the way they organise online communication. Russian LiveJournal, and more recently Facebook and Twitter, are well-known for fierce political discussions, while OK and VK remain largely apolitical. TG messenger, which emphasises user security, privacy, and freedom of expression, became a prominent communication and coordination tool during protest waves in Hong Kong, Russia, and Belarus. Social media are an important part of Russia’s authoritarian media ecology. According to 2021 data collected before the invasion, Russia’s most popular social media included VKontake (44%), Youtube (37%), Instagram (34%), and Odnoklassniki (30%), followed by Tik-Tok (16%), Facebook (10%), Moi Mir (Mail.ru) (5%), and Twitter (4%). Telegram’s audience has been steadily rising and had reached 21% by this time. Two platforms – Facebook and Instagram – were banned in March 2022 but remain accessible via VPN services. In addition to social media users, these platforms are populated with semi-automated and automated accounts, such as bots (automated accounts that produce predefined content) and cyborgs (humans who rely on scripts to produce content). After Western platforms were banned in Russia in Spring 2022, VK, OK, and TG solidified their dominant position as the major social media in Russia. The audience of VK increased to 62%, and OK’s audience increased to 42%. The audience of TG has increased to 55%.

Sometimes referred to as ‘Odnoglazniki’ (the one-eyed), OK is often considered a space for Putin’s electorate. Its audience is much older than the audience of other social media platforms. According to 2021 data, 7.4% of OK users were under 24, 16.9% between 24 and 34, 25.2% between 34 and 44, and the dominant 49.5% were older than 45. Public groups on OK often express anti-Western and pro-Kremlin sentiments and constitute the regime’s ‘Virtual Russian World’ not only in Russia, but in other countries with significant Russian-speaking populations, such as the Baltic States. While VK is a key part of the Kremlin’s online propaganda machine, its audience is much younger. According to 2021 data, a dominant 31.3% of VK users were under 24, 29% were between 24 and 34, 21.8% between 34 and 44, and only 18% were older than 45. Finally, the audience of the relative newcomer TG is slightly older than the audience of VK. According to 2021 data, 29.6% of TG users were under 24, the majority of 30.6% were between 24 and 34, 21.3% were between 34 and 44, and only 18.5% were older than 45.

Both qualitative studies and surveys suggest that age strongly correlates with support for Putin’s regime and the invasion. Older people are much more likely to support Putin’s ‘special military operation’ and more often express support for Putin’s decision to start ‘partial mobilisation’. Given these clear socio-demographic differences, we would expect to see clear ideological differences between online discussions about the war on OK, VK, and TG.

Ideological Spin on OK, VK, and TG Compared

What are the differences in ideological spin in online discussions about the war on OK, VK, and TG? To understand pro-war or anti-war attitudes across platforms, we compared the frequencies of words and phrases which contributed to dehumanising discourse, anti-war language as well as the distribution of terms used by the regime to justify the invasion, such as ‘denazification’ and ‘demilitarisation’. Dehumanisation is a tool often used by governments at war. It provokes hate while also undermining compassion for the enemy. Portraying Ukrainians as Nazis is a key pillar of the Kremlin’s narrative. We built a vocabulary of derogatory terms used by state propaganda to dehumanise Ukrainians, such as ‘Nazi’, ‘ukrofascist’, ‘ukronationalists’, ‘banderovetsi’, ‘ukropi’, ‘naziki’, ‘Zelensky’s gang/junta’, etc.

Figure 1. Dehumanisation vocabulary across platforms (normalised frequencies)

Figure 1 shows that there is a significant amount of dehumanising language on all three platforms. There were two significant spikes in the last two weeks of September 2022: immediately after 21 September and on 30 September corresponding to the announcement of mobilisation and the decision to formally annex four Ukrainian regions. Markers of dehumanising discourse are more frequent in posts on OK. As of 30 September, the amount of dehumanising language on OK was more than five times higher than that on VK and TG – less pro-regime platforms according to their socio-demographic profiles. However, the sudden appearance and disappearance of an enormous amount of dehumanising language on OK, when compared to the stable frequency on VK and TG, points to the artificial nature of this communication. Most likely, these fluctuations suggest that dehumanising language on OK comes largely from astroturfed communication via bots and paid influencers.

In order to further understand the difference in, and dynamics of, pro-war and anti-war discourse, we explored the usage of anti-war vocabulary. Figure 2 shows the aggregated frequencies of keywords typical for Kremlin opponents who critique the invasion, such as ‘Russian aggression’, ‘annexation’, ‘occupation of Ukrainian territories’, ‘Russian invasion’, ‘occupation of Donbas’, ‘occupation of Crimea’, ‘Russian occupants’, etc. The graphs reflect the fluctuations in the use of anti-war language across platforms between July and September.

Figure 2. Anti-war language across platforms (normalised frequencies)

As expected, TG is the most anti-war platform among the three, followed by VK then OK. While there are many pro-Kremlin channels on TG, independent media channels also have a strong presence and generate a substantial amount of anti-war content. Anti-war vocabulary is the least present on OK — the most pro-regime network according to its socio-demographic profile — with a steady decline of this language towards the end of September.

The frequency profiles of the key terms used by the regime to justify the invasion – ‘denazification’, ‘demilitarisation’, protection of the people in Donbas and of the Russian language – demonstrates the artificial astroturf nature of pro-war social media discourse on OK even more clearly (see Figures 3a, 3b).

Figure 3a. Justifications for the invasion: ‘denazification’ (normalised frequencies)

Figure 3b. Justifications for the invasion: ‘demilitarisation’ (normalised frequencies)

As we have previously demonstrated, the terms ‘denazification’ and ‘demilitarisation’ disappeared from TV coverage back in spring 2022 as they did not resonate with the public. However, they have remained a part of the astroturfing messages spread by the regime via state-controlled accounts on social media throughout the summer months of 2022. TG remains a relatively free space, devoid of official rhetoric, and VK users exhibit a mild tendency to re-produce official narratives about the war. However, the frequencies of ‘denazification’ and ‘demilitarisation’ on TG and VK are both steady and low in comparison to the abnormal, spiky patterns registered across OK. Figure 3 captures this intermittent use of ‘denazification’, where the peaks correspond to five times higher frequencies in OK posts than in VK posts. Note that ‘demilitarisation’, which was actively used by state-controlled accounts in the summer, disappeared from these astroturf messages towards autumn 2022. The inconsistent structure of communication on this topic across time and abnormal spikes (as compared to other social media platforms) are suggestive of its forced and artificial nature.

Pro-mobilisation and Anti-mobilisation Discourse Across Platforms

While differences in pro-war and anti-war discussions about the invasion allow us to get some understanding of how the war is framed both by social media users and the Kremlin, ‘partial mobilisation’ announced by Putin on 21 September appeared to be a litmus test for the Russian public. While it may have been easy to support an abstract war fought by professional soldiers far away, it is a completely different story when the war intrudes into the private lives of citizens and claims their lives. How did social media users react to ‘partial mobilisation’ across platforms? The reaction to ‘partial mobilisation’ on social media was diverse. The pro-war users praised the decision arguing that it had been long-awaited. Anti-war netizens quickly turned the word ‘mobilisation’ into ‘mogilisation’ (putting in graves) and launched a campaign against this decision. The lineplots in Figures 4a, 4b and 4c reflect the frequencies of three terms used by the critics of mobilisation – ‘mogilisation’, ‘stop war’, and ‘cannon fodder’ – in September 2022.

Figure 4a. Anti-war discourse at the end of September 2022: ‘mogilisation’ (putting in the grave) (normalised frequencies)

Figure 4b. Anti-war discourse at the end of September 2022: ‘stop war’ (normalised frequencies)

Figure 4c. Anti-war discourse at the end of September 2022: ‘cannon fodder’ (normalised frequencies)

Figures 4(a-c) demonstrates that the reaction to mobilisation was much more negative on TG and VK than OK: all of the solid lines which correspond to OK frequencies are found at the bottom of the graph which indicates how low these frequencies were. TG was the primary platform for spreading anti-mobilisation sentiment — the calls to ‘stop the war’ were most visible on TG. TG and VK adopted the terms ‘mogilisation’ and ‘cannon fodder’ to emphasise the cruel treatment of draftees by the army and the regime more frequently than OK.

Unlike OK, both TG and VK framed draft evasion and fleeing the country as desirable responses to mobilisation. The aggregated frequencies of phrases associated with the two strategies to avoid the draft – ‘fleeing the country’ and ‘dodging call-up’ – plotted in Figures 5a and 5b emphasise the differences between the social media platforms.

Figure 5a. Reaction to mobilisation in September: ‘fleeing the country’ (normalised frequencies)

Figure 5b. Reaction to mobilisation in September: ‘dodging call-up’ (normalised frequencies)

As can be seen in Figures 5a and 5b, TG users were more active in discussing leaving the country even before the mobilisation was announced. After 21 September this ongoing topic was superseded by discussion of ‘avoiding the draft’. VK users followed this trend, although at a lower scale. The massive OK network showed very little interest in the topic, at least publicly. Given that 51% of Russians do not have any savings, and 72% do not own a valid foreign passport, it is not surprising that on all three platforms the frequencies of phrases associated with dodging the call by avoiding interactions with military commissions (ukloneniye ot prizyva) got more traction than emigration.

Putin Fans or Kremlin Bots?

Russia has a diverse social media space. After Western platforms were banned in spring 2022, VKontakte, Odnoklassniki, and Telegram increased their share of audiences and became the dominant social media platforms. As there are significant socio-demographic differences between them, we believe that differences in how social media users discuss the war and the mobilisation across these platforms hint at the strategy behind the Kremlin’s online astroturfing. While Odnoklassiniki has the oldest audience and is a much more pro-regime space, Telegram has a younger audience and both pro-Kremlin and oppositional channels are widely present. VKontakte has the youngest audience. In this context, one would expect the regime to use its extensive networks of bots and paid accounts to target primarily TG and VK in order to persuade users with anti-war preferences or without a clear preference. However, as we have demonstrated, the regime disproportionately targets the most pro-war platform – OK. These findings allow us to hypothesise that the main strategy of the regime’s online astroturfing is similar to more traditional propaganda. Instead of persuading new recruits, propaganda is much more efficient at affirming beliefs of existing supporters. It provides them with arguments which they can use to minimise their doubts and deflect opponents’ attacks.

These findings also emphasise several issues relevant for discussions of Russian media and propaganda in the context of the war. In recent months, Telegram has received the lion’s share of attention from experts and commentators. Due to the presence of influential pro-war bloggers, it is often seen as the main Russian social media space. However, it is only one pocket of the vast Russian media environment. While 55% of Russian citizens report using Telegram, 42% report using Odnoklassniki, and 65% report using VKontakte. As we have demonstrated, the dynamics of both war-related discussions and the regime’s online astroturfing varies widely across these platforms. In fact, Odnoklassniki has much more pro-war sentiment than Telegram. Thus, if one wants to understand how war propaganda shapes public opinion online, it is important to approach the social media space holistically and take into account all its elements.