China’s viral villages: Digital nationalism and the COVID-19 crisis on online video-sharing platform Bilibili

When the COVID-19 virus broke out in China, foreign observers speculated whether the Chinese leadership was facing its ‘Chernobyl Moment’. China’s leadership, however, defied foreign expectations about its ostensibly floundering legitimacy and instead turned the crisis into a national success story. This article explores the role that digital media played in cementing this success, specifically how various actors mobilized nationalist sentiments and discourses on the online video-sharing platform Bilibili. By focusing on visual discourses, online commentaries, and the affordances of the digital platform, the article analyses the role that ‘hip’ and youthful content played in the authorities’ attempts to guiding online audiences to rally around the flag. The results of these efforts were viral villages of community sentiment that created strong incentives for conformity, and in which the official party line was able to reverberate with pop-culture memes and popular nationalism.


Introduction
In mid-February 2020, roughly 10 million Chinese internet users were watching a music video that was making the rounds on social media. The video shows scenes from the city of Wuhan. Medical personnel in full-body protective gear assist patients in a hospital. Men and women in military fatigues board a plane, carrying equipment and relief goods. Construction works build new facilities. A romantic pop melody plays as a group of young singers, social media influencers, and 'V-Tubers' (virtual video celebrities) take turns hushing inspirational lyrics: 'don't think too much about terrifying things', 'don't lose that smile under that face mask', 'let's join with confidence from north to south, from east to west, everyone can be their own hero'.
Then, spliced in among scenes of epidemicstricken Wuhan, a line-up of celebrities and entertainers addresses viewers directly. Almost 40 actors and athletes, pop singers and idols, cosplayers and voice actors each deliver an inspirational message to their audience. Aged between 20 and 80, all adult generations are represented, though many of the contributors are part of the young generation that China's viral villages: Digital nationalism and the COVID-19 crisis on online video-sharing platform Bilibili was born into the hyper-dynamic society of post-Tiananmen China. Each is an influencer in their own right. Many have followings on the microblogging service Sina Weibo that range in the hundreds of thousands, in some cases in the tens of millions; collectively, these celebrities reach about 150 million social media accounts. Now, they are holding handwritten signs of encouragement into the camera while addressing their fans. Many raise their clenched fist as a sign of strength and solidarity as they shout: 'Go Wuhan, Go China!' (武汉加油, 中 国加油!). The sentiment is reiterated in a constant barrage of user text messages that are superimposed over the video, and the comment sections across platforms likewise show thousands of comments that encourage Wuhan and China to 'hang in there'. Emojis of smiley-faces hugging a heart that reads 'Wuhan' are ubiquitous.
This viral music video, called 'I Sing for You' in English and 'Love Is by Your Side' in Chinese (爱在 你我身旁), illustrates how propaganda efforts have been changing in 21st century China, and especially in the wake of the COVID-19 crisis. When the epidemic broke out in China in early 2020, foreign observers speculated whether the People's Republic of China (PRC) was facing its 'Chernobyl Moment' (Topaloff, 2020). However, the Chinese Communist Party (CCP) defied foreign expectations about its ostensibly flagging legitimacy and turned the crisis into a national success story, using both effective governance measures and its propaganda capacities to strong effect.
In the wake of the Wuhan lockdown, China's internet became awash with patriotic messages. Much of this content had been seeded by the state and party through its propaganda initiatives, and those initiatives also involved private actors who supported the official crisis response. The video 'I Sing for You' is a case in point. The producer behind the video is an official agency of the CCP: the Communist Youth League (共青团, specifically its provincial chapter in Guangdong Province). Official CCP and state-media accounts posted and reposted the video in February 2020, and many produced similar contents of their own. At the height of the COVID-19 pandemic, China's internet had a motivational patriotic song available for every music taste, be it rap, rock, or acoustic pop. Much of this content was targeted at young internet users, using popular platforms such as the short video apps Douyin and Kuaishou or services that support longer formats like Youku and iQiyi. Li (2017) calls such platforms 'contact zones': spaces where users interact with media contents through digital interfaces. The authorities have made efforts to insert themselves into these zones, creating or supporting social media productions that heavily feature the celebrities and aesthetic sensibilities that younger generations have come to expect in an entertainment environment saturated by 'anime, comics, and games' content, or 'ACG'.
One such zone is the platform Bilibili, which caters specifically to China's ACG crowd, and which has seen substantial growth. 1 While much of the video content on the platform can appear frivolous and apolitical at first glance, there seems to be a strong, general undercurrent of nationalism that informs interactions on the platform. As Wu (2020) argues in a recent study: [. . .] nationalist sentiment is extremely prevalent on Bilibili, so much so that it can intrude into completely irrelevant videos, such as food documentaries. It is clear that the government will not miss an opportunity to use this platform as a way to spread its message and exert control. And the government's influence can be seen from the way the website operates to the content featured and audience reactions. (p. 133) It is then indeed no coincidence that the video 'I Sing for You' was first uploaded on Bilibili, and that it features prominent Bilibili 'uploaders', or 'B-Tubers', among its creators and singers: the platform promises a direct channel to Chinese youths, some of whom seem to appreciate patriotic content, provided it also suits their aesthetic and thematic tastes.
This article explores the way in which digital media content mobilized nationalist sentiments on Bilibili in the wake of the Wuhan lockdown. By focusing on visual discourses, online commentaries, and the affordances of this specific digital platform, it asks how different actors engaged with and interacted through 'hip' and youthful content. The article first provides a brief overview of the research that has explored the platform Bilibili and its connection to youth culture, leading into a discussion of how the core platform features that define Bilibili generate a specific community experience among users. Next, the article examines how the authorities leverage such platform features and media aesthetics to rally users around the flag. To explore these activities, I have analysed the top videos associated with three different search terms that related to the COVID-19 outbreak in Wuhan, with each term offering a different type of sentiment (respectively positive, negative, and neutral). As this analysis will show, the affordances of the platform generate distinct discursive spheres that do not necessarily overlap in their discourses, but which instead create what I call 'viral villages': networks of socio-technical engagement that generate recursive feedback loops of community sentiments. My argument is that authoritative actors from the state and party adroitly activate these networks in their attempts to promote specific politics, though possibly with consequences that are more dire than they anticipate, especially at a time of increased 'biopolitical nationalism' (De Kloet et al., 2020) around the world.

Bilibili's digital affordances: the mechanics of imagined community
Since Bilibili's inception in 2009, scholarship has frequently focused on the platform as a case study of digital multi-media communication in contemporary China. Leaving aside the literature on technical and computational issues, the main focus in Chineselanguage research on Bilibili seems to lie with Bilibili's business and advertising model (e.g. Ceng & Zhang, 2015;Ling et al., 2019). There is certainly a strong interest in how the platform connects with youth culture (e.g. Chen, 2014;F. Zhang, 2017), but explicit discussions of communication and politics are fairly rare, and when they do take place, they are often centred around the perceived risks that 'vulgar content' (低俗内容), 'spiritual erosion' (精神侵蚀), and 'addiction' (上瘾) ostensibly pose, for example, to young users of the platform (e.g., Wang, 2020, my translations).
The research focus in such publications is then also primarily on how party and government organizations can resolve these perceived problems and instead use the platform effectively to 'guide public opinion' (舆论引导, see R. Peng, 2020, p. 39). As a consequence, authors often unapologetically applaud actors like the Communist Youth League for setting a good example by 'promoting the construction and development of a burgeoning community of (. . .) 'patriotic youths'' on Bilibili (Z. Lu, 2017, p. 6o, my translation). Such endorsement is itself noteworthy, as it illustrates how authoritative attempts at utilizing the platform for political ends are driven by a moral imperative to assure 'correct' and 'healthy' media usage, especially for youths, and how this imperative is shared among officials and at least some academics. Interestingly, these actors attribute a strong potential to the platform as a vehicle for patriotic community construction, and as we will see, they are not entirely wrong in this assessment. The propaganda activities during the COVID-19 crisis should in any case be understood as an outcome of such ambitions.
English-language research on Bilibili illustrates rather different scholarly concerns. Granted, this research often shares with its Chinese-language counterparts an interest in youth culture and subaltern activities on the platform, but the studies are less driven by moral questions about perceived spiritual pollution and more interested in the socio-psychological, economic, and political implications of the platform's design features and user activities. At the risk of oversimplifying what is becoming a growing and nuanced research agenda that cuts across numerous projects, many of them still ongoing at the time of writing, recent studies have explored the role that the platform plays in community construction (Chen, 2014;Huang, 2020;Wu, 2020), in contestations between young people and the state (Shang, in press), in fandom and subaltern cultures (Chen, 2014;W. Peng, in press;Xu, 2016;Zheng, 2017), and within the evolving dynamics of China's complex digital political economy (Cao, 2019;Wu, 2020).
Of particular concern in this present context are the 'digital affordances' of the platform, that is: the way in which the latent environmental cues of the digital system 'hold possibilities for action' for those who interact with it (Parchoma, 2014, p. 360). What possibilities emerge from socio-technical interactions on Bilibili, and how do design choices on this specific platform invite or imply certain usages rather than others? To answer this question, and to show how Bilibili drives community association through its digital affordances, it is worth unpacking the core design features that govern this platform, specifically its comment functions, its monetization scheme, and its 'gamified' motivational mechanics.
A functionality on Bilibili that sets the platform apart from many other video-sharing and livestreaming services is its comment system, which enables users to have their comments flash across Bilibili videos as superimposed text messages. This 'comments-over video' feature, known as a 'bullet curtain' or as 'barrage subtitling' (弹幕), was originally designed for the Japanese video platform Niconico (Zheng, 2017). Known in Japanese as danmaku, the feature has since seen wide-spread adoption across East Asia (Steinberg, 2019, pp. 200-201). To scholars of the bullet curtain mechanic, the ability to comment on-screen 'introduces new ways for user interaction that allow for instantaneous feedback which reinforces community formation and mutual collaboration' (Xu, 2016, p. 444). Zheng (2017), for instance, has argued that the barrage subtitles allow sub-culture communities to create a sense of shared meaning and belonging. Li (2017) similarly comments on the affective power of such interactions, arguing that The pseudo-simultaneity of the danmaku interface creates a sense of community, because it unites viewers with a collective temporal experience of simultaneous viewing, a sense of 'virtual time' of livenness [. . .]. Because the comments have no id attached, they seem to come from nowhere yet appear right in front of you, creating the feeling of an organic and mythical existence of a highly immersive community that is immediately present and intimately welcoming. (p. 252) There appears to be a consensus that these asynchronous commenting activities generate a feeling of belonging, and that they can be empowering to users. However, as Cao (2019) has argued, such sentiments have a dark side. Participation through the bullet comment feature is deeply entangled with the monetization mechanics that govern this specific platform, which relies heavily on the sort of unpaid, playful labour or 'playbour' that many social media services facilitate in their quest for profit (see the contributions in Scholz, 2013). As Cao (2019) puts it: 'if users can afford the hours or the cash, they gain a sense of superiority' (p. 13). Viewed this way, community sentiments on the platform may well be a form of false consciousness. Cao (2019) goes on to argue that (. . .) there is little evidence of or meaning in this sociality except that the pleasure of texting comes from what is believed to be a communal viewing experience. Each audience member becomes a crowd of their own. (p. 8) Cao's observation is reminiscent of Anderson's (1983Anderson's ( /2006 arguments about what he called 'imagined communities', meaning nations, religions, and other groups such as fan communities or professional associations that are so large, their members will never actually meet each other, and yet the communicative exchanges of ostensible group members create a para-social sense of belonging. This 'imagining' does not have to mean that the community sentiments these groups inspire are not 'real', or that they cannot be powerfully deployed for political gain. To Anderson (1983Anderson ( /2006, it was precisely the psychological propensity of people to imagine likeminded others 'out there' who were engaged in the same kinds of media and communication practices that helped cement in the heads of the members the idea of a political collective. Those media users did not necessarily have any meaningful commonalities, but through their shared engagement in the ritualistic consumption of print-capitalism's products they invested themselves in the idea of a community, which in turn allowed political elites to make appeals about how that community should be constituted and governed. The community-building mechanics of digital platforms hold a similar potential, but now buttressed by interactive, participatory affordances that earlier mass-media formats did not possess, and that promise to make the emotional buy-in of users that much more powerful. One such crucial set of features relates to what Koivisto and Hamari (2019) call 'motivational information systems', or what is more commonly known as 'gamification' (see McGonigal, 2011). As Zhao and Tang (2016) show, gamification on Bilibili relies on four interconnected technical features: an 'experience' point system that awards users for certain activities (logging in, watching and sharing videos, etc.), leader boards that display the most accomplished users, a deep system of crowdfunding mechanics that allows content creators to generate monetary revenue, and platform features that are only available to users who have scored enough experience points to achieve certain 'levels' (Bilibili distinguishes between six such levels). Zhao and Tang (2016) are concerned mainly with how effective these digital design choices are for generating engagement, and they do not provide a critique of such 'gamified' designs (for critical discussions, see Bogost, 2011;O'Donnell, 2014;Woodcock & Johnson, 2018 and -specifically with regards to Bilibili -Cao, 2019). However, the 'motivational affordances' (P. Zhang, 2008) that analysts of Chinese platforms like Bilibili discuss, have wideranging implications for social interactions on those platforms. The socio-psychological dynamics, and especially the emotional investment that they try to inspire, create not just the foundation for 'successful adoption and continual use of danmaku websites' as proponents argue (Zhao & Tang, 2016, p. 468). They also provide the basis for precisely the national community construction on which official actors like the Communist Youth League rely.

The actors behind viral nationalism on Bilibili
What, then, can the cultural productions and online activities surrounding COVID-related videos tell us about digital nationalism and its construction on Bilibili? To answer this question, I examined the features and commenting behaviours of videos that Bilibili listed for three search terms, with each term covering a topic that had a distinct tone. These were: 'Go Wuhan' (武汉加油), which projected positive, inspirational sentiments; 'Fang Fang's Wuhan Diary' (方方武汉日记), a topic that solicited mostly a negative, critical and at times aggressive tone; and 'New Corona Pneumonia' (新冠肺炎), which provided a stream of content about the virus with a primarily informative tone that suggested neutrality.
For each topic, I have focused on the top five videos that Bilibili listed for that search term at the time of this study, which was February through March 2021, so roughly a year after the Wuhan lockdown. While we cannot know in retrospect what dynamics played out in these topical categories at the time of the crisis, the distance in time has the advantage that the content has 'settled': comments, likes, and shares have accumulated, the platform provider and the authorities have had time to censor unwanted content, and what remains is what Bilibili itself considers most representative of the topics in question.
Indeed, to understand how the platform itself prioritizes content, I have selected the top five videos that Bilibili lists under its default sorting category, which it calls 'synthesized ranking' (综合排序). It remains intransparent to users which variables exactly go into this ranking, and how they are then balanced algorithmically, but it seems that the platform aggregates data such as number of views, comments, and other interactions, and possibly also editorial choices about relevance and appropriateness, to generate these listings. To focus the analysis on these synthesized rankings has distinct consequences: selecting a different sorting mechanism yields different results. For instance, ranking the videos of the 'Go Wuhan' hashtag by number of views generates more irreverent and seemingly trivial content than the default. These outcomes of different sorting options would certainly be worth further study in their own right, but I am bracketing them here since my interest indeed lies with precisely the default. As scholarship on digital media environments has demonstrated, default settings are powerful (see e.g. Jürgens & Stark, 2017). They guide how most users experience these environments. Following Bilibili's default sorting mechanism means 'following the medium' (Rogers, 2013, p. 27), which in turn promises to illustrate how the platform itself biases users towards certain priorities.
What does the default selection look like for the hashtag 'Go Wuhan'? I have compiled the main characteristics of the top five videos in Table 1 general terms, it is telling that all results were inspirational music videos, and that they had been uploaded between 23 January and 15 February 2020, so during the first weeks of the Wuhan lockdown. This already illustrates the selection bias of Bilibili's synthesized ranking, which narrows the content to a specific genre, time period, and level of impact. The ranking then also privileges videos with high numbers of views (ranging from ca. 213 thousand to well over 1.6 million), as well as large numbers of comments, though the scale of the commenting clearly varies: on-screen comments range from roughly 1300 to 24,000, while regular comments (or 'replies', in Bilibili's nomenclature) vary between a low 709 to a more substantial 3355. The accounts represented in this list are diverse. They include individual users with significant followings, such as the Hong-Kong-based video editor Eliot Axel, a level-six user with twelve uploads and more than 8000 'fans' (the Bilibili terminology for followers), almost half a million likes across user activities, and nearly nine million views of their content. Another example is the self-proclaimed film fan Shen Datou from Hangzhou, a level-6 user with 48 uploads, 11,000 fans, more than half a million likes, and 7.6 million views. However, the top five videos also include users with much lower impact, such as the user 'onion candy', a level-five account with four uploads, 302 fans, 23,000 likes, and close to half a million views, or the account account called 'celestial maiden raising pigs', a level-four user with the account by-line 'Go Wuhan, Go Hubei, Go China *heart heart heart*', who only ever uploaded this one video, garnering 320 fans, ca. 10,000 likes, and just over 213,000 views at the time of writing. While these numbers are not entirely negligible, they pale in comparison to those that high-profile users generate, and especially to the followings of professional media accounts. This is apparent from the Communist Youth League presence in the top five, here represented by its Guangdong account, which at the time of writing had roughly 1.3 million fans, 51 million likes, and a staggering 861 million views.
The users behind the uploads are then indeed of a quite diverse make, even if the video contents themselves are rather homogeneous. This homogeneity is also visible for the top content in the hashtag 'Fang Fang's Wuhan Diary', at least thematically: all topfive videos are dedicated to discrediting the author Fang Fang and heaping aspersions on her book about life in lock-down Wuhan. The format, length, and production quality of these efforts differ, as do the number of views (ranging from as low as ca. 13,000 to nearly 2.6 million) and the number of comments (between a mere 63 on-screen remarks and 16 replies in the comment section to as many as nearly 84,000 on-screen messages and 18,000 remarks in the comment section). I have compiled the core features for these videos in Table 2.
The user backgrounds reveal a similar split as for the 'Go Wuhan' content: it includes private users of varying impact alongside high-profile professional media accounts. An example of a private account of arguably high impact is 'funny bright', a level-six account that had posted 39 videos, many of them promoting nationalist and anti-liberal sentiments, and which had achieved a following of 23,000 fans, a total of about 466,000 likes, and nearly seven million views. In comparison, the level-four account 'Turn around and you'll be prosperous and wealthy' remained far less impactful; at the time of writing, that specific user had uploaded 144 film clips and academic lectures related to Chinese culture, usually of a culturally essentialist bend, but the account had amassed only a modest number of 3197 fans, 8235 likes, and 488,000 views in total. Similarly, the level-five account 'a mouse' had uploaded 8 short, mostly humorous video clips of widely varying popularity, and ultimately reaching a total number of only 314 fans, 9300 likes, and 343,000 views.
Next to these accounts sit professional content creators such as 'College Daily', a hip online media outlet targeted at Chinese students abroad, and which the New Yorker called a 'post-truth publication' due to its frequent use of false and misleading information, usually with a nationalist message (H. Zhang, 2019). That account had 747 fans, 8.4 million likes, and 93 million views. Even these numbers are eclipsed, however, by 'The Observer', an online news aggregator with a tabloid-like quality that has been producing a steady stream of content since its Bilibili account was launched in 2015, totalling almost 10,000 videos and amassing a total of 6.7 million fans, 201 million likes, and almost 4.5 billion views. Regardless of the differences between these five accounts, the message nevertheless remains remarkably consistent, and the videos even recycle much of the same footage and memes. The content from these five accounts then reveals a thematic bubble, produced by institutionally diverse actors. The final hashtag stands somewhat apart, as it portrays a much wider array of content. I have represented the data for these five videos in Table 3. Granted, we see a similar range in the popularity of, and engagement with, these videos as for the other hashtags, and the actors represented here are likewise diverse, with amateur, professional, and even one official account (CCTV News). Thematically, however, the content covers significantly more ground than the other hashtags. Contributions include a video that explains COVID-19 scientifically to laypersons, a CCTV interview with a man who had contracted and recovered from the disease, a montage of CCTV news greetings throughout the year 2020, an American B-Tuber offering his views on the death of whistle-blower Li Wenliang, and another American interviewing people in the streets of New York about their understanding of the pandemic. While the videos contain predominantly pro-PRC, anti-liberal arguments, the discourse works somewhat differently than in the other hashtags, especially since much of it comes from two foreigners.
The real outlier, however, is the factual video that introduces the disease itself, which had received over 6 million views and had generated more than 31,000 on-screen comments and 21,000 posts in the discussion section. That video comes from the popular science account 'Paper Clip', which was later banned from social media platforms along with other science blogs. Despite its generally apolitical educational tone (Shen, 2021), the account had apparently fallen foul of nationalist sentiments (for an example of accusations, see Global Times, 2021). At the time of this study, the level-6 account had still been active on Bilibili, and it had produced 197 videos, attracting a following of three million fans, and generating 7 million likes and 160 million views.
In general, the private B-Tuber accounts represented in this hashtag were all producing similarly professional, high-impact content, and they all maintained level-six accounts. Next to Paper Clip, this included the account 'dog friends having a chat', which had released 181 videos, followed by 45,000 fans and receiving 1.1 million likes and 33 million views. It also included the two American users, Nathan Rich and Jerry Guo, who are part of a foreign influencer crowd that profits from producing partyconform online content (see Allen & Williams, 2021). Rich, who also goes by 'Hotpot King' and who produced content that arguably catered to China's 'new left', had attracted 1.6 million fans, 5.4 million likes, and 53 million views. Jerry Guo, who frequently interviews Americans in the streets of New York on contemporary affairs, had produced 433 videos and amassed a following of 6.6 million fans, who had provided a total of 32 million likes and 740 million views. While all these accounts were outshone by the official CCTV News presence, which had 5.6 million fans, more than 100 million likes, and over 820 million views, they could certainly hold their own in terms of the followings and interactions they generated.

Engineering viral sentiments through Bilibili user interactions
An open question then is what kind of interactions these hashtags facilitated. Do commentaries on these videos reproduce the discourse that the videos construct? Do they tend towards uniformity in that regard? And are there patterns that extend across the three separate hashtags? To answer these questions, I have explored the bullet-screen commentaries of the videos. 3 As it turns out, the on-screen remarks largely affirm whatever is being said at any given time in the videos. What is more, comments frequently serve as catalysts that inspire copy-cat comments in the same vein almost 'immediately' (recall that these comments only appear immediately after each other on screen but are actually left at different times, i.e., whenever the respective viewers watched the video). This creates precisely the kind of 'barrage' that the platform is famous for, and it leaves the impression that specific phrases are highly contagious. The effect is then indeed homogenizing within each video, and where the hashtag was governed by a This is most visible in the videos in the 'Go Wuhan' category. I have mined the first 1000 bullet comments for each video and examined the most common remarks. 4 Users most commonly flood the videos with the comments 'Go China', followed by the comments 'Go Wuhan', 'We have won', and simply 'Go'. Variations on this theme abound, for instance 'China won' or 'Go Beijing', and some users express their tearfulness (泪目) or post more elaborate patriotic phrases such as 'I will enter China without regrets in this life, and still be a Chinese in the next life' -a phrase popularized by the patriotic webcomic and anime 'Year, Hare, Affair' (那年那兔 那些事儿; on this anime, see Tan, 2018 andGuan &Hu, 2020). 5 Overall, the comments generate a clear consensus on how to appropriately react to, and comment on, this specific crisis.
The hashtag on Fang Fang's diary shows a similar pattern of comment convergence, though here the nature of the comments is decidedly different from those visible in the music videos. In fact, the comments that played such a key role in the music videos, such as 'Go China' and 'Go Wuhan', do not make an appearance here. Instead, comments are overwhelmingly derogatory and derisive, for instance referring to Fang as an 'old feme fatal' (老 妖婆). Viewers frequently react directly to comments made in the video, for example accompanying discussions of Fang Fang's socio-economic background and upbringing with the comment 'family misfortune' (家门不幸, implying that Fang is a disgrace to her family). Maybe most interestingly, comments repeatedly affirm what the speaker says by exclaiming 'exactly this!' (正是在下), creating both a sense of self-righteousness and the impression that this sentiment is shared among viewers. Especially the sequential posting and reposting of this kind of legitimacy signal illustrates how users actively construct an imagined community around the respective issue.
Matters are somewhat more complicated for the neutral hashtag that deals with the epidemic itself. Each video again illustrates a move towards homogeneous user commentary, and some of the comments appear in multiple videos (reaffirming statements like 'exactly this' are one example). However, the remarks are generally not uniform across these videos, and they do not necessarily reproduce to the same nationalism that is on display elsewhere. The video by Paper Clip is particularly revealing in this regard. While user frequently use the term 'Go!' as a marker of solidarity, that solidarity is not tied to 'Wuhan' or 'China'. Comments abound that suggest 'taking care' (保护) more generally, and the main community that users flag is 'humankind' (人类). In fact, the most common comment in the sample was 'The hymn of humankind is the hymn of courage' (人类的赞歌就是勇气的赞 歌), which is a line from the popular Japanese animation series JoJo's Bizarre Adventure. It seems that Paper Clip viewers are plucking cultural content from a cosmopolitan array of 'ACG' resources to express humanist worries, and my qualitative follow-up analysis of the video comments revealed that this sentiment indeed carried through to the very end of the video.

Discussion: digital nationalism and the viral village
Several patterns emerge from this analysis, and they suggest at least some tentative conclusions about three issues: the inner workings of Bilibili communities on themes related to the Wuhan crisis, the workings of the platform itself as it generated incentives for such community dynamics, and the strategies that political actors in state and party adopt to use those dynamics for their own ends. I will go through each issue in turn.

Community signalling
When it comes to engagement with the videos, the analysis yields three important findings. The first is that the use of memes and sub-cultural intertextualities is pervasive across Bilibili, and this is true both for the video content and the user references. In line with the findings from the scholarship discussed earlier, the videos and comments analysed here used highly specific cultural tropes that served users to signal membership in an 'elite' community of people who are 'in the know' about such cultural tropes, specifically the 'ACG' tropes for which Bilibili is famous. More importantly for this present context is how these tropes interacted with nationalist sentiments to slot neatly into the rationale and functioning of the wider national community: both groups are 'imagined', and they generate their attraction and their capacity to inspire a sense of belonging via cultural contents, their circulation, and their ritualistic deployment in social interactions. 'Fandom publics' (W. Zhang, 2016) on Bilibili can then serve as a potential foundation for nationalist mobilization, along the lines that scholarship on China's 'fandom nationalism' has illustrated in various contexts (see the contributions in H. Liu, 2019).
A second observation is that the content of individual videos generates relatively uniform user reactions, and often as a landslide of similar or even identical comments. In some instances, reactions even develop uniformly across a specific theme, for instance, when viewers of videos that criticize certain perceived enemies of the nation repeatedly post the same invectives, slurs, and accusations. Such sharing of tropes, and particularly the impression of overwhelming support for these shared tropes that the barrage system enables, directly feed back into the impression that there is a community of likeminded viewers 'out there'. In fact, the very act of ritualistically typing homogeneous and repetitive communal messages that collapse asynchronous behaviours into a seemingly simultaneous show of community support may be part of what reinforces a sense of unity in the first place; such activities promise to socialize users into the community through what Randall Collins (2004) has called 'interaction ritual chains': a series of activities that 'entrain' sentiments much like circadian rhythms. The barrage system and its affordances in any case create a viral dynamic that generates spirals of self-reinforcing communal signalling. Importantly, users frequently signalled that the nation should serve as the most important reference of such communal belonging. The reaction 'Go China' is a case in point: while the encouragement 'Go Wuhan' was also common, it was ultimately the reference to the national community that became the go-to comment for viewers of patriotic music videos as they wished the people of Wuhan well during the lockdown.
However, and this is the final observation, these viral, homogenizing dynamics that drive community construction in specific videos or themes do not always cross thematic boundaries. Context matters. Commentators on patriotic music videos use different language and reference different memes than users who comment on the pandemic more broadly, or users who comment on those who criticized the relief effort in Wuhan. Even within the top-five videos that Bilibili lists for a specific search term, such as 'New Corona Pneumonia', the discourse can be fractured. This was eminently on display in the scientific explainer-video, where viewers who commented on the pandemic eschewed the potentially available nationalist context in favour of more humanist, cosmopolitan narratives. The intra-group dynamics that the barrage system promotes can then work in very different directions, and it would seem that the outcome hinges to no small extent on the resources that the videos themselves make available for users through their discursive inputs as well as on the early comments that serve as inspiration for later barrages of copied sentiment.

Platform mechanics
This brings me to the next issue area: the power of platform mechanics. Based solely on the video content and comments, it might be tempting to argue that anything goes, on Bilibili. After all, the content is highly diverse, its user base is broad, and its appeal cuts across numerous sub-cultures. Indeed, user interactions can range from liberal-minded to chauvinist (see also S. Lu, 2021). However, the design of Bilibili creates a particular socio-technical system through which all interactions must run. The design features then guide usage into conservative and conformist patterns. This happens because of three interrelated mechanics.
The first of these mechanics is the power of the 'default'. Diversity exists on Bilibili, but it must be actively sought out. In the absence of an overriding user action, the system will display aggregated rankings of content based on intransparent choices that include the company's editorial oversight and its understanding of how to measure relevance. Those choices, in turn, must comply with state guidelines, which the authorities have further consolidated since the crisis. Official missives and notices from 2021 show, for example, how the state is changing the very nature of influencer status, and how it is starting to target the algorithmic rationales of platforms (for translations of such official documents see Creemers, 2021 andCreemers et al., 2021). In this way, the state and enterprises collaborate to assure that platforms push 'correct' content, especially on issues of political concern, and it should then come as no surprise that the top returns for topics related to COVID-19 had only positive things to say about the Chinese response to the pandemic while reserving criticism for foreign countries and domestic detractors.
The second mechanism is the unifying quality of the bullet curtain. Except for the very first commentators, the vast majority of viewers are never only reacting to the video itself; they are reacting to the visible reactions of others. The asynchronous nature of the comment stream leaves the impression that this happens in real time, when actually it happens after the fact. The effect is somewhat akin to that familiar from arboglyphs, that is: from the practice of travellers carving their names or initials into trees and other wooden features, with subsequent travellers then adding their marks in turn, and so on. Since these glyphs are reactions, they frequently reproduce the patterns that are already on display, generating, in this case, a conformist stream of what we might call videoglyphs. Such symbols, even those that may seem banal, are a crucial component of what makes shared community sentiments work (Billig, 2009). The ability of the digital system to enable and encourage collective use of videoglyphs should be understood in that context.
The third and final mechanism that deserves mention here is a set of mechanics that have not been visible in the analysis so far but that are familiar from the literature on the political economy of social media, and specifically on the political economy of Bilibili I discussed above. These are the platform's interlocking monetization mechanics. To maximize income from their videos, B-Tubers must create content that will float to the top of the rankings and will compel users to donate to them. This means that content creators are always navigating the algorithmic and policy environment that the platform has set up while also assuring that their productions appeal to user sensibilities. Clickbait headlines, novelty content, and 'genuine' emotions sell, but at the same time the discourse can never markedly depart from the mainstream. B-Tubers flourish if they balance the expectations of the platform and of their user base; the two American B-Tuber accounts in this study are excellent examples of this, as they both superficially surprise viewers with foreign faces, voices, and views, only to then cater to popular understandings of what especially nationalist viewers are already likely to believe: that the decadent, crazed, and egotistical 'West' constitutes a fundamental 'other' to the ostensibly measured, reasonable, and socially righteous China.

Nurturing the viral village: political communication strategies on Bilibili
Bilibili of course features significantly more design features than just the three mechanics I have discussed here, but it is the interplay between these technical features and the users that creates what I call 'viral villages': networks of socio-technical engagement that generate recursive feedback loops of community sentiments. Creating and nurturing such viral villages serves political actors as a channel for promoting their respective brand of political discourse, which is precisely what organizations like the Communist Youth League are doing on Bilibili. When they succeed, as is arguably the case for the interactions surrounding patriotic content like the 'Go Wuhan' music videos, it is because these actors have understood the dynamics between the digital ecosystem and the people who populate it. Specifically, these actors rely on the following sequence of moves, made possible by the socio-technical setup: 1. They mobilize influencers who benefit financially, and in terms of their celebrity status, from joining propaganda efforts that are framed as righteous. 2. They promote their core cultural productions through the platform's algorithmic and editorial systems.
3. They bait and mobilize users by relying on their 'fear of missing out' (FOMO) in the face of hip, viral content. 4. They stifle relevant alternative information through communicative and technical strategies, such as censoring unwanted voices, drowning out counter-discourses, and relying on digital biases within the platform to marginalize oppositional actors, and 5. They count on the algorithmically enabled interactions between users to recursively amplify the 'viral' message within a specific channel.
This is a powerful set of strategies that promises to make efficient use of the dynamics between people, technology, and politics to create communities of shared viral activity. Even if some of the online activities that these strategic moves rely on may seem to be token at best (posting 'Go China', using patriotic emoticons, etc.), research on so-called 'clicktivism' or 'slacktivism' has repeatedly shown that digital virtue signalling among online users generates awareness for the issue, legitimates and even normalizes the behaviour, and ultimately has the potential to desensitize such users to the option of more extensive political actions (see Chapman & Coffé, 2016;Halupka, 2018;Kristofferson et al., 2014 andGeorge &Leidner, 2019). Seen in this light, the activities of, for example, the Communist Youth League may very well create more nationalist political subjects.
And yet, this strategy is not without flaws. Subcultures are only ever 'hip' and 'cool' as long as they remain sub-cultures; once they are mainstreamed, they lose their appeal. In many ways, constructing mainstream CCP narratives on the basis of sub-cultural tropes is a contradiction in terms, and it may well prove to be a fool's errand, at least if the goal is to reach the young generation that builds its identity around such tropes. Follow-up research would have to explore who actually engages with officially-promoted content on Bilibili, and who views themselves as part of the CCP-led patriotic community on that platform. It may not be the ACG crowd that the authorities think they reach in this fashion.
Maybe most worryingly, since the viral villages that official actors are helping construct on social media platforms are governed by recursive loops of user engagement, there is an ever-present risk that these feedback loops will spiral out of the control of those who initiated them. Such spaces function as amplification chambers for whatever content is injected into them. Especially where the input premise is that such a closed 'village' community needs to be protected from enemies within and without, the output may turn out to be supremely toxic.
As There is of course much to criticize about McLuhan. 6 Nevertheless, his often-astute observations about the ways in which communication technologies shape human communities remain relevant in a time of near-ubiquitous digital communication, and they provide a warning to political actors who try to activate and manipulate human tribalism in advanced socio-economic systems to serve their political projects, whether in China or elsewhere.

Conclusion
As this article has shown, the Chinese video-sharing platform Bilibili became a prominent vehicle for mainstream political actors to promote their brand of nationalism during the COVID-19 crisis. Especially the CCP capitalized on the social and commercial dynamics of Bilibili to sell its particular brand of nationalism and inspire solidarity among users, hoping in particular to reach young media consumers of hip 'ACG' culture.
It remains an open question how these efforts were perceived by this audience. Indeed, we know very little about the demographics that drive specific user groups on platforms like Bilibili, where viewers and commenters remain largely anonymous. It is at least plausible that one of the major target audiences of CCP efforts, young fans of ACG culture, are not the main consumers of nationalist propaganda and are instead turning elsewhere for their entertainment. Ethnographic fieldwork would need to unpack which types of audiences engage with which types of content on platforms like Bilibili. However, the actual make-up of the audience may not be that important to the discursive outcomes of socio-technical interactions on digital platforms. If content producers, platform providers, and regulators all imagine a specific target audience, then their communicative choices will be informed by that imagination, and this may create an audience of its own, effectively locking everyone involved into self-fulfilling prophecies that have little to do with actual audience preferences.
Whatever the actual nature of the viewership on Bilibili, it is clear from this study that the interlocking socio-technical systems on which authoritative actors relied in their propaganda efforts generated viral content on that platform that resonated through dedicated channels, creating what I have called viral villages: networks of socio-technical engagement that generate recursive feedback loops of community sentiments. Granted, some of these spaces were governed by cosmopolitan, humanist narratives rather than by toxic nationalism, reflecting the inputs that content creators had injected in such instances. Most discourses, however, fell back on nationalism to generate community sentiments, and those sentiments seem to have interacted fruitfully with the sub-cultural communal signalling prominent among many platform users.
I have suggested that official actors like the Communist Youth League are fairly effective when it comes to generating viral, nationalist content, but I have also made the case that the strategists behind this approach at political mobilization are underestimating the toxic potential of imagined communities such as nations. While it is difficult to draw a direct causal link between, for example, specific nationalist videos on Bilibili and the often aggressively chauvinistic attacks that are visible elsewhere on the platform (see S. Lu, 2021), or more broadly throughout digital China, we should acknowledge that the constant flagging of the national community creates a discursive and symbolic backdrop before which certain users feel empowered to take actions that they themselves interpret as 'patriotic'. Follow-up studies should then explore the dynamics of digital China more broadly, for instance through cross-platform comparisons of video-sharing sites like Douyin and Youku, but also in comparison with microblogging and chat-app platforms like Sina Weibo and Tencent's Weixin, to empirically assess whether and how China's socio-technical platform ecology generates a shared 'symbolic universe' (Berger & Luckmann, 1966 Digital nationalism in China is arguably a peculiar animal, seeing as it emerges from a complex home-grown digital ecosystem governed by a oneparty political apparatus (see also Schneider, 2018, ch.8), but despite the specifics of this context, such cyber-nationalism is not unique. At the time of writing, humanity was witnessing a global moment of aggressive 'biopolitical nationalism' (De Kloet et al., 2020) that has been driven by populists who manipulate socio-economic grievances, perceived or real, and stoke them through clever manipulations of digital systems. To interpret aggressive nationalism in China as merely a function of the PRC's one-party system and its attempts to legitimate itself means misinterpreting the politics of our times. Those politics are deeply entwined with technology, and technology is never neutral (Kranzberg, 1995). Its design creates affordances that encourage specific political behaviours, and those behaviours should not be encouraged to fall into nationalist patterns. It is then up to those who design the interfaces of our social world, as well as those who regulate those design activities, to examine the Chinese experience and carefully consider what mechanics generate digital nationalism, and which mechanisms can be deployed to defang and dismantle it.