The Political Economy of Live Streaming in China: Exploring Stakeholder Interactions and Platform Regulation

Yuanbo Qiu

A thesis submitted in fulfilment of the requirements for the degree of

Doctor of Philosophy

Faculty of Arts and Social Sciences

The University of Sydney

2021

i

Statement of originality

This is to certify that to the best of my knowledge, the content of this thesis is my own work. This thesis has not been submitted for any degree or other purposes.

I certify that the intellectual content of this thesis is the product of my own work and that all the assistance received in preparing this thesis and sources have been acknowledged.

Yuanbo Qiu

February 2021

ii

Abstract

Watching online videos provides a major form of entertainment for internet users today, and in recent years live streaming platforms such as Twitch and Douyu, which allow individual users to live stream their real-time activities, have grown significantly in popularity. Many social media websites, including

YouTube and Facebook, have also embedded live streaming services into their sites. However, the problem of harmful content, data misuse, labour exploitation and the burgeoning political and economic power of platform companies is becoming increasingly serious in the context of live streaming. Live streaming platforms have enabled synchronous interactions between streamers and viewers, and these practices are structured by platform companies in pursuit of commercial goals. Arising out of these interactions, we are seeing unpredictable streaming content, high-intensity user engagement and new forms of data ownership that pose challenges to existing regulation policies. Drawing on the frameworks of critical political economy of communication and platform regulation studies by Winseck, Gillespie,

Gorwa, Van Dijck, Flew and others, this thesis examines regulation by platform and regulation of platform in a Chinese context. The thesis asks how Chinese live streaming platforms control and mould communicative activities through their proprietary algorithms, contracts and policies, and examines how multiple stakeholders work together to regulate the platforms in order to defend and maintain the public interest. To examine these issues, the historical transformation of live streaming in China, the role of the state and live streaming regulation and the complexity of the platformisation have all been explored in detail through the lens of the selected theoretical frameworks. The thesis concludes with some reflections on contemporary platform studies and recommends that future researchers take a more holistic view that explains the complexity of the cultural and political structures underpinning the platform ecosystem. It points out that exploring platform regulation in the context of Chinese live streaming can contribute to the existing body of platform theory and, in that process, ‘de-Westernise’ existing platform regulatory studies.

iii

iv

Contents

Statement of originality ...... ii Abstract ...... iii List of figures ...... x Acknowledgements ...... xi Chapter 1 Introduction ...... 1 Live streaming and its development ...... 1 Literature review and theoretical frameworks ...... 4 Live streaming and interactivity ...... 4 Political economy of communication and platform regulation ...... 6 Research questions of this thesis ...... 11 Conceptualising governance and regulation ...... 13 Definition of key terms ...... 15 The concept of individual live streaming in this thesis ...... 15 Existing studies and their definitions of live streaming ...... 16 The meaning of ‘live streaming’ in this research ...... 16 Streaming terminology in contemporary discourse ...... 18 Official event live streaming ...... 18 Streaming video sites ...... 19 Channels on live streaming platforms that do not provide live streaming content ...... 19 Other terms related to live streaming ...... 20 Streaming channels/rooms ...... 20 Tipping/gifting/donation ...... 20 Game streaming ...... 20 Live shows ...... 21 Partnered streamers ...... 21 Bullet-screens ...... 21 The organisation of this thesis ...... 24

v

Chapter 2 Platformisation and the historical development of live streaming in China ...... 27 Platformisation in China ...... 28 The role of the Chinese state in platformisation ...... 29 Empowerment and exploitation in a different social context ...... 31 The development of live streaming in the United States ...... 33 Lifecast and Twitch.tv...... 33 The practice of interactivity: Social TV and live streaming ...... 36 Sociability of watching ...... 37 Co-viewing and socialisation on a single platform ...... 38 The popularity of live show performances: Different trajectories in the development of Chinese live streaming ...... 42 From voice chat to live video entertainment ...... 42 The socialising of watching in China: From pseudo-synchronous interaction on Acfun to synchronous interaction on Douyu ...... 43 The characteristics of Chinese live streaming industries ...... 45 The quasi-monopoly of Twitch in the West and the oligopolistic market in China ...... 46 Different viewer demographics ...... 47 Genres of streaming content: Enthusiasm for erotic performances ...... 49 From advertising to tipping: The differences of core business and sources of revenue ...... 52 Twitch: Relying on subscription and ads ...... 52 Chinese platforms: Profiting from paying viewers rather than ads ...... 53 Modes of partnership and employment: A unique talent agency model ...... 58 Conclusion ...... 62 Chapter 3 The role of the state and live streaming regulation in China ...... 66 Regulating digital media in a global context...... 67 The context of live streaming regulation in China ...... 69 The focus of Chinese internet regulation ...... 69 The dilemma for platform regulation ...... 71 Platform regulation in the Chinese context ...... 73 Live interaction and the imperative for new regulatory policies ...... 74 Live interactions ...... 75 vi

Community Building and Collective Action...... 77 Regulation of live streaming in the United States: Twitch and Facebook Live ...... 78 Twitch ...... 78 Platform regulation of Twitch for streamers and viewers ...... 78 Channel regulation from the communities ...... 80 Facebook Live ...... 82 Contemporary issues of live streaming regulation on Western platforms ...... 84 Challenges to regulation by the state in China ...... 85 An overwhelming workload for real time monitoring? ...... 85 Conflict between government agencies ...... 87 Outdated legislation, vague policy mandates and inefficient penalties ...... 91 The evolving regulatory strategies ...... 93 A central strategy: Forcing the platforms to take up self-regulation ...... 96 Stakeholder interviews ...... 96 First-tier self-regulation: Self-regulation through industry associations ...... 99 Second-tier self-regulation: Self-regulation from the live streaming platforms ...... 100 Third Tier self-regulation: Peer regulation from viewer communities ...... 107 The complementary measures of the state: More subtle forms of control ...... 110 The bottom line: News content control ...... 110 Admission to the market: Licensing ...... 112 Becoming an inspector: The changing role of the government ...... 114 Discussion: Combining self-regulation with direct government control? ...... 116 Issues arising from complementary government control ...... 118 Hurdles for self-regulation ...... 122 Towards a more nuanced analysis ...... 127 Chapter 4 The complexity of platformisation ...... 129 Filling the gap: Live streaming for the empowerment and gratification of marginalised social groups ...... 130 Live streaming as a new production tool for the grassroots ...... 130 Gratification for the non-celebrity streamers ...... 134

vii

The joy of streaming and interactivity ...... 134 A decent job with considerable income for full-time streamers ...... 136 ‘Down to earth’: Streamers and the social needs of viewers ...... 140 The state as a key stakeholder for poverty reduction ...... 145 From suspicion to support: ‘Peppa Pig’, e-commerce streaming and a shift in attitude ...... 146 Summary ...... 148 Rethinking the audience commodity: An audience-led narrative and the unconscious ‘broadcasters’ .... 150 Audience commodity theory and the change of monetisation channel ...... 150 The enclosure of virtual currency in China ...... 154 Two paradigms of commodification in live streaming ...... 156 A content-based paradigm: Live streaming as enhanced street performance; donation as reward ...... 156 A platform-based paradigm ...... 159 Summary: The commodity audience and donation inflation ...... 168 Beyond precarity: The inherent dilemma for professional ‘amateur streamers’ ...... 170 Sponsorship and e-commerce live streaming as unique forms of advertising ...... 172 Interventions into production spaces ...... 175 The professionalisation of leisure activities and homogenisation ...... 175 Standardised and opaque evaluation ...... 179 Unfair contracts and power asymmetries ...... 181 Unfair contractual obligations ...... 182 Ever-changing agreement ...... 183 Platformisation and the concentration of ownership ...... 186 From the war between the thousands to oligopoly...... 187 Oligopoly, platformisation and labour ...... 190 Conclusion ...... 195 Chapter 5 Conclusion ...... 198 Supervised cultural production: Paternalism over ‘lowbrow’ endeavours ...... 198 Anti-vulgarity campaigns and stereotypes ...... 198 Promoting goodness with positive energy ...... 200

viii

A paternalistic, principles-based model of regulation ...... 202 Calling the community: An absence of protagonists for self-management ...... 204 Community regulation in Chinese live streaming ...... 205 Co-regulation models ...... 207 Trusted flaggers and the Facebook Oversight Board (FOB) ...... 208 The Discipline Committee of Bilibili: Signalling an alternative approach? ...... 209 Building up a consistent, cooperative regulatory system ...... 213 Engaging with the world: International and cross-cultural negotiation via live streaming governance ...... 216 An anti-monopoly campaign on platform economy: The real beginning for platform governance? ...... 216 Digital sovereignty and live streaming in a global context ...... 218 Conclusion and directions for future research ...... 221 References ...... 226

ix

List of figures

1 Waste of Time ...... 3 2 Example of bullet-screen (white text) ...... 22 3 The lifecast of Justin Kan ...... 35 4 Interface for the live interaction on Twitch ...... 41 5 Interface for the live interaction on Douyu ...... 44 6 The categories of live streaming content ...... 49 7 Percentage of gaming streamers not paid for content they create in 2019 ...... 50 8 Advertisement display of Twitch...... 52 9 Cost and revenues of Douyu from 2016 to 2018 ...... 54 10 Cost of revenues of Douyu from 2016 to 2018 ...... 55 11 Cost of revenues of Douyu from 2017 to 2019 ...... 55 12 Cost and revenues of Douyu from 2017 to 2019 ...... 56 13 Paying users of Douyu ...... 56 14 Key functions of streamer agencies ...... 61 15 Transition of state focus on media regulation ...... 71 16 Reciprocal practices of live streaming ...... 76 17 Functionality of administrations on live streaming ...... 88 18 Evolving national policies on live streaming in China ...... 95 19 Registration process of Douyu...... 105 20 Licences and related government agencies ...... 113 21 Actual appearance and filtered appearance ...... 139 22 Flows of virtual gifts ...... 155 23 Frontpage of Douyu ...... 161 24 Frontpage of Twitch ...... 161

x

Acknowledgements

There are many people I would like to thank for their contributions. This thesis could not have been completed without their kind support, and I extend my sincere thanks to all of them.

First and foremost, I would like to express my utmost gratitude to my supervisor, Professor

Tim Dwyer, who was always accommodating and patient throughout the process. This thesis would not have been possible without the support and guidance I received from him. Appreciation also goes to my supervisor, Jonathon Hutchinson, for his support and helpful advice regarding my research.

I would also like to express my deep appreciation to Joyce Nip and Chris Chesher, who gave me valuable opportunities for teaching and learning during these years. I also appreciate the help I received from other staff and colleagues in the Department of Media and Communications and the Faculty of Arts and Social Sciences, including but not limited to Alana Mann, Benedetta

Brevini, Fiona Martin, Gerard Goggin, Mark Johnson and Penny O’Donnell. And I would like to thank Susan Jarvis for her final editing of my manuscript.

Special thanks to scholars and friends who provided precious feedback about this thesis and constant and considerate encouragement to me: Xiaojie Cao, Yunjuan Luo, Xuanzi Xu, Huang

Huang, Diwei Zhang, Yuming He, George Qiao, Kyle Moore and many others; I cannot do justice to all of them here.

Last but certainly not least, I would like to thank my family, especially my parents and grandparents, for everything they have given me over the years. And I would like to thank Ying

Lin: without her love, time and attention, I would never have achieved so much.

xi

xii

Chapter 1

Introduction

My dream is to become a professional dancer, and I started learning dance in my

childhood. However, I was too short to be a professional dancer when I was

growing up. I was pretty sad at the beginning, then I found live streaming as an

alternative. I can fulfil my dream on live streaming platforms – to be a dancer,

showing myself and earn the attention and admiration of the others, without

worrying about the defect of my height … We have a good relationship online, and

we also play popular computer games together after streaming hours.

— Interviewee S2, a young woman who pursues her dream of becoming a dancer on a live streaming platform

Live streaming and its development

Watching online videos is major form of entertainment for people these days, and in recent years the popularity of new forms of video platforms – user-generated live streaming platforms such as

Facebook Live, Twitch and Periscope (the Twitter Live streaming platform) – has grown significantly. These platforms allow individual users to live stream their real-time activities for audiences. Launched in 2011, Twitch became the first successful live streaming platform to focus specifically on game streaming – it allows users to broadcast their gameplay, as well as providing large game tournaments for professional players. The popularity of watching competitive games is a major impetus for the development of live streaming (Kaytoue et al., 2012; Nascimento et al.,

1

2014; Smith et al., 2013). For game players, the gratification of watching e-sports games

(competitive games) is similar to watching traditional sports, and many game players claim that they are more interested in watching others’ live-streaming than playing themselves (Kaytoue et al., 2012). As the first successful game streaming platform internationally, Twitch was once ranked

4th in Peak US Internet Traffic, just behind Netflix, Google and Apple (Nascimento et al., 2014); it was acquired by Amazon in 2014. Following Twitch, Facebook, YouTube and Twitter all started their live streaming projects from 2015 (van Brandenburg et al., 2016). Unlike Twitch, Facebook

Live, YouTube Live, Instagram Live and Periscope (owned by Twitter) were launched to broadcast a variety of content genres, rather than just games. These include indoor live shows (cam show), outdoor live shows (exploration, hunting, etc.), sports streaming and educational streaming.

2

Figure 1: Waste of Time1

In China, streaming platforms such as Douyu and YY were founded for game-playing following the success of Twitch. While game streaming continues to constitute a large proportion of streaming content, in China – unlike Western streaming platforms – these live shows have always dominated streaming content. Developed from live video chat rooms, YY and 9158.COM are regarded as the first two live streaming platforms in China; from their outset in 2012, both were dedicated to live cam shows (Xie, 2016). In a live show, streamers usually sit in front of a computer with a camera and talk to the “comments” post by audiences. Singing and talk shows were two typical categories of performance of streamers in this early phase. After two years of development, the broadcast content of live streaming in China diverged after Douyu TV was

1 Gredor, LoadingArtist.com, 11 September 2015, https://loadingartist.com/comic/waste-of-time

3

launched in 2014, allowing individuals to freely broadcast anything without restrictions. Moreover, the development of digital recording, mobile internet bandwidth and transcoding technologies now makes it possible for individuals to stream via mobile devices; this allows everyone with a phone to live stream from anywhere at any time (Harcha & Pereira, 2014; Pires & Simon, 2014; Zhang

& Liu, 2015). According to one report, there are half billion live stream users in China in 2019, and this number was predicted to grow dramatically in 2020–21 during the COVID-19 pandemic

(iiMedia, 2020). While most industries shrunk in the pandemic, e-commerce streaming and education streaming benefited from the C19 shutdown and quarantine (Z. Li, 2020; M. Shen, 2020;

Xinhua, 2020; Yuan, 2020).

Literature review and theoretical frameworks

Live streaming and interactivity

The prosperity of live streaming does not just reflect the increasing interest of audiences in watching live video, but also reflects their call for live interaction. Live interactions in live streaming include discussing content with other audience members in the chat zone, posting comments to streamers and tipping favourite streamers. Watching live streaming is thus not simply about spectating, but entails a sense of ‘social’ interaction, which is empowered by the evolution of interactive technologies (Baruah, 2012). Nascimento (2014) and Gehl (2014) further indicate that live interactivity makes live streaming distinct from other forms of social media entertainment because it enables synchronous streamer–spectator and spectator–spectator interactions, rather than merely asynchronous interactions with audiences.

Existing studies show that viewers gain new forms of pleasure through live interactions.

Viewers can interact with streamers by texting or posting emojis, and streamers can respond to

4

viewers by facial expression, conversation or actions (Nascimento, 2014). These live interactions can produce feelings of ease, satisfaction, confidence and connectedness for the viewers

(Cunningham et al., 2019; Guarriello, 2019; Johnson & Woodcock, 2019a, 2019b), and thus make watching live streaming addictive. A wide range of research focuses on the evolution of interactive technologies, examining how broadcasting and recording technologies facilitate live interactions among viewers, thus changing the nature of spectatorship and the media industries (Baruah, 2012;

Consalvo, 2017; Johnson & Woodcock, 2019a; Kaytoue et al., 2012; Martini, 2018a; Shamma et al., 2012; Summa, 2011; Weisz et al., 2007; Zhang & Liu, 2015). Understanding these changes through the concept of affordance can avoid technological determinism and an oversimplification of social changes. Norman (1999, p. 39) defines affordance as perceived possibilities of actions.

Deriving from this connotation, Wellman (2001) notes that the new technology has social affordances – it provides the possibilities to change social relationships and social structure. In this sense, live streaming, which is characterised by live interaction, provides possibilities to change social relationships and social structures. How these possibilities can be realised is still in the hands of human actors and determined by other social factors. Inspired by these studies, I argue that any research on live streaming should put live interaction in a central position when conducting further analysis, as live interaction provides new social affordances. Many recent studies on live streaming have considered gender politics, the fandom economy and labour issues, but most do not provide a nuanced analysis or contextualisation of live interactions before offering empirical findings and theoretical discussions. As a consequence, such research tends to overlook the distinctness of live streaming and render the analyses repetitive, as similar findings on other forms of social media and video websites are now abundant. In short, contextualising live interactivity should be a

5

primary step for live streaming research, as it can provide deeper insights and avoid putting ‘old wine in new bottles’ for different forms of social media.

Political economy of communication and platform regulation

Theories related to the critical political economy of communication (CPEC) and platform regulation (PR) constitute the main theoretical framework of this thesis. This section aims to link these two theoretical strands to identify how they can help us to better understand the political economy and regulation of individual live streaming platforms in China. I begin with a short overview, which is followed by a synthesis of these theories to identify the theoretical and practical significance of this study. Detailed discussions of CPEC, platformisation and platform regulation will be associated with empirical findings in the next few chapters.

CPEC, deriving in large measure from Marxist political economy, builds on claims that the ruling classes seek to control not only the material production in society, but also spiritual production. This idea was developed by theorists such as Dallas Smythe, Herbert Schiller, Graham

Murdock and Harold Innis to analyse distribution, exchange, power and control in the processes of communication production and consumption (Winseck, 2017). It can usefully be applied to both physical and non-physical production modes, such as ‘material’ ICT infrastructures or ‘immaterial’ media content. Mosco (2009, p. 2), in a review of these theoretical frameworks, concludes that

‘Political economy is the study of the social relations, particularly the power relations that mutually constitute the production, distribution and consumption of resources, including communication resources.’ The main objective of CPEC, as Mosco notes, is to reveal the unequal power relations and advance democratic communication by building public systems of communication (Mosco,

2009, p. 79).

6

One starting point of the studies of CPEC is the commercialisation of mass media and the audience commodity. In the 1970s, as proposed by Dallas Smythe (1977), the audience-commodity concept saw the audience of media as a commodity that had been commodified by media and then sold to advertisers, with the mass media being the free lunch to attract the commodity/audience.

Based on this theory, audiences can be commodified and have economic value for the advertisers and media companies since the audience has the ability to purchase the advertised product and to be influenced by advertisements (Schiller, 1989). In this sense, audience commodity theory offers a critical perspective on the economic model of media based on the exploitation of audience labour.

The audience commodity theory has been developed further in the contexts of social media.

Jenkins (Jenkins & Deuze, 2008, p. 331) argues that users can participate in content production and distribution; thus participatory culture, ‘which fans and other consumers are invited to actively participate in the creation and circulation of new content’, characterises the active role of social media users. However, social media and the active role of users have not fundamentally changed the exploitive relationship. The concepts of ‘produsage’ (Bruns, 2008) and ‘prosumption’ (Toffler

& Alvin, 1980) are often used to describe the dual role of social media users as both producers and consumers. Yet, while the users participate in content production and distribution, they neither own their products nor have the right to govern and control the structures that affect the productions

(Fuchs, 2013, p. 57). Moreover, users’ consciousness and behaviours have been shaped during the content production and consumption; social media thereby amplify the commodification of media users (Mosco, 2009, pp. 134–37). Drawn from this perspective, the concepts of digital labour, playbour and affective labour have all been used to refer to the spontaneous, pleasurable and free labour that is created by the activities of internet users for social media (Fuchs, 2014a, pp. 34–36;

2014b; Hardt & Negri, 2000; Kücklich, 2005; Terranova, 2000). In other words, media companies

7

gain profit through advertising but do not pay for the labour of the content providers: users, and thus the different forms of labour mentioned above, are exploited by social media, thereby constituting surplus value extraction.

Targeted advertising, which comes with big data analytics and extraction of user information, prevails in today’s social media, and has become an advanced way of commodifying social media users. Marx notes that there are two forms of surplus value: absolute surplus value and relative surplus value. Surplus value produced by prolonging the working day is absolute surplus value, while that produced by increasing efficiency and productivity is relative surplus value (Marx, 1867, p. 431). Targeted advertising inevitably intensifies surplus value production partly because ‘the advertising company’s wage labourers and partly and Internet users, whose user-generated data and transaction date are utilised, produce the profit generated from these advertisements’ (Fuchs, 2013, p. 106). More effective ads are produced in the same time period, and thus more relative surplus value is generated – the effectivity is attributed to more unpaid labour of a company’s employees and users who generated transaction data (Fuchs, 2013, p. 106).

In addition to audience commodity theory and the detailed process of labour exploitation, the second focus of CPEC is the political-economic system underpinning this exploitation. This strand of studies is concerned with inequality and control of the social relationships from a macro perspective, examining the expansion in power wielded by giant media corporations and the subsequent unfair power relationship between different actors including industrial stakeholders,

NGOs and governmental agencies (Mosco, 2009; Winseck, 2017, p. 97). The imbalanced power relationships are largely caused by the privatisation of the material infrastructure of these giant media corporations (Winseck, 2017, p. 97). Companies such as Google, Amazon and Baidu are eager to build, incorporate and control the information infrastructure to shape communicative

8

practices of users of their platforms, which opened up a new research agenda around the platform companies from the early 2010s (Gillespie, 2014; Langlois & Elmer, 2013; Plantin et al., 2018).

‘Platform’ is defined by Poell (2019, p. 3) as ‘programmable digital infrastructures that facilitate and shape personalised interactions among end-users and complementors, organised through the systematic collection, algorithmic processing, monetisation, and circulation of data’. Platform studies is a relatively a new field of communication study that coincides with the rise of the digital platforms themselves, the evolution of information and communication technology and the changing role of active users on digital platforms (Benkler, 2006; Fuchs, 2013; Poell et al., 2019). Platform studies aims to explore how platforms, as intermediaries, bring together different human and non- human actors, and transform culture production, distribution and circulation with the logic of the platform economy to build a new ecosystem (Poell et al., 2019; Van Dijck et al., 2018). According to

Poell et al. (2019), there are four different research perspectives for platform study: software studies, business studies, critical political economy and cultural studies. The CPEC perspective of study on platforms is becoming increasingly significant, given the intensification of platform power on a global scale. This exponential growth of the platform economy has led to a concentration of ownership. The domination of platform companies such as Facebook, Apple, Amazon, Google and Netflix does not just lead to an accumulation of economic power; they also exert significant social influence (Srnicek,

2017; Van Dijck et al., 2018). These platforms and related entities own the means of communication

(Winseck, 2020, p. 263), and have both the ability and willingness to redefine power relations between stakeholders and influence their communicative activities (Poell et al., 2019; Srnicek, 2017) through enhancing, formatting, encoding and diagnosing communication (Langlois & Elmer, 2013, p. 3). In this sense, the public interest is at stake when platforms are controlled by profit-seeking commercial entities and the means of communication production become proprietary; these developments are

9

calling for regulatory studies to explore the power dynamics and to assess how best to keep the platforms on the right track (Gillespie, 2018). Drawing from this latter point, while infrastructuralisation of platforms and market reorganisation are the two main dimensions of platform study, platform governance has become an emerging research dimension from a political economic perspective (Poell et al., 2019, pp. 6–8).

Platform governance studies identify the political economy of how cultural production, distribution and circulation are shaped through platforms (Nieborg & Poell, 2018). Through algorithmic sorting and display of the user interfaces, platforms can privilege flows of data, thereby making certain types of content prominent and thus shaping the communicative practices of platform users (Bucher, 2018; Chun, 2011). Moreover, terms of service (TOS) and user guidelines can justify or even legalise the platforms’ extraction of user data and control of user behaviour (Gorwa, 2019b;

Langlois, 2013). In short, user behaviour is governed by platform companies, and these forms of governance are usually not fair, transparent, ethical or even legal (Gorwa, 2019b, p. 859). Therefore, the first question of any platform governance framework requires understanding this mechanism of control by the platforms: ‘How do platforms control and mould communicative activities through algorithms, interface, contracts and policies? (Poell et al., 2019, p. 8).

Second, platforms structure communicative practices, and their power has extended to societal rather than just economic influences; platforms have thus emerged as regulatory entities and powerful gatekeepers for their own commercial benefit (Gorwa, 2019b, p. 858). However, platforms are still subject to official regulators, including state and local laws (Flew, 2016; Gorwa,

2019a). NGOs, including civil society groups and academics, also play vital roles in supervising platform practices (Gorwa, 2019a). Under these circumstances, we can ask, ‘How should platforms be regulated by public sectors in order to defend and maintain the public interest?’ (Flew, 2021).

10

This becomes another key question for platform governance, which is complicated by the contentious discussion in play about the shifting balance between content moderation under public interest obligations and democratic rights of free speech (Cohen, 2017). Therefore, it is important to examine the key aspects of the research on regulating platforms, including the kinds of intervention and communication policy from public sectors that should be applied to regulate platform companies, how we can increase the legitimacy of this regulation, and the regulatory roles of the state and other stakeholders (Gorwa, 2019a, 2019b). In short, platform governance studies are concerned with exploring both governance by platforms and governance of platforms (Gillespie,

2017). Their main goals are to understand the complex governance relationships and public policy challenges and consider how to make these relationships more beneficial and sustainable for the benefit of the majority rather than the few (Gorwa, 2019b, p. 865).

Research questions of this thesis

In terms of studying the politics of live streaming itself, an existing body of literature has examined the democratic potential of live streaming (Bingham, 2020; David et al., 2015; Martini, 2018b; van Ditmarsch, 2013) and gender politics (Ruberg et al., 2019; Witkowski, 2011). Issues of inequality and labour issues in live streaming have also been researched by scholars (Johnson &

Woodcock, 2019a, 2019b; Taylor, 2016; Wang, 2020; Zhang et al., 2019). However, there are only a relatively few studies on the platformisation of individual live streaming in China, and even fewer of these have conducted a nuanced analysis focusing on live interactions and the links with platform regulation in a Chinese context.

Live streaming has its own social affordances and provides possibilities to change social relationships. First, live interactions put viewers and streamers into a mode of continuously and synchronously creating and circulating content, and these practices of live interaction are

11

structured by platform companies and are following their commercial logic from a CPEC perspective. Platform capital plays a major role in utilising or limiting the possibilities afforded by live interactions. This might lead to new models of capital accumulation and new ways of

‘governance by platforms’, which need to be examined. Second, the traditionally free social media, which are mostly advertising-based, are transitioning to obtain revenue from subscriptions and a pay-per-view model rather than advertising (Winseck, 2017, p. 75). But live interaction entails a novel revenue model called donation, which cannot fall easily into any of these categories

(Recktenwald, 2016), calling for a rethinking of audience commodity and labour exploitation.

Third, the increasing popularity of live streaming, accompanied by intensifying issues about live streaming content moderation and freedom of speech, are proving to be of ongoing concern to scholars and policy-makers (He, 2016; Liu, 2016; Ruberg, 2020). In the global context, while globalisation and digital networks seem to weaken the regulatory capacity of the state, it has been noted that it is the state that still holds dramatic power and continuously shapes digital media through regulation, surveillance and control of infrastructure (Flew, 2016; Winseck, 2017, pp.

101–04). Moreover, the role of the state and the implications of the concentration of media markets, as key questions for platform regulation studies (Flew, 2016; Winseck, 2017), become even more complicated in the Chinese context, where the role of the state is deemed overpowered, and global market expansion is rapid. Therefore, exploring the ‘governance of platforms’ in the context of

Chinese live streaming can contribute to the existing platform theory and ‘de-Westernise’ existing platform studies (Davis & Xiao, 2021; Wang & Lobato, 2019, p. 367). In these circumstances, this thesis draws on theories of platformisation to examine the political economy of individual live streaming in China; platform regulation will be the main dimension of this examination. Three main questions will be analysed:

12

1. What is the distinctness of live interactivity in Chinese live streaming? How does it

influence the platformisation of live streaming in China, rendering it different from its

Western counterparts?

2. How do the processes of platformisation in China influence communicative practices in

live streaming and trigger new regulatory issues?

3. What are the roles of different stakeholders in live streaming platform regulation? How do

regulatory activities and legal mandates from the state and their agencies affect commercial

organisations?

Conceptualising governance and regulation

In the field of platform studies on social media platforms, the concept of governance has increasingly been relied on since the 2010s (Langlois & Elmer, 2013). Compared with the concept of regulation, governance is regarded as a more inclusive, broader concept that can better describe the multi-stakeholder dynamics of the contemporary digital media ecosystem (Gorwa, 2019b;

Haggart, 2020; Napoli, 2019, pp. 17–18). However, to say that the regulation is an outdated practice for digital platforms is arbitrary as Terry Flew (2021) notes. Drawing from discussions of these two terminologies by Flew, this thesis uses the term ‘regulation’ instead of ‘governance’ to narrow down the research scope.

There are two main reasons why recent literature prefers governance over regulation. First, it is widely accepted that while regulation only focuses on the regulatory policies from state regulatory agencies, governance recognises the involvement of different stakeholders and the dynamics between them, proposing to study how they achieve alignment with internal departments and external stakeholders (Ginosar, 2013; Langlois, 2013; Napoli, 2019, pp. 17–18). However,

13

this distinction ignores the fact that the complexity of dynamics between multi-stakeholders and networked architecture has already been recognised in new modes of regulation (Flew, 2021). For example, accounts of the ‘new regulatory state’ are characterised by the diffusion of the regulatory activities and responsibility to government departments, industrial agencies and NGOs (Scott,

2017).

Second, governance is regarded as more inclusive than regulation, as it values non-human actors, including algorithms, architectures and artificial intelligence; these have become increasingly influential in digital media ecosystems, and have been treated as essential actors that equate to human actors in studies of platform governance (Gorwa, 2019b; Napoli, 2019, pp. 17–

18). It is undoubtedly the case that non-human actors now have significant implications for the changing modes of regulation, and digital platforms and other entities are increasingly governed by non-human objects such as algorithms and artificial intelligence. However, human actors continue to play a decisive role in the formation of these non-human objects: code, algorithms and architectures are designed to meet particular goals of the human actors and they manifest existing social relationships; (Galloway, 2004), and these objects automatically bring together disciplinary and sovereign power through the production of self-enforcing rules (Chun, 2011, p. 27). Even for the decentralised digital platforms where power is diffused, high-profile individuals in the platform companies, such as CEOs and shareholders, are still in charge of the non-human elements of the operation (Flew, 2021).

Although traditional media regulatory frameworks are insufficient for understanding peer- to-peer, decentralised and so-called ‘neutral’ digital platforms, regulation theory is also evolving and is ‘itself increasingly framed around identifying non-regulatory means of shaping behaviour’

(Flew, 2021). Some regulatory research uses the word ‘regulation’ as well as exploring multi-

14

stakeholder relationships and the role of technology (Winseck, 2020). In this sense, I adopt the concept of platform regulation identified by Flew to narrow down the research scope, focusing instead on how the regulatory activities of public sector agencies impact private sectors entities, and how the related laws, policies and actions derive their legitimacy (Flew, 2021).

Definition of key terms

The concept of individual live streaming in this thesis

The term ‘live streaming’ has been used widely in the media industries and academia. However, it has been deployed by different corporate organisations and individuals in various ways in their discursive practices, and now is becoming ambiguous in the contemporary dialogues. Given the popularity and commercial success of live streaming around the world, while many platforms label themselves as ‘live streaming platforms’, or at least claim that they are providing live streaming services, their meanings of live streaming and the services they provide are often very different.

In this situation, there is an urgent need to give ‘live streaming’ a clearer definition.

Although promotional gimmicks used by businesses are intended to attract viewers by deliberately using the term ‘live streaming’ regardless of these different meanings, in this section

I review such variations in the term from academia to industry, therefore comparing and shedding light on the most often used terminologies related to ‘live streaming’. My intention is to clarify and minimise the chance of the term being misused. It is expected that clarifying these vague concepts about live streaming will help to reduce the possibility of misunderstandings, and thus contribute to the quality of scholarly discourse in this area.

15

Existing studies and their definitions of live streaming

Existing research has tended to give live streaming brief, underdeveloped definitions and has outlined its characteristics in the context of specific studies. There are two key components of live streaming: the activity of streamers that are being streamed, and the chat related to the activity between viewers and streamers (Pellicone, 2016). Anyone can be a streamer and host a channel on a live streaming platform to broadcast activities such as gaming (Johnson & Woodcock, 2019b;

Lu et al., 2018; Zou, 2018). Yet, for game streamers, in addition to showing their screens, sometimes the detailed movements of the streamers during the activities are also streamed

(Witkowski et al., 2016); this may include their facial expressions, movement of their upper body and even their fingers. We can see that live streaming is different from traditional video sites because it not only enables real-time watching, but it also enables real-time and interactive communication between streamers and viewers regarding activities that are generally referred to as video-mediated communication (Recktenwald, 2017; Zou, 2018). Such forms of spectatorship and interaction are not limited by the social circle, and anyone in the world including a random stranger can watch the streaming content of any streamers without restriction (Lu et al., 2018).

The meaning of ‘live streaming’ in this research

Given the diverse literature discussed in the previous section, and the context of this thesis, the term

‘live streaming’ that is being used is an abbreviation of Social Media Live Streaming, which is derived from social TV. It combines social media and video watching to facilitate the interactions between viewers based on the video content they are concurrently watching (Shamma et al., 2012; Weisz et al.,

2007). The practice provides a co-viewing experience and lets the viewers have conversations based on the video content, or even facilitate interactions between viewers and streamers (Martini, 2018a).

For example, when a streamer is live streaming gaming activity, the viewers will be discussing their

16

gaming strategies in the chat room. Indeed, some active viewers can also propose strategic suggestions to the streamer. Streamers, in turn, can follow the suggestions, or reject them and tell the viewers why the strategy may not help in the particular situation. These forms of interactions are essential for us to understand social media live streaming, and I would argue that the primary rationale underpinning the term ‘social live streaming’ is that of socialisation, or reciprocal social interactions between individuals, rather than mere viewership. I would also suggest that this is the key to this variety of live streaming

(Woodcock & Johnson, 2019b, p. 326). My argument is that, in order to satisfy social live streaming, a service must meet at least four requirements:

1. It should be conducted by individuals. The content is mainly created by individual

anchors/streamers of the channel, not by a group of companies with replaceable streamers.

In this way, there is a connection between the specific streamer and the viewer community

(Woodcock & Johnson, 2019b).

2. There should be no limitation for streaming and viewing. Any individuals can freely

broadcast their activities anywhere and at any time with their recording devices, including

PCs and mobile phones. In addition, all the audiences, not just the friends of the streamer

or those are selected, can watch freely on their PC or their portable devices when streamers

are streaming (Lu et al., 2018).

3. There should be real-time broadcasting and viewing. The activities of streamers, including

the recording and broadcasting, and the video watching itself, happen simultaneously

(David et al., 2017, pp. 503–04). Although there might be delays of a few seconds, the

video is not a replayed one.

4. There should be live interactions between streamers and viewers. Members of the audience

can chat freely with each other in the chat room, and there are real-time interactions

17

between the audiences and broadcasters, which are usually co-produced and text-based or

audio-based. In this sense, the viewers have the chance to influence the narrative direction

of the streaming (Smith et al., 2013; Zhu et al., 2017).

Streaming terminology in contemporary discourse

Many commercial organisations or government agencies have used ‘live streaming’ to pursue fashion and improve their influence, but their meanings are different. For clarification, this section identifies several services related to live streaming that have frequently been misused.

Official event live streaming

Traditionally, when people talk about live streaming they are referring to the sports streaming or event streaming that are offered by official broadcasters. This form of live streaming does provide a new form of watching experience as the content is broadcast live in real-time – the activities are both recorded and broadcast simultaneously, which makes it very immersive for the viewers.

However, as noted, the live streaming discussed in this thesis is social live streaming, which also offers various forms of sociality and turns the broadcasting into real-time conversations between the streamers and viewers (Johnson & Woodcock, 2019b, p. 341). Compared with social live streaming, the anchors of official event live streaming are more professional, only providing a one- way communication without instant feedback from and to the audience.

It is worth noting that some major events such as game tournaments or sports contests are using their streaming platform as a channel to broadcast. Similarly, the event holders usually employ several hosts/hostesses and a whole professional production team, making an official ‘TV program’ for live streaming platforms (Smith et al., 2013). These kinds of professional broadcasts

18

usually would not include live interactions as the hosts do not respond to audiences’ comments, and thus they are not the focus of this research.

Streaming video sites

Streaming media/sites is another term that has been widely used, but is sometimes confused with live streaming. The ‘streaming’ here refers to streaming technology that converts video into a continuous flow, and transmits it to the computers or mobile devices of end-users. The transmission and the retrieval of content are processed by a remote server, and then the content is stored in the server instead of the users’ device (Spilker & Colbjørnsen, 2020). The video is therefore ‘streamed’ because it is being transmitted and being played by the viewer simultaneously without downloading the full file. Almost all contemporary video sites, including Netflix,

YouTube and iQiyi, fall into this category. Unlike event live streaming, most videos on streaming video sites are not recording ongoing activities; instead, the video has already been made and then uploaded after the activities have been filmed.

Channels on live streaming platforms that do not provide live streaming content

It should be noted that there are some official channels on live streaming platforms, which are curated and managed by commercial companies or the authorities. These channels have both recorded videos and live content for the publicity; however, the anchors of these channels are interchangeable. Usually, the content on these channels has a fixed narrative and the anchors do not have any interaction with the viewers. In addition, viewers can only interact with other viewers in the chat room when watching and the anchors would not reply to the questions asked by viewers, with no possibility of viewers influencing the narrative of the streaming content. In short, these companies or authorities just use live streaming as a new means to broadcast their videos for publicity purposes. While such channels are being hosted on live streaming platforms, their

19

services can hardly be called live streaming – although this does remain a traditional form of (one- to-many) mass communication.

Other terms related to live streaming

Streaming channels/rooms

Every streamer has a unique web page in the live streaming platform named a channel or room, where they show streaming content and let viewers watch it. It is similar to YouTube channels, which are provided by platform companies but managed by individual streamers for displaying their content and interacting with viewers.

Tipping/gifting/donation

Tipping, gifting and donation are ways for audiences to give money to their favourite streamers when they find the content interesting. In Twitch, there is a “tip” button under the streaming screen: when audiences click this button, they donate their money to streamers via PayPal or any other payment methods. On most Chinese live streaming platforms, viewers use RMB to buy virtual gifts (price ranging from 0.1 to 500 RMB) from the platforms in the first place (Zhu et al., 2017, p. 276), then they can click a ‘gifting’ button to send the gifts to the streamers. After receiving considerable numbers of gifts, streamers can use them to exchange for RMB from the platform.

Tipping, gifting and donation are distinct forms of commercial activity in live streaming, and they account for a substantial amount of revenue for both streamers and platform companies.

Game streaming

Game streaming is a process by which streamers broadcast their in-game activities. Streamers live stream their gameplay sessions and talk to their audiences, who are looking from the in-game perspective of streamers. Usually, streamers record their upper body via webcam and place the

20

video in a small block at the bottom corner of the gaming content, in order that audiences may see streamers’ real-time facial expression and body reaction as well as gameplay (Johnson &

Woodcock, 2019b).

Live shows

Live show is a form of live streaming in which streamers broadcast their real-life activities via indoor webcam or via outdoor mobile phone camera. A live show includes, but is not limited to, talk shows, singing, live chat, teaching, book and movie reviews, outdoor adventure, shopping, hunting and crafts (Lu et al., 2018; Zhou, 2017).

Partnered streamers

Partnered streamers,2 first introduced by Twitch, have signed contracts with platforms and are working as ‘employees’. Partnered streamers can share the revenue from ads, subscriptions and tips, while non-partnered streamers always rely on tips as their primary form of monetization

(Moiz, 2017). When individuals gain popularity, platforms will contact them and ask whether they are willing to be partnered streamers. Sometimes platforms recruit streamers to promote certain content such as new mobile games; these streamers are employed and become partnered streamers.

When streamers have partnerships with platforms, they are paid a monthly salary – but they must fulfil minimum requirements for subscriptions, viewership or streaming hours (Johnson &

Woodcock, 2019b; Pan, 2018).

Bullet-screens

Known as Danmanku functions (Zhu et al., 2017, p. 283), this refers to the ability to synchronously project viewers’ comments onto the main screen as a foreground. Audiences can then customise

2 From https://www.twitch.tv/p/partners/

21

how the comments are presented on screen – usually the comments fly into the screen from the right side and fly out to the left side. These Danmanku affordances are a typical way for viewers to interact with streamers and other viewers. The detail of this activity in live streaming will be discussed further in Chapter 2.

Figure 2: Example of bullet-screen (white text)3

‘Agent company/broker’ is a term used to describe a formal or informal organisation that manages streamers. It describes the person who is dealing with streaming platforms and government. The agency of live streaming is functioning much like a talent spotting agency in the performance industry – detecting promising streamers, training them, supporting them with streaming devices and handling their legal affairs for them.

In China, there is another category of informal agent ‘companies’, known as the family/guild. Families or guilds are usually informal associations to manage streamers, and their functions are similar to formal agent companies but without a legal agreement. A guild is formed

3 From http://v.17173.com/news/10292015/090435668.shtml

22

by an assembly of streamers and managers, and its main objective is to help members to operate their streaming channels and bargain with live streaming companies (Cao, 2019). Sometimes families/guilds help streamers to build up popularity via methods such as cross-channel promotion.

The partnership is a loose arrangement, so there are rarely legal bonds between guilds and streamers. However, streamers generally like to share their income from streaming as a return for the guilds. Liu (2016) has called them a union, but according to her description ‘family/guild’ would be a better term. Most families/guilds are commercial organisations that aim for profitability, whereas unions refer to non-profit workers’ unions that aim to protect the rights and working conditions of workers in a general context. In this thesis, these families or guilds are seen as agents, given their similar functionalities.

Method

This thesis mainly adopts interview and document analysis as the main research methods. For the contextualisation of platformisation and platform regulation in China, relevant documents including existing academic studies, governmental policies, industrial reports, and court documents were collected and analysed in detail. Furthermore, in order to explore the complex relationships between different agencies and the power structure in the Chinese live streaming industries, different stakeholders are interviewed. The interviews was conducted over a period of approximately 20 months, from 2018 to 2020. The interviewees were individual streamers, staff from streaming platforms, talent agent companies, and representatives from government agencies.

These interviews were semi-structured, and they were conducted face-to-face or by telephone.

To answer the research questions of this thesis, the interview questions focus mainly on four aspects: participants’ personal experience in live streaming activities; their relationship with

23

the other stakeholders in the industry; the regulation of live streaming activities; their general understanding of the industry and future development. All the personal information in the thesis has been de-identified to protect the privacy of the interviewees, given the highly sensitive nature of the data. The selection criteria, justification, and other detailed procedures of the interview will be illustrated in chapter three, after introducing different stakeholders and their regulatory roles.

The organisation of this thesis

Chapter 2, ‘Platformisation and the Historical Development of Live Streaming in China’, introduces the concept of platformisation and explores how it is unfolding in China. This is followed by an examination of the historical development of live streaming, both in the Western and the Chinese contexts. By comparing these different development trajectories, the distinctness of Chinese live streaming is identified from five aspects: in terms of market share; viewer demographics; genres of streaming content; revenue model; and labour relationships.

Chapter 3, ‘The Role of the State and Live Streaming Regulation in China’, explores the synchronous interactivities in live streaming, analysing how it is distinctive from previous interactions and how it influences communicative practices; it is thus examined in relation to the existing regulatory framework. By contextualising platform regulation in a global context, this chapter points out that in the Chinese context, an overwhelming workload for real-time monitoring, conflicts between different agencies, and lagging legislation and inefficient enforcement are all challenging the ‘command-and-control’ system. In this complex and dynamic situation of live streaming, with changing social relationships and communication technologies, the regulatory strategies governing live streaming platforms have been updated from 2015 and now reply on platforms’ self-regulation, while the state has changed its role from one of direct regulator to that

24

of an inspector, managing the live streaming platforms in more subtle ways. Although this model has turned out to be efficient for governmental agencies, drawbacks such as excessive and ambiguous self-moderation, conflicts in accountability and risks associated with monopolisation have emerged. These all signal the need for more nuanced analyses of the dynamic between stakeholders.

Chapter 4, ‘The Complexity of Platformisation’, situates live streaming in the unique cultural and political structure of China, and tries to take a more holistic view of the complexity of the platformisation, exploring platform regulation beyond the focus of moderation of inappropriate content. It is argued that three key aspects can help researchers to understand the multi-stakeholder system and the underlying logic. The first is empowerment. Driven by platformisation and governmental facilitation, live streaming has the potential to reduce the digital inequality gap and ameliorate the existing polarisation in China. This empowerment, differing from the traditional understanding of ‘empowerment’ by social media as driven by the platforms, is driven by a content-based paradigm of streaming by both the small streamers and grassroots viewers. These streamers are countering the existing social and economic structures that were formulated by commercial actors, but this practice is still subject to the political power of the state.

The second aspect is commodification. A platform-based paradigm of live streaming implies a new mode of commodification that goes beyond advertising and a ‘pay-per-view’ model. Powerful live streaming platforms control all the actors through distribution, and in that process the platforms turn users into commodities, selling themselves through the practice of ‘donation’. Third, we examine professionalisation. While individual live streaming is inherently amateur-based, platformisation accelerates professionalisation, causing irreconcilable conflicts between content- based and platform-based streaming, authentic and staged performance, informal and formal

25

agreement, precarity and stability. All these characteristics bring out issues such as inequality, precarity, illegal contracting and vicious competition. By revealing the roles of different stakeholders in platformisation and the mechanisms of platform governance, this chapter offers a critical understanding of the platformisation of Chinese live streaming and the origins of regulatory issues. These nuanced analyses point out the possibilities for the regulatory activities of public sector agencies to turn the tables, and thereby to build a more democratic, pluralistic and sustainable system of communication than incorporating platforms.

Chapter 5 follows on from the discussions in the previous chapter, concluding by critically assessing the political economy of live streaming in China and offering detailed suggestions for platform regulation. It recaps by referencing the changing attitudes and regulatory framework of the Chinese government and proposes the construction of a nationally uniform, more consistent regulatory system. It also examines how more informal regulatory arrangements might be the key to shaping the behaviour of commercial companies on content moderation. Finally, this chapter further theorises the implications and meanings of regulatory interventions and links these to the engagement of Chinese platforms in the world, including a discussion of the potential economic, political, social and technological impacts.

26

Chapter 2

Platformisation and the historical development of live

streaming in China

A platform is not simply an application or a website that we discuss in our daily conversations.

Nor is it just a business model, a new technology or a new form of governmental framework. A platform is an intermediary, potentially a part of a new integrated ecosystem, bringing together different actors including customers, advertisers, service providers, producers, suppliers and even physical objects (Srnicek, 2017, p. 25; Van Dijck et al., 2018).

Platforms do not just enter or expand the market, but may fundamentally change it into a new integrated ecosystem (Cohen, 2017). This process, which is similar to the concept of industrialization or electrification, can be called platformisation; it is fundamentally transforming culture production, distribution and circulation (Poell et al., 2019). This process is not only driven by the technological development; the driving forces behind this evolving dynamic process emanate from both non-human and human actors (Srnicek, 2017; Van Dijck, 2020).

Existing studies of platformisation have expressed concerns over this process (Van Dijck et al., 2018). They make the claim that the expansion strategies of platforms rely on three processes: datafication, commodification and selection of users’ data and activity, which lead to new forms of exploitation. The concept of platforms has been connected to ‘global capitalism’, which has been theorized as ‘platform capitalism’ by Srnicek (2017). From this point of view, platforms are not only exploiting human labour, but are also exploiting networks by privatising and modifying control over the global infrastructure. As a result, platforms place human rights and the privacy of

27

users at risk. Moreover, users may often be ‘stuck’, adhering to platforms without an option to opt out (Cohen, 2017). Although the participants can look for alternative intermediaries, many intermediaries are becoming integrated into bigger ecosystems, which become ‘hard-to-avoid necessities’(van Dijck et al., 2019). For example, Facebook is connected to numerous websites and apps, controlling social media transactions on a global scale, and it has evolved into a platform by building complex ecosystems to capture users’ attention and personal data (Plantin et al., 2018).

If many human and non-human actors become part of such an integrated ecosystem, people have little choice but to participate. In short, studies on platformisation explore three dimensions: how infrastructuralisation, market reorganisation and governmental extension, led by digital platforms, shape cultural production, distribution and circulation (Poell et al., 2019).

Platformisation in China

Platformisation is not a uniform process; rather, it follows different trajectories depending on infrastructure, governance and societal practices (de Kloet et al., 2019). China is not a solely a capitalist country and nor does it share the same culture or forms of governance as seen in the

United States and European Union. The original framework of platformisation and platform capitalism may therefore be unsuitable for analysing platformisation in China. Existing studies usually investigate the platformisation process through the lens of ‘datafication’ and

‘commodification’, but platformisation in China is also largely influenced by policies from the government, culture and regional economic development. An obvious manifestation is that Baidu,

Alibaba and Tencent (BAT) have led the process of platformisation in China, while platformisation in other countries has been dominated by the Big Five tech giants of the United States (Van Dijck,

2020). These Chinese companies are so-called digital champions, and the new ecosystem of

28

platforms has been built around them. The Chinese state not only regulates, but also supports the development of digital champions through tax reductions, government loans and barriers to foreign competitors (Hong, 2017a). In this way, the digital champions are mobilised by the Chinese state despite their globalisation, financialization and privatisation (Hong, 2017a, p. 136). Thus, the platformisation is somehow officially mobilised by the Chinese state. From this perspective, the role of the Chinese state in China’s platformisation needs to be examined carefully.

The role of the Chinese state in platformisation

While the US government usually take a neutral position towards technology and platforms, the

Chinese state does not position platforms as neutral intermediaries, and it does not treat algorithms as neutral objects (Z. Zhang, 2020). Instead, the Chinese state focuses on technological development, treating the development of digital technology and entrepreneurship as the solution to China’s social, economic and cultural problems (Keane & Chen, 2019), thereby hoping the digital platforms will facilitate ‘cultural empowerment’ in the ‘Internet Plus’4 agenda (Keane,

2016). In order to achieve this, not only digital champions such as BAT, but also smaller digital companies in China are partners with the cultural and political projects of the state (Wang &

Lobato, 2019). If these companies meet the goal of the state, they can gain support; if their development trajectories do not fit the plan of the state, the state will intervene – the Chinese state is also a facilitator and supporter of the digital economy (Z. Zhang, 2020). For example, Z. Zhang

(2020) has used the case of TikTok to analyse the platformisation of video entertainment in China.

In this case, teenagers were found to be addicted to short video watching on TikTok because of the designed algorithms of the platform, and the Chinese government criticised it and urged the

4 Applying internet and other information technologies to connect the traditional industries in China.

29

platform to make a change. Under pressure from the government, TikTok had to start a

‘infrastructural turn’ to find a balance between the expectations of the state and the business interests of the platform. It then started to cooperate with the government, universities, online celebrities and tourism agencies to promote Chinese traditional culture, online education and tourism. This change met the aims of the state and TikTok finally won endorsement from the government (Z. Zhang, 2020).

Another study shows that the bottom-up grievances of internet users towards platformisation can also trigger state intervention. Guan and Zhang (2020) exemplify the platformisation of e-commerce in the Chinese context from another perspective to show how the dynamics between actors in China are different. During the platformisation, Alibaba increasingly gained a monopolistic position. To cover its costs for promotion and expansion when competing for this monopolistic position, it accelerated its commercial monetisation by charging sellers high annual service fees for keeping their online stores, leaving the existing sellers with no choice but to pay the fee or give up their established online stores, losing all their customers (Guan & Zhang,

2020). This is a typical platform strategy: depending on venture capital for growth during the platformisation process, then once the platform gains a dominant position in the market and users become dependent, starting the process of monetisation by increasing the prices of services to make up the losses (Srnicek, 2017, pp. 26, 45). However, the result was different in China. This mandatory payment irritated the sellers, and given that there was no formal way of contesting the fees, the sellers to engage in started informal protests and boycotts – they went to the online shops of top sellers on Alibaba and purchased items, giving them bad reviews and then refusing to pay, in order to disrupt the algorithm and paralyse the website. Thousands of invalid orders were generated, leading to 9.5 million RMB in losses (Liu, 2012). To calm down the sellers, the

30

government required Alibaba to compromise, and the Ministry of Commerce even facilitated the negotiation between Alibaba and the protesters. Finally, Alibaba agreed to fund the small businesses that were qualified to keep their stores (Guan & Zhang, 2020).

From these cases, it is clear that the Chinese government plays a more central role in platformisation than do its Western counterparts. In China, platforms are required to have social responsibility, helping to maintain the stability and serve the development goals of the state. In the meantime, the flipside of the platformisation, including data acquisition and commodification, are ignored (Z. Zhang, 2020). Even though the Chinese government cannot be treated as an actor with inflated power, as many media and studies have claimed, the fragmented governance characterises the contemporary governance of China and the way the Chinese state has had to compromise with different actors in the Chinese and global markets, as well as local Chinese bureaus (Hong, 2017a, pp. 77–78; Hong & Xu, 2019). In this sense, as Z. Zhang (2020) claims, the Chinese state is not a pure regulator, but simply one of the stakeholders in the process of platformisation. Therefore, nuanced analysis of the power of Chinese state is necessary to examine how the state interacts with different actors in the platformisation process when it comes to live streaming.

Empowerment and exploitation in a different social context

In addition to the different role of the state and government, the divergence of social actors and social conditions in China can also provide a new lens into the studies of platformisation. While platformisation opens up new forms of exploitation, it also affords possibilities for the empowerment of certain groups of users. Theoretically, platform-based ICTs are crowd-enhancers, and platformisation helps bottom-up organizing and crowd-sourced, democratic action. However, the configurations of those technologies within the framework of political economy of

31

informational capitalism make these actions manipulable. And algorithms themselves are not inherently neutral, but rather created to achieve particular ends and controlled by their initial distributors (Chun, 2011, pp. 27, 53; Galloway, 2004, p. xii). Thus, these user activities led by the algorithms can be exploited or even directed by the key stakeholders for their own interest, which might not be democratic and sustainable (Cohen, 2017). Therefore, in the Western context, scholars tend to worry about the issues raised by platformisation and the risk of capitalisation, rather than being optimistic about the potential of empowerment in this process.

However, platformisation might have different implication in a Chinese context. For instance, while the privacy issues of a social credit system (SCS) concern citizens of the West,

Chinese internet users instead understand it as a system to boost moral integrity and ensure honest dealings in a Chinese context (Kostka, 2019). As the SCS punishes those who misbehave and rewards citizens for their good behaviour, it could potentially empower good citizens. Another study conducted Lin and de Kloet (2019) shows that Kuaishou, as a platform targeting marginalized Chinese people in undeveloped cities and rural areas, turns grassroots citizens into creative workers, empowering these people and opening up opportunities for creative entrepreneurship in remote areas. These studies show that not just the state, but the other stakeholders and the dynamic of actors in China, are diverse, so the existing framework of platformisation may not be applicable in China.

Recently, the tensions between the United States, China and the European Union has manifested as a conflict over infrastructural power and ideological clash in digital space (Van

Dijck, 2020). America’s inhibition of Huawei and Tik Tok is typical of this conflict. Given this background, analysing the platformisation of live streaming in China can help us to understand its divergence and thus can contribute to an understanding of these tensions.

32

Platform power does not lie with individual companies but rather in coordinating the rule- setting power of an integrated ecosystem (Van Dijck, 2020). Therefore, to understand the platformisation of live streaming in China, we cannot just look at the individual platform; instead, a more precise analysis of the integrated ecosystem of platforms is needed. The interrelational, dynamic and often oligopolistic structure of the whole ecosystem (van Dijck et al., 2019) should be examined when we analyse the platformisation of live streaming in China. In addition, nuanced analyses of the power of the stakeholders at different levels and layers, and the dynamics between actors, need nuanced analysis. Drawn from this framework, this thesis will first look at the history of live streaming in order to provide an understanding of the ecosystem.

The development of live streaming in the United States

To better understand the contemporary situation of live streaming, it is necessary to examine the historical development of the industry. In particular, the political economy approach proposes a historically informed and dynamic way to understand social transformation (Mosco, 2009, p. 3).

Moreover, platform theory is ‘sensitive to the historical origins of the particular platforms’ (Wang

& Lobato, 2019). Therefore, this section will analyse the evolution of live streaming platforms and practices. It is notable that the development of Western and Chinese live streaming has taken different trajectories over recent decades, building up different live streaming ecosystems that call for closer analysis in terms of their political and economic implications.

Lifecast and Twitch.tv

In 1996, Jennifer Ringley created a website called JenniCam to show her daily activities through static shots from her camera. The most recent images from the camera would be uploaded to the

33

website every two to 20 minutes (Jimroglou, 1999), fully documenting her life (Andrejevic, 2004).

Inspired by the lifecast concept, in 1998 the movie The Truman Show portrayed the life of Truman

Burbank, who had spent his entire life in an enormous set near Hollywood, with thousands of video cameras to record his every move, beamed to a global television audience 24/7 (The Truman Show,

2020). While the movie seems like an exaggeration of the recording technology and a sense of voyeurism, what happened over the next two decades showed that the live show was far more than an intriguing concept. After JenniCam, people were inspired to use webcams to reveal their daily lives, and we have witnessed the lucrative nature of pornographic cam shows (Andrejevic, 2004, pp. 194, 205).

From March 2007, a 23-year-old young American internet entrepreneur, Justin Kan, and his colleagues created a single channel to broadcast Justin’s life 24/7. He put a webcam on his head to ‘lifecast’ his eating, sleeping and even dating (Guynn, 2007). Viewers could see everything that Kan saw, and Kan also interacted with the audience through live chat rooms, emails and even phone calls (Guynn, 2007). The stream lasted eight months, and afterwards Justin.tv got funding to become a platform that allowed anyone to live broadcast their videos online (Popper, 2014).

Justin.tv can be regarded as the first live streaming platform for amateur individuals. The site has various categories, including tech, sports, entertainment, gaming and others.

34

Figure 3: The lifecast of Justin Kan (Guynn, 2007)

Shortly after this, video gaming – and especially e-sport gaming and spectatorship – gained its popularity, and began to outweigh film and music industries in its viewership from the early

2010s (Kaytoue et al., 2012). Since then, watching gameplay video and watching live game playing increasingly developed as popular forms of entertainment (Nascimento et al., 2014). As one of the categories, gaming streaming grew dramatically and became the most popular content category on Justin.tv (Lynley, 2011). Then Justin.tv developed the game streaming platform known as Twitch as a separate site in 2011. After three years, on 10 February 2014, Twitch’s and

Justin.tv’s parent company was rebranded as Twitch Interactive. On 5 August 2014, the original

Justin.tv was closed down and from that point the company focused solely on Twitch (NDTV,

2014).

Following Twitch, other social media companies such as Facebook, YouTube and Twitter started their own live streaming services from 2015 (van Brandenburg et al., 2016). Services such as Facebook Live, YouTube Live and Periscope (owned by Twitter) were launched and dedicated to letting broadcasters stream a variety of content, as Justin.tv did, including indoor/outdoor live

35

shows, education and event streaming (concert and sports), rather than just games. From that point, live streaming escalated beyond the ‘webcam subculture’ and became mainstream (Andrejevic,

2004).

The practice of interactivity: Social TV and live streaming

It needs to be pointed out that the kind of live streaming that we are currently discussing does not just enable viewers to experience new forms of watching, but also enables new forms of social interaction between streamers and viewers. Live news, sports streaming and event streaming have been around for many years, and they provided a new form of watching experience as the content was broadcast in real-time. However, the social media live streaming discussed in this thesis also offers different forms of sociality as it turns the broadcasting into conversations between the streamers and viewers.

For some people, watching itself was not meant to be a social activity. However, for others, sharing watching experiences could enhance enjoyment and initiate conversations, providing a core purpose for watching. As Weisz et al. (2007, p. 878) point out, ‘socializing around media is perhaps just as important as the media itself, and supporting social interactions during media consumption can significantly affect, and we hope enhance, the viewing experience’. Along with the transition from video watching, the interactions during the watching practices is also evolving simultaneously and has become an indispensable part of the experience. This section therefore traces how the practices of sociality are attached to watching, and sheds light on the core of live streaming.

36

Sociability of watching

From the very beginning, watching TV shows at home on the couch with family members or friends, and discussing the content with physically present co-viewers, provided the original form of social interaction during watching. This form of interaction has been coined ‘direct sociability’

(Ducheneaut et al., 2008). After the watching, usually the viewers would discuss the shows they had watched with colleagues or friends, a practice that is termed ‘indirect sociability’ (Ducheneaut et al., 2008). This form of indirect sociability is also known as ‘water cooler moments’, referring to informal conversations during breaks at work (Lochrie & Coulton, 2012).

With the help of communicative devices, via texting or making calls to friends or families remotely when watching the same TV show or afterwards, viewers extended their direct and indirect sociability. In this way, the communicative devices helped to extend the sociability of watching, which came to transcend spatial limitations.

With the development of social network sites, people could share their watching experience and discuss TV programs with others online, both instantaneously when watching or afterwards with both acquaintances and strangers (Hambrick et al., 2010; Krämer et al., 2015, pp. 255–56; Lochrie &

Coulton, 2012). Increasingly, audiences have tended to use mobile devices to access social media, with mobile phones or tablets for social interactions becoming supplementary to the first screen of television when watching – this is known as ‘second screening’ (Lochrie & Coulton, 2012). Reports indicate that nearly half of US mobile devices users are using portable devices as second screens to interact with other audiences while watching television programs (Nielsen, 2013). Elements of these second screen conversations are related to the topic on the first screen, and in this case the activities become an extension of concurrent group viewing or water cooler moments. For example, soccer fans can post their feelings on social media when watching UEFA European Championships. In China, nationwide

37

viewers of Spring Festival Gala praise or criticise the show on their Weibo or WeChat while they are simultaneously watching the show (Chen, 2020). In short, the use of social media helps increase the sociability of watching, transcending the limitations of both space and personal social circles.

Consequently, video watching is becoming increasingly social, initiating the concept of integrating watching and socialising on a single platform.

Co-viewing and socialisation on a single platform

For the purpose of bringing together socialisation and watching to enhance the experience of social interaction, video sites have enabled on-site interactions for viewers. Almost every video watching site has a comment zone or chat zone to facilitate conversations about the videos shown on the same site. It is common to see viewers posting comments on video-sharing platforms such as

YouTube to discuss the video content asynchronously: they reply to the comments or replies that were posted a few seconds or even a few days earlier, starting the conversation with the other viewers in an asynchronous way. Watching, commenting and replying asynchronously on video sites could be seen as a beginning of combining the first and second screens.

In fact, media companies were working on putting together the activities of watching and interaction on a single platform. For example, Social TV combines social media and television

(Shamma et al., 2012; Weisz et al., 2007), which enables practices whereby viewers interact with each other based on the television content they are concurrently watching on the same platform.

Activities such as voice chat, text chat and peer rating are enabled on Social TV, these can all significantly enhance the watching experience (Summa, 2011, p. 9). Notably, the examples of the most innovative features of Social TV that facilitate the rise of live streaming are Time-Tag and

Viewer-Broadcaster interactions.

38

Time-Tag, for the pseudo-synchronous interaction

Among the different functions of Social TV, Time-Tag is a noteworthy feature. It comprises personal annotations in a video, which can be created by individual viewers and made publicly visible. This feature is good for explaining the plots and sharing watching experiences (Shamma et al., 2007, p. 279). In a video, when viewers find a memorable section or a notable scene, they can make annotations. Similarly, viewers can also make comments on the annotations of the other viewers, which constitutes a form of asynchronous interactions based on the scene. People may notice that the YouTube comments zone offers a similar function, in which viewers can create a clickable timestamp to highlight a particular scene in order to initiate a conversation on the comments zone based on that scene.

While the timestamp is only limited on the comments zone, a more vivid form of Time-

Tag is bullet-screens, which incorporate annotations or comments into the video itself. In a bullet- screen video, the annotations or comments will be archived based on the timeline of the video, and will be displayed exactly when other viewers are watching the same scene, making the viewers feel they are surrounded by co-viewers. Later viewers can also post bullet-screens to react to existing comments, thus creating annotated conversations for the future audience to view.

While this interaction is still asynchronous, it creates a co-viewing experience and the viewers feel as if they have viewing companions when watching. In this sense, it can be regarded as pseudo-synchronous interaction. In short, Social TV extends ‘sharing the couch’ and ‘water- cooler moments’ (Cameron, 2016) and has become the new ‘digital water cooler’ (Auverset, 2017)

– thus people can share watching experiences through pseudo-synchronous interaction without the constraints of distance and their own social circle.

39

Instant viewer–broadcaster interaction, empowering viewers and changing the narrative of content

Traditionally, the narrative of a video has been decided by the makers and it is usually not changeable after release, although that is not always the case. In fact, there have been cases where the interactions between audience and broadcasters/program makers have allowed changes in the narrative. For example, the well-known American drama television series Lost has changed its narrative to fit the needs of the audience. Fans groups of Lost used social media such as Facebook and Twitter to discuss the program. For example, in Season 3, there were complaints about two newly joined characters, who fans found annoying. They expressed their dissatisfaction on Facebook. Finally, the directors catered to the audience and decided to remove these two characters by ‘killing’ them off (Summa, 2011, p.

32). In addition, some reality shows occasionally launch surveys and change the narrative of future episodes according to the preferences of the audience. In these cases, the response from audience influences the narrative of the content, and in more recent Social TV and live streaming further empowers audiences to change the narrative to cater to their needs.

Stemming from social TV, the interaction in live streaming becomes both live and interactive. Viewers are sharing their watching experience synchronously; furthermore, these interactions are not only between viewers, as the streamers are also involved in the interaction.

The individual streamers can react to the comments or follow the suggestions of viewers, constituting synchronous viewing and streamer interactions. The reflexive activity enabled by viewer/streamer interaction affects the original live streaming narrative, making it more engaging and constantly changing for both streamers and viewers. Martini (2018b) explores how synchronous and asynchronous forms of watching can trigger different forms of connected activity.

The study compares traditional online video watching and live stream video watching, and the results show that the interactions – both user/broadcaster and user/user – generated by online

40

videos provide a long term, low-intensity, passive and minimally reflexive activity and user– broadcaster interaction. However, the interaction generated by watching live streaming is characterised by high-intensity user engagement with significant reflexive activity and user– broadcaster interaction, which can also facilitate community building around streamers (Martini,

2018b, pp. 16, 17). As a result, YouTube gameplay video uploaders finally turned to streaming on

Twitch because they found that the feedback of YouTube viewers was not instantaneous. To them, the interactions on Twitch are more engaging and interactive because streamers can receive instant feedback as content creators (Johnson & Woodcock, 2019b). Now it is common to see viewers and streamers building up both online and offline relationships after live streaming. Thus connectivity and social ties build on live stream platforms, enabling significantly stronger interactivity than occurs on more static older platforms (Johnson & Woodcock, 2019b).

In conclusion, Time-Tag and synchronous viewer and broadcaster interactions changed socialisation during video watching, empowered viewers and contributed to the growth of live streaming.

Figure 4: Interface for the live interaction on Twitch5

5 Taken from streamer Plalism from Twitch, 15 July 2020, https://www.twitch.tv/plalism

41

The popularity of live show performances: Different trajectories in the development of Chinese live streaming

Like its Western counterpart, live streaming in China also started from ‘lifecast’, but the development of live streaming in China followed a different trajectory. While game streaming is the core element of Western live streaming, on contemporary Chinese live streaming platforms, the dominating content is an updated version of ‘lifecast’, which is termed ‘live show performances’ in this thesis. On ‘live show performances’ channels, individual streamers show their skills either indoors or outdoors, with singing, dancing, talk shows and cooking typical forms of performance

(Lu et al., 2018; Zhou, 2017). To further understand this phenomenon, we must take a look at the original form of Chinese ‘showroom performances’ in the early 2000s.

From voice chat to live video entertainment

Voice chat platform YY was launched in 2005, and is regarded as the predecessor of live streaming in China (Xue, 2014). In the beginning, YY was used mainly for the communication between players in massively multiplayer online games (MMOGs) such as World of Warcraft. Players of

MMOG formed guilds or clan channels on YY and use the channels for real-time voice chat with their community members when playing games. From 2009, YY transitioned from voice chat software between players into another form – a UGC platform on which users could perform a singing or talk show, sharing pictures and videos while talking to other users. A typical form of performance is a broadcaster sitting in front of the cam to host the show, trying to say something funny, sing a song or show their talents, for example, B-box6. The performance could be either video or audio-based (Xue, 2014). In this earlier form of ‘showroom performances’, ‘cam girls’

6 B-box is a form of vocal performance that a performer mimicking musical instruments using the mouth, lips, tongue, hands, and voice.

42

was a term used to describe those female performers, which is now regarded as the pioneer of the burgeoning industry of live interactive web entertainment (Yin & Fung, 2017). Not long afterwards, the popularity of showroom performances outweighed game communication in YY, and the number of entertainment channels soon surpassed the number of game channels. Consequently, in

2012 YY changed into a video streaming platform featuring live performances.

The socialising of watching in China: From pseudo-synchronous interaction on Acfun to synchronous interaction on Douyu

In addition to the development of YY, Chinese video sharing sites also underwent significant changes in terms of their interactive functions. Acfun and Bilibili were two Chinese video sites that were launched around 2010. They learnt from Japanese video site Niconico Douga, featuring the sharing of anime, and Danmuku (known as bullet-screen). Soon the websites became subcultural hubs for remixing videos and popular internet memes (Yin & Fung, 2017, p. 138).

Both Acfun and Bilibili are similar to Social TV in that they focus on the social conversations around video watching. The most notable function of these two sites is not the videos themselves, but the bullet-screen created by viewers mainly for socialisation. The video records the time tag when viewers post comments. When future viewers watch the video, the comments of previous viewers are displayed at the same timeline of the original video show, making viewers feel the

‘existence’ of co-viewers. For example, when watching horror movies, former viewers tend to post comments like ‘be careful’, ‘high energy alert’ to alert viewers a few seconds before the horror scene. In this way, later viewers can prepare for the scene psychologically, and sometimes they post comments to thank posters for their reminders, or express their instant reaction to the specific scenes. However, the co-viewers could also become spoilers and ruin other viewers’ viewing experiences (Yin & Fung, 2017) – just as real life spoilers can.

43

In this way, the viewers on these platforms feel that they are watching with other viewers on a couch, and thus sharing viewing experiences together. The bullet-screen has become a new digital water cooler that connects asynchronous viewers. For some popular videos, there would be numerous comments on the original video, and what viewers are actually watching might be the comments rather than the original videos themselves (Yin & Fung, 2017, p. 139). This form of pseudo-synchronous interaction initiated by bullet-screen became the core feature of Acfun and

Bilibili and resulted in their rising prosperity.

Stream content

Bullet screen Chat room (with donation notification)

Figure 5: Interface for the live interaction on Douyu7

In January 2014, Acfun released a separate site known as Douyu; like Twitch.tv, it focused on gameplay streaming in the beginning, and it also enables live comments for viewers to interact synchronously. The difference is that the comments are not only displayed in the chat room, but also as a bullet-screen on the video. This means viewers can chat with both streamers and other

7 Taken from streamer Russian Godfather from Douyu, 15 July 2020, https://www.douyu.com/848337

44

viewers live when they are watching the streaming content and live comments simultaneously.

Douyu successfully attracted substantial investments – notably, Tencent invested millions of dollars and became the largest shareholder (43.1 per cent) of Douyu (DouYu International

Holdings Limited, 2019). Today, Douyu can afford to let everyone freely broadcast anything in addition to games.

The characteristics of Chinese live streaming industries

In the Western context, platformisation could fundamentally transform culture production, distribution and circulation (Poell, Nieborg & van Dijck, 2019), and scholars worry about the issues resulting from the process of datafication and commodification (Cohen, 2017; Srnicek, 2017;

Van Dijck et al., 2018). However, China is not a capitalist country, and as mentioned previously, the role of the state is different; there are also cultural differences. Therefore, the process of platformisation may have a different trajectory (de Kloet et al., 2019). In this sense, the divergent social conditions and historical development of the live streaming platforms may help to explain the significant variations between contemporary Chinese and Western modes of platformisation.

The following section highlights the most important divergences in the two industries, initiating further discussions about the underlying implications about platformisation in China. Five different perspectives of live streaming industries are examined: the market share of the platforms; the demographics of viewers; streaming content; monetization channels; and partnerships between platforms and streamers. In a careful examination of streaming practices and the underlying commercial forces from these five perspectives, the power relationship between different agencies in the live stream industries is revealed.

45

The quasi-monopoly of Twitch in the West and the oligopolistic market in China

Twitch is almost a monopoly in the live streaming industry in the West. It has 140 million monthly users (Khan, 2019) and takes about 75 per cent of the market share. The remainder of the market is shared among YouTube Live (20 per cent) and other smaller platforms such as Facebook Live and Microsoft’s Mixer8 (Yosilewitz, 2019).

According to the report from CNNIC, the number of total live streaming users in China reached 433 million, which represents about 50.9 per cent of overall Chinese internet users

(CNNIC, 2019, p. 22). Driven by the huge market, Chinese live streaming industries have gone through a rapid expansion from 2015, and more than one thousand live streaming companies have been formed in a fiercely competitive market (Laoyu, 2019). However, from 2016 many companies died out quickly, partly because of the escalating restrictions and penalties enforced by the Chinese government from 2016 (CAC, 2018b; SAPPRFT, 2017a; Wei, 2017), and partly because of the mergers and acquisitions by the giants (D. Li, 2020; Laoyu, 2019). Right now, while there are still more than one hundred live stream platforms in China, only a few of them dominate the whole industry. By market share, the top live streaming platforms in China are YY, Huya

(owned by YY Inc), Kuaishou and Douyu, resulting in an oligopoly structure. Douyu and Huya have160 million and 120 million active monthly users respectively, and both are listed on the New

York Stock Exchange (Yeung, 2020, p. 104). Unlike their Western counterparts, these Chinese live streaming platforms are competing fiercely and provide similar forms of content. Top streamers are the key resource for which they are competing, and these streamers are required to

8 Mixer has been closed since 22 July 2020. The streamer and viewers of Mixer have moved to Facebook, Microsoft’s current partner for game streaming. See https://www.theverge.com/2020/7/22/21334082/mixer- shut-down-twitch-facebook-gaming-partnership and https://mixer.com

46

sign exclusive contracts to stream on one platform only. Successful signed streamers will be rewarded with huge signing bonuses, with up to US$2 million per year (H. Liu, 2020). However, the streamers of Twitch do not enjoy such high signing bonuses, probably because of the quasi- monopoly situation of Twitch – there is no alternative platform for those top streamers.

While the government intervention in the market economy is less frequently in the United

States, the Chinese government tends to believe that intervention in the market for the political and cultural goals is necessary, especially when it comes to digital media platforms (Wang &

Lobato, 2019). The Chinese government does not just restrict or regulate the media, however, but also supports them, depending on whether the goals and development of the platforms are in alignment with central and local government (Z. Zhang, 2020). In this context, Twitch has been blocked in China since 2018. Although the underlying reason is unclear, Twitch now has few

Chinese domestic users after being banned. Yet some Chinese live streaming companies are extending their international markets. For example, YY has acquired Singapore-based live streaming platform Bigo Live, which was the biggest SEA live streaming platform in 2019 (Cheng,

2019). More recently, Bigo announced that it was planning a global expansion starting from India, to the Middle East, to the United States (Singh, 2019).

Different viewer demographics

The number of viewers on Twitch is huge – it has an average of 953 thousand average concurrent viewers. At peak usage, the number of overall concurrent viewers of Twitch can reach one million.

For example, in 2017 over one million concurrent viewers watched ELEAGUE’s Counter-Strike:

Global Offensive Major Grand Final together (Needed, 2018). Compared with Western viewership,

Chinese audiences are huge. According to the report, China is the world’s largest game streaming

47

market, with approximately 4.9 times the monthly average users of the US market in 2018 (Douyu,

2019, p. 1). In terms of the total number of live streaming users, China reached 560 million in 2020

(Xia, 2020). In 2019, the highest number of concurrent users for live streaming was reported as 67 million, with the Tencent League of Legends Pro League Finals (Douyu, 2019, p. 126).

Although China has more viewers, the demographics of users of China are similar to the rest of the world. For all the live streaming platforms, the viewers are mostly teenagers: 55 per cent of Twitch users are aged between 18 and 34 (Needed, 2018), which is similar to the demographics of China. These viewers of live streaming are mostly males for both Western and

Chinese platforms, while the proportion of male viewers of Twitch is higher than the others.

According to one marketing report, 81.5 per cent of Twitch users are male (Influencer, 2020).

The use of different languages is the main factor that separates the streaming communities, given the high level of synchronous interactions among viewers and streamers. Different languages coexist on Twitch, although English is still the dominant language, with about 62.87 per cent of

Twitch activity being in the English language (Needed, 2018). There are few bilingual channels, and non-English languages are in the minority, so streamers will usually add an English title or

English channel description to draw more viewers (Recktenwald & Du, 2016). In China, most streamers speak Mandarin or their local dialects, so viewers will tend to just use Mandarin to communicate with streamers and other viewers. Although English is also used on Chinese platforms, it is mainly used by the channels featuring language education or lifecasts of living abroad. As live streaming hinges on live interactions, language becomes a major barrier for the international viewership on live streaming platforms.

48

Genres of streaming content: Enthusiasm for erotic performances

Live streaming platforms classify live streaming content into different categories; however, there has been no consensus about the classification for live streaming content. Given that different forms of live streaming content have significant variations, this research classifies current live streaming content into four different categories for clarification and further study. The criteria for the classification draws from both the sorting of live streaming platforms and the categorisation given by official government reporting (CNNIC, 2019, p. 22; Douyu, 2019, 2020b; YY, 2020).

Figure 6: The categories of live streaming content

Categories Examples Game streaming Gameplay Outdoor live cast/show (Huwai) Hunting, survival, hiking, shopping Indoor live cast/show (Xiuchang) Indoor dancing, cooking, singing, pets Event live streaming9 Sports, e-sports, concerts E-commerce live streaming Introducing and promoting goods: Taobao

On Western live streaming platforms, game streaming is the most popular category, with far more streamers and viewers than for other categories of content. For example, both Twitch and

YouTube feature game streaming, while other forms of live streaming on these two platforms are the less popular categories of content. However, in China non-game-centric live streaming makes up a higher proportion of the content, with indoor and outdoor live show/cast being the most popular.

9 The event live stream here means an individual streamer/streamers live streaming an event from their cam, not from official broadcasters of the event. As this study focuses on individual live streaming, the professional streaming of events is excluded.

49

This phenomenon can be attributed to viewership habits. On Twitch, ‘non-endemic content’ is forbidden. For a long time, this has meant that any non-gaming materials are not welcomed

(Recktenwald & Du, 2016). Due to the dominance of game-centric streaming, Twitch has more male viewers and streamers. In contrast, the female streamers are in the minority and are frequently harassed with discriminatory terms such as ‘cam girl’ or ‘titty streamers’, especially when they are not skilful enough and are regarded as drawing undeserved attention and donations from viewers (Ruberg et al.,

2019). In these circumstances, female streamers on the Western platforms are somehow perceived as the disruptors of the male-dominated video game culture (Cross, 2016; Phillips, 2017).

However, in China, the female streamers are welcomed by both platforms and viewers.

Figure 7 shows the clear example of the gender variation in live streaming platforms of different regions. Chinese viewers and platforms are more willing to pay for the content of game streamers, and they are more likely to pay the female streamers. However, in the United States, game streamers – especially the female game streamers – are usually unpaid, which means that their efforts are not as yet widely acknowledged by the viewers and platforms.

50

Figure 7: Percentage of gaming streamers not paid for content they create in 2019 (Yeung, 2020)

Female streamers in China not only get more recognition, but also represent a higher proportion compared with the Western platforms; for the top streamers, the ratio of male to female streamers reaches 50:50 (Institute, 2019). It has to be pointed out, however, that while cam shows, or even those featuring pornography, are legitimate in most Western countries, sexually explicit content is strictly forbidden in China. The pornographic cam show is undoubtedly crossing a red line in China. Female live shows become a legitimate and implicit substitute for the illegitimate cam show, and are always regarded as a grey area – although all the platforms in China prohibit

‘illegal content’ such as nudity, ‘streamers are often allowed to perform their sexuality in more subtle and implicit ways in order to lure viewers’ (Zou, 2018). On the other hand, a large number of single men from the subordinate social groups in China need something to deal with their loneliness, so watching and interacting with female streamers becomes a way to cater for their psychological needs (Cunningham et al., 2019). Today many live streaming platforms in China including Huajiao and Inke are still featuring camgirls (Zhang & Hjorth, 2017). Moreover, on every platform there is a category of content called ‘yan zhi’(level of attractiveness), in which most streamers are females with attractive faces and bodies to appeal to viewers.

In short, while game streaming of the male streamers dominates the western live streaming platforms, Chinese platforms and viewers show an enthusiasm for the live shows of female streamers. However, Douyu as a distinctive platform has turned its focus to game-centric streaming, along with various forms of content including outdoor live shows from 2016, rather than merely focusing on live shows (Zhang & Hjorth, 2017).

51

From advertising to tipping: The differences of core business and sources of revenue

Twitch: Relying on subscription and ads

Twitch generates its revenue in three ways: advertising, subscriptions and tips. While Twitch has not disclosed the precise portions of its revenue, analytics indicate that Twitch makes money mainly through advertising and subscriptions, while donation is just a prospective source of revenue (Fortney, 2019; Lqbal, 2019). The ad revenue of Twitch in 2018 was US$230 million

(Anand, 2020).

In terms of the advertising formats, the advertising model on Twitch is similar to those on video websites such as YouTube. Sponsors can contact Twitch and pay the advertising fee to promote their product. Advertisements will be integrated into every channel, from Homepage

Carousel and Headliner to Bounty Board, as Figure 8 shows. Moreover, short video ads would be displayed before viewers watch the streaming content (Twitch, 2020c).

Figure 8: Advertisement display of Twitch (Twitch, 2020c)

52

In addition to advertising, subscription fees from viewers are another major revenue source for Twitch. Viewers can pay a couple of dollars per month to enjoy premium services. The monthly

Twitch subscription costs between US$4.99 and US$24.99. There are different subscription tiers, depending on the fee. Higher tiers of subscriptions come with more benefits for the viewers. For all tiers of subscriptions, Twitch will cut down the advertisements for the viewers, so the subscribers can fully engage the streaming content. In addition, subscribers can enjoy a few aesthetic upgrades including special emojis and customisable messages when they are chatting on the channels, so they can easily attract the attention of streamers and other viewers (Twitch, 2020b).

Tips and donations have not yet emerged as a main alternative source of revenue for Twitch.

To date, the main revenue sources for Twitch remain subscription and advertisement fees, similar to the business models of other social media or video websites. However, in China, donations constitute a significant part of the revenue while advertising revenue is a secondary source, as analysed in the following section.

Chinese platforms: Profiting from paying viewers rather than ads

Among all the platforms in China, the viewers of indoor live shows had around 24.0 per cent of the market in 2019, an increase of 4.3 percentage compared with 2018. Unlike in the West, game streaming viewers in China represent only 28.4 per cent, a proportion that has remained stable over the past few years (CNNIC, 2019). What is surprising is that, while game streaming has more channels and viewers, non-game live streaming gains more total revenues. For the total revenue of live streaming in China in 2018, the revenue from non-game live streaming was 34 billion RMB, accounting for approximately 72 per cent. However, game streaming only earned 13 billion RMB, which accounted for 28 per cent (Douyu, 2019, p. 127).

53

While non-game streaming is more profitable than game streaming in China, another important point to note is that advertising has never been a core business for Chinese live streaming platforms. Rather, Chinese live streaming platforms develop various types of monetisation and mostly rely on the revenue from paying viewers. As a leading streaming platform in China, Douyu published the Prospectus of DouYu in 2019 and was listed afterwards. The prospectus and the annual report from 2019 to 2020 (Figure 9) show the details of its core business, which is representative of Chinese live streaming platforms.

Figure 9: Cost and revenues of Douyu from 2016 to 2018 (Douyu, 2019, p. 14)

The cost of operating a live streaming platform is particularly high. Douyu was launched in 2014, but the cost of revenues has outweighed net revenues in the past few years. In 2018, for the first time,

Douyu reported a profit amounting to US$22.5 million. As shown in Figure 9, the net revenue was

US$544.5 million, but the cost of revenues was US$522 million, leading to a gross profit of just

54

US$22.5 million (Douyu, 2019, p. 14). Therefore, although Douyu has been expanding dramatically over the past few years, the venture capital investment remained its main funding for the operation until 2019.

In 2019, the gross profit of Douyu finally reached US$171 million, signifying the beginning of its self- sufficiency (Douyu, 2020a, p. 1).

Figure 10: Cost of revenues of Douyu from 2016 to 2018 (Douyu, 2019, p. 97)

Figure 11: Cost of revenues of Douyu from 2017 to 2019 (Douyu, 2020a, p. 80)

Among all it costs, Douyu’s main expenses are revenue-sharing fees, content costs and bandwidth costs. Not surprisingly, bandwidth costs are constantly increasing as the user base and user engagement grow. However, the ratio of revenue sharing fees and content cost among total costs have been increasing dramatically by about 7 per cent every year. In 2019, the total cost was

US$870 million, revenue-sharing fees and content cost were 85 per cent and the bandwidth cost represented only 10.1 per cent. As Douyu explains, revenue-sharing fees are the payments made to streamers based on tips and subscription fees of their channels. The content cost includes copyright purchases, compensating and recruiting streamers, and the cost of generating self- produced content (Douyu, 2019, p. 97). In short, significant fees are paid to the content generators

55

on the platforms, while the labour of streamers are not free but paid or even sometimes overpaid because of the fierce competition between platforms. As the report shows, salaries and recruitment fees paid to the top streamers have been the major cost for the platforms, leading to cash shortages for some platforms and their closure – for instance, the failure of Panda TV, which was founded by the son of a Chinese tycoon (Yeung, 2020, p. 104).

From the perspective of revenue: profiting from paying viewers rather than advertising

Unlike other media industries and Western live streaming platforms, live streaming platforms in

China have to date only received a small proportion of revenue from advertising. According to the documents from IPO, the main monetization sources of Douyu are live streaming revenue – sales of virtual gifts (tips) and subscription fees (Douyu, 2019, p. 93).

Advertising only generates a minor proportion of the total revenue for Douyu, and that proportion is decreasing. As shown in Figure 12, the advertising revenue was 9.1 per cent of the total revenue in 2018, while live streaming revenue was 90.9 per cent in 2019 (Douyu, 2020a, p. 79).

Figure 12: Cost and revenues of Douyu from 2017 to 2019 (Douyu, 2020a, p. 79)

56

Figure 13: Paying users of Douyu (Douyu, 2019, p. 94)

In 2017, 2018 and 2019, Douyu generated US$35.5 million and US$48.9 million,

US$73.4 million US dollars respectively from the sale of advertising, representing 13.2 per cent,

9.4 per cent and 7.0 per cent of the total revenues for this same period (Douyu, 2020a, p. 9). In contrast, tips and subscription revenue in 2017, 2018, and 2019 was 80.7 per cent, 86.1 per cent and 90.0 per cent of total revenues respectively (Douyu, 2020a, p. 79). According to this report, the paying users (who tip and subscribe) increased from 1.7 per cent in 2017 to 4.4 per cent in

2019, reaching 7.3 million in 2019 (Douyu, 2019, p. 93; 2020a, p. 49). In short, the main revenue of Douyu comes from the tips and subscriptions from paying viewers, and paying viewers only make up a small proportion of the overall viewers. Most viewers on Douyu are still free riders.

It might seem unbelievable that a platform does not rely on the advertising fees, but Douyu is not a special case. Another top Chinese live streaming platform, Huya, has a similar business model. Its major revenue comes from tips and subscriptions; furthermore, Huya did not broadcast any forms of advertising in the first few years after its launch in 2014. In 2019, according to its annual report, the revenue generated from tips and subscription fees reached US$1146 million, accounting for more than 95.2 per cent of the overall revenues, while advertising and other revenues was only US$57.2 million, or about 4.7 per cent of total revenue (Huya, 2020, p. 4).

Advertising of Chinese live streaming platforms

Although the proportion of advertising revenue is low, this does not mean Chinese platforms disregard advertising altogether. Douyu also values advertising, and different categories of advertisements, including full-screen advertisements, slides, banners, links, videos, logos and buttons, are displayed on

Douyu in different forms. In addition to traditional forms of advertising, it has an innovative e- commerce-focused model of advertising. In this form of advertising, streamers are encouraged at the

57

instigation of advertisers or platforms to play an active role to promote products during their online show or offline activities (Douyu, 2019, pp. 97, 146). Sometimes streamers are required to sign advertising contracts with the platforms, and may share a portion of the advertising revenue. However, streamers cannot display advertisements on their own without the approval of the platform, and a direct contract between the advertiser and streamer is forbidden (Douyu, 2019, p. 146).

We can see that Chinese live streaming platforms such as Douyu are still developing strategies to improve their ad revenues. However, although China has a much larger market than the United States, the revenue from advertising is much lower compared with Twitch – in 2019,

Twitch generated about US$300 million from advertising, while Douyu and Huya received

US$73.4 million (Douyu, 2019, p. 96) and US$57.2 million (Huya, 2019, p. 4) respectively. While tips are not a major revenue source for Twitch yet, for Chinese live streaming platforms this is a significant source of income that outweighs advertising. Therefore, it can be seen that profiting from the payment from viewers rather than ads fee from advertisers is the key feature of Chinese platforms. So the sources of revenue are divergent: Twitch continues to develop different models of advertising and premium services for subscribers, while Chinese live streaming platforms are developing multiple strategies to lure donations from daily viewers. As they are catering for different stakeholders, this leads to distinct power relationships between agents in the live streaming industries. The political economic implications of these arrangements will be explored in Chapter 3.

Modes of partnership and employment: A unique talent agency model

Last but not least, Chinese and Western platforms have similar strategies regarding the management of streamers. Most streamers are still freelancers – they earn via donations and stream

58

occasionally at a time of their own choosing. In other words, they treat live streaming as their part- time job or just a hobby. In addition to the freelancers, both Chinse live streaming platforms and the Western platforms have another group of streamers – so-called ‘partnered’ or ‘affiliate’ streamers, who need to stream for consistently longer hours, but then share the revenue from ads and subscriptions with the platforms (Needed, 2018; Zhang et al., 2019). However, these partner streamers are not employees of the platforms, but rather independent contractors who ‘cooperate’ with the platforms. They have no protections and do not enjoy the rights of actual employees, rendering their jobs precarious. For example, there are clauses in their contracts that demand the partnered streamers meet requirements such as minimum streaming hours and subscription numbers, in order to fulfil the contract and receive their payments. If the streamers do not meet the specified requirements of the period, they are then regarded as breaching the agreement, and may even be fined according to the contract. There have been contractual disputes between streamers and platforms in recent years (Douyu, 2020a).

It is notable that Chinese live stream platforms, such as Douyu, also have another form of partnership with streamers – the talent agency model. Now most streamers in China are affiliated with agents rather than streaming independently (Douyu, 2019, p. 2). Agents, or guilds, are intermediary stakeholders between platform and streamers. They are commercial organisations that function like management agencies in the entertainment industry. They identify promising streamers and train them, support them with streaming devices or facilities, and recommend streamers to platforms and handle their legal affairs. These talent agencies are similar to the multichannel networks (MCN) on YouTube, which serve as a kind of intermediary between celebrities and platforms, to manage, monetize and professionalize UGC (Cunningham, Craig and

Silver 2016; Lobato 2016). While MCN usually operates around advertising infrastructure (Lobato

59

2016), agents in China operate around the donation infrastructure instead, because of the divergence of revenue mode. Moreover, Chinese agents do not only target online celebrities but also offline talents, and the management extends to every aspect of streamers’ daily life. Reports show that they are the most profitable sectors of the live-streaming industries (Ding, 2016). These agencies have a number of functions:

• First, agents select promising individual streamers, both online and offline, at times

identifying those who are yet to participate in streaming, but appear to have the talent to be

a streamer (Liu, 2016).

• Second, agents sign contracts with streamers and provide training for streaming skills

(Ding, 2016). To manage streamers, some agencies will even rent an ‘office’ (usually an

apartment or a big house) and divide it into numerous separate rooms for streamers to ‘work

on the sites’. Agencies also provide broadcasting devices (PC, lights, cams, microphone)

and personalized decorations in the rooms, as well as making stylistic suggestions and

providing make-up (L. Zhao, 2017).

• Third, agents sign contracts with platforms and provide different kinds of streamers as

required by the platforms. Agents also put effort into building up streamer popularity,

sometimes by using bots or employing fake viewers. In this way, agents can boost the rank

of streaming channels and let their nominated streamers be highlighted on the platforms.

In addition, agencies will even cooperate with platforms and use ‘promotion’ methods,

such as making mass purchases and donations to act as incentives to lure more tips from

real viewers (Ding, 2016).

• Last but not least, agents regulate their streamers in accordance with their internal

regulatory policies, in compliance with the policies of both platforms and the government.

60

In some cases, the agents also need to deal with legal affairs related to live streaming.

Providing streamers to platforms

1.Scout potential 3.Support 4.Handle legal or existing 2.Provide training streaming affairs streamers facilities

Figure 14: Key functions of streamer agencies

In the Western context, streaming is regarded as a form of precarious employment without insurance and stability, and many streamers seek another professional career for a regular income instead of continuing their more precarious streaming career (Johnson & Woodcock, 2019b).

However, streamers are becoming professional in China due to the popularity, profitability and flexibility of this form of entertainment, especially since the ‘e-sport professional’ and ‘livestream sellers’ have been officially recognised as occupations by China’s Ministry of Human Resources and Social Security (Chen, 2019; X. Shen, 2020). Therefore, increasing numbers of streamers are tending to work as full-time employees for agencies or live streaming companies. Corporations such as Tencent even have their own agencies to provide facility support and training to promising streamers (Lu et al., 2018). Some ‘streamers’ are not only working on live stream platforms, but also working as performers on other social media sites or in real life, depending on the demands made by their agent companies. This group of streamers is usually called Wanghong – internet celebrities – and even less well known streamers imagine themselves as micro-celebrities

(Marwick, 2013; Zou, 2018). As Yin and Fung (2017, p. 4) point out, the Wanghong ecology in

China is based on an attention economy of ‘internet celebrities’, which penetrates the entire contemporary Chinese mediascape.

61

However, these so-called Wanghong do not enjoy as many rights as formal employees.

They are still treated as contractors, like their Western peers, and it is not uncommon for the existing employment policies of the agent companies and the partnership of the talent agency model to violate relevant labour laws (Douyu, 2019, p. 56). The agent companies and the platforms may be subject to penalties, but according to an evaluation by Douyu, while compliance with labour law and any revision of contracts may benefit the streamers, it could also significantly increase operating expenses, resulting in financial risk for the platforms (Douyu, 2019, p. 56). The conflicts between the streamers, agencies and live stream platforms are becoming an increasingly important issue, especially in such an oligopolistic environment.

Conclusion

From this brief overview of the historical development of live streaming and comparison of the contemporary situation of the Western and Chinese live streaming platforms, it is clear that socialising rather than mere watching is the key aspect of live streaming. In China, live streaming industries have their own historical development and distinct characteristics, and they are now heading towards diversified directions in platformisation compared with their Western counterparts.

First, the dominant forms of live streaming are live shows and gaming. Live shows originated from the West, but have become marginalised over the past two decades. In a Western context, live streaming is like a showcase of the activities of streamers. As a dominant form of western live streaming, game streaming is focused on the in-game activities of streamers, including their upper-body and manual operation or gesticulation of their hands. However, there are fewer interactions between streamers and viewers as the streamers are concentrating on gameplay. The

62

live interactions are mostly between viewers in the chat room. The game streamers usually do not play games deliberately to entertain the viewers. In other words, without live streaming, the game streamers would still play – gaming is part of their daily life, and they are just using live streaming as a new channel to reveal real-time gameplay to viewers.

However, as argued in this chapter, live shows have been the mainstream content for live streaming in China, and it is much more profitable than game streaming. In a Chinese context, live streaming has a number of different forms with a variety of content, but they all have something in common: the core of the Chinese live streaming, from its origins, has been based on instant interactions rather than just showcasing personalised activities. Most streamers – especially those live show streamers – are performing deliberately to provide entertainment for viewers in order to generate more profits from these paying viewers, and the synchronous streamer–viewer interactions are the most important part. Even for game streaming practices in China, viewers are more engaged in the streamer– viewer interactions as a result of the specifically designed chat and donation system (Recktenwald &

Du, 2016). This prompts a question: is it possible that these higher level of live interactions in China intensify the process of commodification and datafication in platformisation?

Second, advertising is still the main revenue source of Twitch, but in China the revenues from live streaming come mostly from tips and subscription fees from paying users. This may imply that the traditional audience-as-commodity mode that hinges on advertising is not appliable in the context of live streaming in China; this will be discussed in detail in Chapter 4. Moreover, if tips and subscription fees are much higher than the advertising revenue, then the key stakeholders who shape the live streaming practices might be the paying users, or ‘gold owners’, rather than advertisers or normal viewers. This is a fundamental change that is reconfiguring the influence of advertisers and empowering viewers, challenging the traditional understanding of audience commodity theories. Can

63

we interpret it as a sign that the live streaming in China actually empowers the audience and weakens the influence of advertisers and big capital?

Lastly, it is clear that there are more stakeholders in Chinese live streaming industries and their relationships are more complicated than in other media platform contexts. In addition to the advertisers, viewers, streamers and platforms that they share in common, Chinese live streaming industries are rendered more complex by the agent companies and the governmental authorities that have significant power to shape and influence the streaming practices of individuals and whole industries. In addition, while live streaming in Western countries is still precarious paid labour, live streaming in China is emerging as a professional career pathway. The payments made from the Chinese platforms to content generators, namely the streamers and agent companies, are generally much higher than those paid in the West. These industrial changes have resulted in the transition to different modes of partnership and employment, and they have created novel situations of conflict between different agencies with significant ongoing regulatory implications.

How these different actors interact and the power relationship between them should be the key issues in understanding platformisation in China.

While these divergences are mainly industrial, they inevitably transform culture production, distribution and circulation of live streaming, and therefore they have significant social and regulatory implications. This transformation is rendered more complex when it is situated in broader political-economic contexts as the next two chapters will explain: the changing regulatory environment, economic polarisation, the commodification of cultural production, and the professionalisation of amateurs. The regulation of live streaming is not only about regulating live streaming companies and users, but also the regulation of digital intermediaries, and the management of the dynamic relationship between all these actors. This brings about complex

64

regulatory issues and priorities, which deserves detailed examination. The following chapters will further analyse these characteristics in detail, addressing the questions derived from them based on data collected from interviews and relevant industry and regulatory documents, to explore the distinctive power dynamics in the Chinese live streaming industry.

65

Chapter 3

The role of the state and live streaming regulation in China

The booming market of live streaming industries facilitates economic growth and empowers users, as van Ditmarsch (2013) and David et al. (2015) highlight. As a consequence, the traditional regulatory policies are no longer applicable to control live streaming content (Liu, 2016; Zhao &

Liu, 2017). This is a matter of concern from a regulatory perspective, particularly when certain categories of streaming content have been deemed detrimental to society (CAC, 2018b; MCT,

2016a; Wang & Linyuan, 2017). First, the Chinese state is trying to control new forms of broadcasting and interactivities enabled by live streaming. Second, the complex relationships between multiple agencies and constantly changing industries has proven to be a challenge to the

Chinese state. In the process of platformisation, a potentially new integrated ecosystem is being established and live streaming platforms are taking up a unique position in the process. Thus, developing new policies for maintaining information control of live streaming is understandably the latest media challenge for the state.

This chapter analyses the regulatory framework of the Chinese state for live streaming in the recent economic, cultural and political context of Chinese society. It first introduces selected studies on regulating the global digital environment, including the more recent regulatory dilemma of platformisation. Then it identifies the distinctive features of regulation in relation to live streaming industries in China and the affordances of live streaming compared with other forms of social media sites. The chapter then provides an overview of the challenges that live streaming brings to existing Chinese regulation modes in the process of platformisation, including how the

66

Chinese state, together with industrial practitioners, is responding to these challenges. Finally, in response to the discussion of regulatory intervention by the state, I will analyse how understanding the case of live stream regulation in China can contribute to current discussions on the changing role of the state in the platformisation.

Regulating digital media in a global context

There are two significant strands of theories in relation to the power of state and the regulation of digital media in the West. One suggests that users of digital media are empowered by virtue of the political, economic, technological and culture changes – although state control on digital media can be weakened in this process. For example, Price (2002, p. 147) suggests that new media technologies change the ‘frame for negotiation, for decision making and for the formation and application of policy’, thus outweighing the regulation of the law and the capacity of the state. In this situation, new modes of governance and standards may emerge. Following the network theories of Castell, Howard (2011, p. 20) argues that power has moved from the state to the media system in the new network structures. Moreover, the power to control information resides in media networks themselves rather than state institutions, as ‘media networks are constituted by social relations and communication technologies’ (2011, p. 20). Thus social relations and communication technologies are the key to understanding how information control of digital media and the power of state are weakened. Similarly, by specifically examining the live streaming of a sports event in a Western context, David et al. (2017) suggest that live streaming as a form of mass self- communication enables new politics among those marginalised by dominant networks of power

(David et al., 2017, p. 30).

67

Another strand of theories argues that western nation-states still retain significant control over digital media. These theories suggest that as the studies of the internet have advanced, states have realised both a need for, and new ways of, controlling new media, and in that process they have even developed new strategies to harness digital media for information control. Zittrain and

Palfrey (2008) have studied the internet filtering of states around the world and point out the newly developed mechanisms for digital media. Flew et al. (2016) have closely examined the power of the state in the context of globalisation and the digital transformation of media. They contest the idea that the capacity of nation-states to regulate media has been weakened, and they argue that the nation-states hold more power than they ever realized. A weak state perception, as they suggest, is as result of a rapid social change making historically based path-dependent nation-state systems

‘react more slowly than anticipated to the forces of change’ (Flew et al., 2016, p. 8). It is notable, however, that rather than traditional interventionist approaches, nowadays the state has revised its strategies, applying distinct rules for different media in respect of ownership, licensing, funding and the setting up of different regulatory environments such as forms of self-regulatory environments (Flew et al., 2016, p. 12). It is seen that, given the rise of the platform companies and the inefficiency of self-regulation, the Western regulatory model is transiting from industry self-regulatory model to increased governmental intervention (Gorwa, 2019a; Helberger et al.,

2018). In this way, it can be argued that the power of state has remained strong. These Western theorizations depart from the Chinese situation, where the state apparatus significantly adjusts its media policy to fit the changing political and economic circumstances after its own economic reform (Hong, 2014; Wang, 2018), urging the media to serve the country’s political, economic and cultural goals under the premise of adhering to both the party line and the bottom line (Zhao, 1998).

68

The context of live streaming regulation in China

The focus of Chinese internet regulation

Like internet studies around the world, research on the debate between ‘states retaining power versus harnessing the internet for propaganda and control’ and ‘the internet empowering its users

[to] bring about pluralization and liberalization that challenge the state’ (Yang, 2009) is also prevalent among Chinese media studies scholars. Compared with Western countries, China has a long history of a dirigiste regulation system for domestic digital media (Bolsover, 2017; Qiu, 1999;

Yang, 2009; Zhao, 2008). Traditionally, the Chinese state has maintained a high level of control over information transmission within the country, mainly for ideological particularism (Bolsover,

2017), and all forms of media are officially considered to be a ‘mouthpiece’ of the Party and should therefore guide public opinion (Latham, 2000). This traditional set of values has been challenged by the dramatic changes wrought by digital media and new social relationships. Most notably, as a reaction, the functionality of media to benefit the Chinese state has changed over time, which provides a starting point for the discussion that follows.

From the 1990s, under the presidency of Jiang Zemin, ‘stress politics’ became a major concern for media control. New laws have been enacted and projects such as the Great Firewall have been launched for managing the new communication tools (Brady, 2017). In addition to political concerns, recent studies claim that the Chinese authorities have shown increasing interest in the economic value of digital media from the time of the presidency of Hu Jintao, and the core

Chinese policies on regulating digital media rely on a ‘two-hands strategy’, whereby the government wants to promote economic prosperity generated by digital media while keeping control of the flow of information via censorship (Lu & Weber, 2007, p. 929; Weber, 2005, p. 2;

Zhang, 2006, p. 9). This consideration of both market-driven forces and political concerns has

69

characterised China’s transition from a planned to a market economy (Zhao, 2008, p. 6). Although research has pointed out that the strict regulation hindered the economic development of China’s media (Brady, 2009; Donald & Keane, 2002; Zhao, 1998), it is also seen that less strict regulation is applied to non-ideology-oriented media platforms such as those for entertainment (Guo, 2014).

At the same time, public interest remains another enduring driving force behind the policy-making process, and sometimes it has outweighed the party’s value and interest for policy-makers during the period (Zhang, 2006).

From 2012, Xi’s administration shifted the focus of regulation again. Traditional media were told to focus on positive news stories and minimise negative reports, and restrictions on social media use were reinforced (Brady, 2017). Core socialist values10 promoted by Xi’s administration and related policies on the internet have shown an emphasis on superstructure, sometimes over the economic base, for building an integral state and maintaining long-term stability. A typical case is that the state tends to suppress social media content with the potential to trigger collective action without permission, regardless of whether or not those involved are critical or supportive of the state (King et al., 2013). For Xi’s administration, the cultural influence of media against competing discourses, both domestically and globally, is becoming a central evaluation criterion for regulating digital media (CAC, 2018a; Gow, 2017; Ho & Fung, 2016; Khudyakov, 2017;

SAPPRFT, 2017b). The “going out” strategy (Keane, 2016) and “cultural empowerment” in the

“Internet Plus” agenda (Keane, 2016) are other concerns, with China seeking to utilise digital media to achieve cultural impact both domestically and internationally. Consequently, ideological

10 Promoted at the 18th National Congress of the Communist Party of China in 2012, including: prosperity, democracy, civility, harmony, freedom, equality, justice, the rule of law, patriotism, dedication, integrity and friendliness. See https://www.chinadaily.com.cn/china/19thcpcnationalcongress/2017- 10/12/content_33160115.htm

70

and cultural governance has increasingly been enforced to mould online expression and behaviour

(Yang, 2014). Specifically, moral goodness, personal security and social stability are key priorities for internet governance at this stage (Cui & Wu, 2016).

Thus, the central policy rationales of the Chinese government towards regulating digital media content are characterized by a complex standard: on the premise of protecting stability and achieving economic goals, digital media have to fulfil the mission of promoting cultural construction based on core socialist values. This is different from Western digital media regulation, which has limited state intervention and it often restricted to issues such as offensive content, child safety, terrorism and privacy violations (Flew & Suzor, 2018, pp. 168–69). These complicated cultural and political economic standards pose challenges to legislation and law enforcement, especially when targeting unfamiliar and rapid changing industries such as live streaming.

Cultural Two-Hands Political Strategy construction Economic value concern Public interest Stability

Figure 15: Transition of state focus on media regulation

The dilemma for platform regulation

In Western digital media studies, there are growing concerns about the regulation of digital platform companies such as Google and Facebook. Platform governance has become an emerging research agenda for digital media regulation (Gorwa, 2019b). The power of the platforms has extended to societal in addition to purely economic influences, so there is a need for regulatory study to take both into account. However, the practical limitations mean the regulatory regimes currently prioritise users’ welfare from an economic perspective, not citizens’ wellbeing from a

71

social perspective, which is why content moderation continues to be an issue (van Dijck et al.,

2019). Existing studies have claimed that user generated content on digital platforms is usually immune to government intervention, particularity in a US context.

First, the First Amendment to the United States Constitution characterises the public sphere as an arena for neutral truth production (Cohen, 2017). Therefore, if the digital platforms are treated as part of this public sphere, it is neither justifiable nor legitimate to regulate the content on them from this perspective.

Second, section 230 of the Communications Decency Act (CDA), as part of the

Telecommunications Act of 1996, grants immunity to online intermediaries for their roles in distributing speech produced by others (Cohen, 2017). Therefore, the measures of moderating user-generated content on these online intermediaries come either from the wider government context or from the platforms themselves – and both are controversial.

Third, in addition to these two forms of immunity, platforms also use tactics to avoid responsibility and protect themselves. For example, the platforms always use detailed and very lengthy terms of services, which are almost impossible for users to read through before clicking

‘agree’; this is done in order to transfer accountability to the users (Cohen, 2017). Furthermore, platforms use the excuse of algorithmic models to avoid legal responsibility, by claiming that the algorithms are deemed neutral and objective, and this cannot easily be examined using the existing regulatory toolkit (Cohen, 2017).

Against this background, when platforms position themselves as neutral intermediaries, they do not need to take social responsibility and be regulated according to the existing law, despite the fact that they have significant social influence (Van Dijck et al., 2018). Another issue is that in the process of platformisation, platforms emerged as legal entities and they are empowered by self-

72

regulation (Cohen, 2017). Notably, platforms conduct editorial functionalities or self-regulation, including content moderation and user interdiction, usually for their own interests, such as

‘defusing public controversy and forestalling direct regulation’ (Cohen, 2017, p. 31); however, the moderation is usually justified by ‘cleaning’ or providing a ‘great user experience’. Essentially, it makes the platforms gatekeepers and means they are far from the neutral intermediaries they claim to be. The opaque and arbitrary self-regulation of platforms has given rise to questions of hidden agendas, double standards and biased points of view (Gillespie, 2018).

In short, despite the fact that the regulation of platforms has expanded from purely economic factors to more social factors, they can mostly escape the public responsibility associated with traditional media. In the meantime, they can also act as regulatory entities in their own right, shaping the media content just as traditional media do, but in alignment with their own interests and agendas. This inevitably becomes problematic as their private interests may frequently conflict with their obligations to society.

Platform regulation in the Chinese context

In terms of platform regulation, the United States prefers self-regulation using a more light-touch and reactive regulatory model, while the European model stresses the importance of the public interest and holds a positive view of state intervention, compared with the United States (Enli et al., 2019). In China, state intervention in digital platforms is prominent, as both the Chinese state and citizens do not position platforms as neutral intermediaries, nor do they treat algorithm as neutral object (Z. Zhang, 2020). While there are discussions about whether platforms should carry the obligation of public utilities in the Western context, in China the platforms are always responsible for the public interest and are treated as public utilities, as examined in Chapter 2 and

73

the previous sections of this chapter. Thus, avoiding legal responsibility is impossible for the platforms in China.

However, the top-down regulation model no longer seems to be effective in China. While the Chinese state has long held ambitions to modernize traditional industries and expand the ICT market, obstacles such as labour discontent, excessive competition and corporate resistance to policies are making it difficult (Hong, 2017a, pp. 71–76). Traditional state strategies of direct intervention in pricing and source allocation are being confronted; the Chinese state has had to compromise with different actors in the markets and local bureaus (Hong, 2017a, pp. 77–78).

Regarding digital media, the Chinese state not only functions as a regulator but also as a stakeholder in the market within global digital capitalism, in order to enhance the competitiveness of Chinese companies (Z. Zhang, 2020; Zhao, 2019, p. 12). Regional variations in policy-making are also an influential factor for the platforms’ regulation, as Enli et al. (2019) argue when analysing the interplay between digital platform and media policy in Northern Europe. As there are significant differences of regional economy and politics in China, analysis of platform regulation in China cannot ignore such regional variations.

Therefore, in order to conduct a nuanced analysis on the regulation on live streaming platforms, the following sections will analyse how the characteristics of live streaming matters from a regulatory perspective and also reflect on how the regulatory framework for live streaming has changed in recent years.

Live interaction and the imperative for new regulatory policies

A wide range of research concerning platform governance and online safety of the social media entertainment sites has been conducted over the past decade. However, as a specific category, live

74

streaming alone has not been given much attention by scholars. Few researchers have outlined the distinctiveness of live streaming or the need to develop specific regulation policies. We can reasonably ask a first set of questions: How is live streaming different from ordinary video-sharing websites and traditional sports/event live streaming? And how do these differences create new challenges for existing regulations?

Live interactions

The distinctiveness of live streaming hinges on communal live interactions (Nascimento et al.,

2014) – it enables synchronous streamer–spectator and spectator–spectator interactions rather than merely asynchronous interactions between audiences. Spectators can ‘talk’ to broadcasters by text or icons, and broadcasters can communicate responses to spectators by facial expression, talking or actions (Nascimento et al., 2014). These synchronous interactions in live streaming include discussing content with other audience members in live chat rooms, posting comments to streamers and donating money to favourite streamers.

Among live interactions, tipping/gifting (donation) is the most notable activity that drives all the live streaming practices. Tipping or gifting is a way for audiences to donate money to their favourite streamers. On Twitch, Tip is a feature button under the streaming screen. When audiences click it, they can donate their money to streamers via PayPal or by various other methods.

On most Chinese live streaming platforms, viewers use RMB to buy virtual gifts (with a price ranging from 0.1 to 500 RMB) from the platforms (Zhu et al., 2017, p. 276); viewers can then click a gifting button to send the virtual gifts to the broadcasters. After receiving considerable amounts of gifts, broadcasters can exchange them for RMB on the platform. Viewers usually tip streamers when they find content amusing or helpful. Tipping, in combination with comment posting, can affect reactions of streamers or even influence streaming narratives (Smith et al., 2013). As Smith

75

et al. (2013) note, ‘It is common to hear the performers answering questions from viewers and reacting to their suggestions.’

Hence, there are two different but simultaneous forms of content in live streaming. The first type is prepared by streamers to trigger live interactions – for example, gaming and cooking.

The script is prepared, and the form is similar to ordinary video on YouTube, except it is live streamed rather than recorded. The second type is the live interactions themselves, which include actions of viewers, reactions of streamers and unscripted streaming content initiated by live interactions, which are also being live streamed; they create reciprocal practices among streamers and viewers, as shown in Figure 16. The system of providing rewards through donations is the major commercial driving force for the live interactions.

Comment

Streaming Donation content

Reaction from streamers

Figure 16: Reciprocal practices of live streaming

However, this rewarding system can easily trigger social problems, as some streamers tend to entertain audiences unscrupulously (typically with sexy dressing or talking, or even abhorrent materials) to earn more tips, and viewers also use tips and comments to lure streamers (Lu et al.,

2018). Compared with ordinary video websites, these reciprocal practices can become serious

76

when eye-catching content dominates, thus generating more profit for streamers and platforms directly (Liu, 2016). The problem of vulgar, erotic and misleading content on live streaming platforms is challenging regulators.

Community Building and Collective Action

Another feature of live streaming video that matters with regard to regulation is that live interactions can trigger different forms of connective activity and facilitate community-building. Martini (2018b, p. 15) compares online video websites and live streaming websites, and claims that, ‘The connective activity generated by online videos is a long-term and low-intensity form of engagement, characterized mostly by a high level of passive sharing and a minimal volume of reflexive activity and user–producer interaction’, while the activity generated by live streaming is ‘high-intensity user engagement’ and includes ‘significant levels of reflexive activity and user–producer interaction’ (Martini, 2018b, p. 15).

This also echoes Powazek’s finding that live synchronous action has a stronger emotional impact

(Powazek, 2002, pp. 11–13). In China, this new mode of high-intensity engagement, together with the fan groups of streamers on instant messaging apps, further affords social interactions between viewers and streamers, and has built communities with large memberships around streamers of both online and offline varieties (Lu et al., 2018). On this basis, purported detrimental effects of ‘inappropriate content’ could be exaggerated, but collective action could also easily be triggered if there was conflict. It is these scenarios of a potential threat to social stability that concern the government (CAC, 2018b; Gao, 2018).

In short, the transmission of illegal content that can potentially threaten social stability concerns the government. When the Chinese state realized the need for regulating streaming content (Wang &

Linyuan, 2017), it quickly started to reinforce restrictions on this new industry. However, new problems can be seen to arise when traditional ways of regulation are not applicable to live streaming.

77

Regulation of live streaming in the United States: Twitch and Facebook Live

From the Philando Castile shooting case in the United States to the Christchurch Massacre in New

Zealand, in the Western context such live streaming videos are regarded as inherently unpredictable and difficult to control for the content moderators, as the content is shared and uploaded simultaneously (Lomas, 2017). Twitch, Facebook Live and YouTube Live are dominating the market in Western countries (Khan, 2019; Yosilewitz, 2019). Moreover, as US companies they are mostly immune to the governmental intervention and thus enjoy a higher level of autonomy. These platforms have the right to enact and implement their own rules, as long as the content does not involve terrorism, child pornography and copyright infringement (Flew &

Suzor, 2018, p. 168). Therefore, it is clear that self-regulation of the platforms is the dominant mode for the content regulation on live streaming in the United States. Platforms employ vast numbers of content moderators and develop rules for content filtering by themselves. However, the result of self-regulation is not currently satisfactory, and is still the subject of criticism: it has not fully prevented illegal content from circulating, but it has also censored legitimate free speech at times (Gillespie, 2018; Poell, 2020).

Twitch

Platform regulation of Twitch for streamers and viewers

Twitch features game streaming and streaming content generated by streamers, which is relatively predictable compared with other forms of content, such as live shows. In addition, as Twitch was one of the earliest live streaming platforms in Western countries, it is very experienced in content moderation. Most notably, Twitch achieved a monopoly position without competitors in Western game streaming, as examined in Chapter 2. Understandably, then, Twitch had no hesitation about

78

impose tough restrictions, given that there are no alternative streaming platforms onto which gamer streamers can ‘escape’.

There are two key sets of regulation for the content moderator of Twitch: terms of service and/or community guidelines. In the terms of service, Twitch outlines the kind of content that is inappropriate, such as unlawful, copyright infringing and pornographic content, although the criteria are not detailed. From Article 9 ‘Prohibited Conduct’ and Article 15 ‘Disputes’, a key point is that the users are solely responsible for their conduct and Twitch is not liable for any user content or any loss or damage resulting therefrom. Thus, users take all the responsibility and risk of prohibited content (Twitch, 2019).

In addition to terms of service, Twitch’s community guidelines provide another framework that is updated more frequently and with more details compared with the terms of service. Twitch’s community guidelines define inappropriate or harmful content, including law violations, suspension evasion, 11 harmful content, harassment and pornography. In addition, as Twitch features in game streaming, cheating and disclosed officially unreleased game content are also prohibited (Twitch, 2020d).

For enforcing these rules and punishing the violators, inappropriate content is to be removed and the user’s account might be suspended. As announced by Twitch, it reserves the right to suspend or ban any account at any time for any conduct that Twitch determines inappropriate

(Twitch, 2020d). There are three different forms of punishment, depending on the violation:

11 Guidelines violation leads to the suspension of an account, and any attempt to circumvent an account suspension will also lead to suspension and will increase the length of suspension.

79

warning, temporary suspension and indefinite suspension. Those who are suspended cannot access all the Twitch services, including viewing, broadcasting and chatting (Twitch, 2020a).

Channel regulation from the communities

The management of specific channels is handed to the users instead of the official platform moderator. Twitch has various ways to let streamers regulate the viewer-generated content. In addition to the common functions such as banning and reporting users, the following section highlights three categories of functions that are exclusive to Twitch-like live streaming platforms.

First, the viewer community can be enlisted to help by becoming channel moderators.

Streamers can ban unwanted viewers and delete the inappropriate chat messages, but they do not need to handle all these by themselves. A typical way is to appoint official channel moderators, which is similar to the Chinese room regulators (discussed later). These moderators are usually the active members from the viewer community. When they are appointed as channel moderators, they have the power to regulate the content of the channel, including setting chat modes and timing out or banning viewers. Streamers can also deploy moderation bots to do the preliminary moderation jobs for filtering, so the bots can reduce the workload of manual moderation of appointed channel moderators. These bots are developed by third parties – usually Twitch community members instead of Twitch developers – but Twitch approves and encourages the use of bots for additional moderation (Help.Twitch, 2020a).

Another typical way for streamers to reduce the manual workload of moderation is to use automatic filters. For example, the Chat Filter system developed by Twitch can block unwanted content automatically. Streamers can set Chat Filter preferences to default settings to moderate inappropriate content such as discrimination, hostility and sexual language. Streamers can also personalise the keyword database by using the channel-blocked terms filter. In this filter, streamers

80

can add terms based on their demand, and the channel-blocked terms filter can block any words related to the terms in this channel. Wildcard functionality can also be applied for this filter so all variations of the terms will be detected and banned (Help.Twitch, 2020a). In these ways, users of

Twitch, particularly active community members, become volunteer moderators, which turns out to be effective and reduces the workload of Twitch staff.

Second, streamers can mute or block certain groups of viewers. The streamer of the channel can block specific users from entering the channel or from posting any comment. The block can range from a few minutes ban to indefinite timeout, depending on the nature of the conduct. In addition, streamers can also use stratification methods to classify viewers and grant them different levels of access to communication functions. For example, when a streamer enables Sub Only

Mode, only subscribers and the appointed moderators of the channel can chat, and all other messages from other viewers would be prevented. Another similar mode is called Follower Mode, which only allows followers to join the chat. In this mode, streamers can set the barrier based on how long viewers have followed the streamers – from zero minutes to three months. Similarly, the

Email Verified Mode requires all viewer to be email verified before chatting (Help.Twitch, 2020a).

No external moderation is needed in these modes – the viewership or communication from unwanted strangers is controlled in the first place, and only the long-time community members have full access to the communication functions. Although the issue of speech rights and discrimination could be raised here, so far it has been an effective way to guarantee the content as it guarantees the quality of viewers.

Third, an innovative and distinct way to achieve content moderation in live streaming is a chat delay system, which is developed to hold user-generated content, giving moderators a chance to review the content before it becomes publicly available. The channel owner can set a non-

81

moderator chat delay, which delays all the messages for a few seconds, instead of them being displayed instantly. As a result, streamers or the appointed moderators have time to review the messages and they can stop harassment from being shown on the chat room. Moreover, streamers can also enable AutoMod, which uses machine learning and NLP algorithms to hold risky messages, preventing them from displaying instantly on the chat room. In this mode, moderators only need to focus on the suspicious content, and they can approve or deny any message that is held by

AutoMod. In this way, the human workload is reduced, and the appropriable content does not need to be delayed.

In short, we can see that Twitch has done good job at moderating the streaming content, partly because it is experienced, and partly because the game streaming content and game community are easily managed. Another key reason that cannot be omitted is that Twitch’s monopoly gives it power to enforce any forms of restrictions, regardless of any complaints from users.

Facebook Live

Facebook generates much more content than Twitch, and its live streaming service, Facebook Live, provides various forms of content that are not limited to gaming. The content is even more unpredictable if it is streamed outdoors. In 2017, CEO Zuckerberg announced that Facebook would add 3000 more reviewers for the content regulation after violent content had been streamed on

Facebook Live, especially after a footage of an 11-year-old girl being killed by her father was broadcast (S. Gibbs, 2017). In 2017, Wuttisan Wongtalay killed his daughter before killing himself in Thailand, live streaming the process in two sets of footage. The two clips were accessible to users on Facebook for approximately 24 hours and have been viewed more than 100,000 times.

Moreover, the videos were reposted on YouTube by viewers. It was impossible to hold Facebook

82

criminally liable because the platform as a service provider had acted according to its protocol in

2017 (Samuel Gibbs, 2017).

However, the implementation of tougher forms of self-regulation did not improve the situation. The widespread viral viewing of the Christchurch mosque massacre video in 2019 indicated that content moderation of Facebook Live was still inefficient. In the case of Facebook

Live, the need for improvement of moderation is even more urgent as it is closely connected to

SNS, allowing this and similar videos to be easily shared, downloaded, re-edited and re-uploaded by other users. In addition, excessive moderation and freedom of speech are another critical concern for the development of content moderation (Filtering Practices of Social Media Platforms

(115th Cong), 2018). For example, Diamond and Silk has been censored by Facebook because it was found to be ‘unsafe to the community’, according to the moderator. The moderation rule in the case is a mystery without enough transparency, according to research by Chasmar (2018; see also Filtering Practices of Social Media Platforms (115th Cong), 2018).

Under pressure to modify its filtering and moderation practices, Facebook is consistently working on improving self-regulation. Compared with Twitch, the Facebook standards are richer in detail – the community standards alone have five categories and 26 clauses, and they clearly identify offensive content (Facebook, 2020b, 2020d). Moreover, Facebook regularly releases a

Community Standards Enforcement Report,12 transparently and comprehensively showing the actions taken against violating content. Clearly, Facebook’s long-term operation and moderation experience have led to this high level of clarification and transparency, thereby improving the efficiency of video moderation. In terms of live streaming regulation, there is no exclusive standard

12 See https://transparency.facebook.com/community-standards-enforcement

83

for Facebook Live yet, and most regulatory measures are devised for its social media features, while the community standards and terms of service of Facebook Live are the same as for

Facebook’s other services. There is only one distinction for Facebook Live: ‘For better regulating live videos, “one strike” policy is also applied to Facebook Live and the violators would be banned for a period of time if they violate the policies.’ Facebook VP of Integrity Guy Rosen also claimed that Facebook was investing in image and video recognition technology for the moderation of live videos (Rosen, 2019).

Facebook Live does have channel regulation from its communities: there are modes that only allow long-term trustworthy community members to post comments. They could be followers, registered accounts or long-time followers (Live, 2020). Streamers can also add community managers to help moderate the comments during and streaming. These two moderation measures are the same as those used by Twitch.

Contemporary issues of live streaming regulation on Western platforms

Given the above analysis, self-regulation by the platform and from user communities can be considered to be the major means of regulating live streaming content in the United States.

However, relying on self-regulation can be problematic in certain contexts. First, inconsistent enforcement is a serious problem on Western platforms. The exercise of discretion rather than the rule of law is the common practice in self-regulation, without clearly articulated terms and conditions (Common, 2020). Second, the infringement of rights of speech is prevalent, and is an issue that can easily escalate when the moderation is carried out by the untrained community moderators – particularity when the streamers rather than the community moderators should be responsible for the moderation. Third, imposing legal liability on platforms could increase their

84

costs, causing local platforms to move to other countries with fewer restriction (Communications,

2019). It could also lead to monopoly if the small platforms become less competitive because they are less experienced in dealing with legal issues, while Big Tech has already built up a comprehensive self-regulation system. Fourth, a recent report shows that the current moderation processes of platforms are opaque and slow (House of Lords Select Committee on

Communications, 2019).

It is tricky to find solutions to these issues, but if we look to Asia, we can see that the

Chinese regulatory system is also facing similar issues brought by live streaming. We can ask several questions: What kind of challenges does live streaming pose to existing regulatory systems in China, particularly given the distinct political, economic and cultural heritage of the nation?

How has the regulatory framework in China been adjusted over time to accommodate live streaming? Have these strategies turned out to be effective or not? What can we learn from these regulatory experiences in China? These questions animate the following sections of the thesis.

Challenges to regulation by the state in China

An overwhelming workload for real time monitoring?

In previous decades, Chinese media review processes for media governance included pre- production topic approval and post-production censorship (Zhao, 2008, p. 27). The script-based part of the content was predictable, but a core part of live interactions in live-streaming is unpredictable, as even the streamers themselves do not know what they will encounter during the broadcasting processes. As a result, pre-production review as a typical way of censorship for the content is impossible for live streaming. In addition, most streaming videos are not archived, making post-production censorship difficult. In this situation, real-time monitoring has become the

85

most efficient way to supervise live streaming content. Using digital tools to automatically reduce manual workload is a typical way to facilitate content censorship. For example, the Golden Shield

Project is aimed at automatically monitoring online user-generated content to make sure the content is legal and proper (Bolsover, 2017). However, while computational monitoring works well for text-based content, it is not good for video-based content according to Huan in Wang and

Linyuan (2017). Manual monitoring is thus still the primary method for video content in China.

There are hundreds of streaming platforms, with tens of millions of users every day in

China (CNNIC, 2019). From a micro perspective, it is easy to become a streamer as no qualification (unless news streaming) or training is required – anyone can start streaming with a mobile phone. From a macro perspective, even with tight control, there are still high numbers of streaming platforms in China. This is partly due to the Chinese government’s ‘two-hand’ strategy to maximise the economic benefit of the media, which aims to prioritise support for the development of digital technology and commercialisation of the industry. The significant growth in domestic mobile devices industries (Lu & Weber, 2007, p. 932) and advanced IT infrastructure

(Liu, 2016, pp. 1–2), together with governmental support to commercial media (Weber, 2005, p. 793) and cultural industries (Ho & Fung, 2016, pp. 107–08), contribute to the prosperity of live streaming, while providing fertile ground for speculative streaming platforms. The number of new live streamers and platforms is rising exponentially, and so is the streaming content for regulation.

Just who should take the main responsibility for the content regulation remains a contentious issue. There has been wide discussion about whether the platform should have liability for the user-generated content (Hong & Xu, 2019). However, in China the platforms are treated as public utilities and are usually responsible for the content published on them, as we have discussed.

86

Therefore, platforms now have to take the responsibility for manual real-time monitoring, which is too costly for them given the huge number of channels.

Conflict between government agencies

In China, the Central Publicity Department is in charge of all ideology, media and public opinion- related issues, while the specific industry supervision work is carried out by different governmental agencies. It is notable that the Chinese government does not work as a monolithic entity, and the reality is that different agencies, leaders or levels of government work for various purposes (King et al., 2013, p. 3). Branches of the state, administrations and courts do not always act in consistent ways, showing that the state is fragmented – as is platform governance itself (Hong & Xu, 2019).

In addition, the turf war between regulatory ministries also brought challenges for more consistent digital media regulation (Zhao, 2013). It is therefore necessary to interpret the Chinese state actions

‘as a combination of contending functions and interests rather than a uniform entity’ when studying the regulatory functions (Hong, 2017a, p. 10).

In the case of live streaming regulation, there are at least five governmental departments regulating live-streaming at the same time, but none of them regards live-streaming as its major focus. Liu (2016) provides an account of the supervisory perspectives of different authorities in

China. From her framework, she has developed the chart shown in Figure 17 and discussion, which correlates the functionality of these agencies and their relations with live streaming legislation, pointing out the conflicts of different governmental agencies on regulation.

87

Figure 17: Functionality of administrations on live streaming

Ministry of CAC NRTA14 (Formerly Cyber Police MIIT Culture13 SAPPRFT)

Objective Evaluate Internet Regulate audio- Stop illegal Internet Cultural governance of visual content online conduct management implication of information and business content and security; real activities censor name inappropriate authentication content

Functionality Regulate Supervise Regulate streaming Supervising Supervise platform streaming content; Licensing major streaming operators and content; manage management platforms & content; VPN streamers; real ID streamers; use; ICP licensing registration Control cyber licensing management system; news crime management licensing management

Legislation Administrative Internet Live Notice on Cybersecurity Internet Measures for Service Strengthening the Law of the Information the Operation Management Management of People’s Service of Network Regulations; Web-based Live Republic of Management Performance Internet News Program Service China Measures Information (existing) (existing) Service Management Regulations

Although all these agencies are responsible for the regulation of live streaming, no

significant actions have been taken at the time of writing, and no legislation was produced for live-

streaming until 2016. In addition, their regulatory responsibilities are vague and overlap, with even

13 In 2018, the Ministry of Culture and National Tourism Administration were merged into the Ministry of Culture and Tourism, subordinate to State Council (Xinhua, 2018). 14 The National Radio and Television Administration (NRTA) was established in 2018, subordinated to the Communist Party of China Central Committee’s Publicity Department. Formerly it was the State Administration of Radio, Film, and Television (SARFT, 1998–2013) and the State Administration of Press, Publication, Radio, Film and Television (SAPPRFT, 2013–18); both were departments under the State Council.

88

the policy-makers themselves being frustrated by such complicated relationships (Zhang, 2006).

Except for the CAC, which established the real ID and blacklist systems, and NRTA, which implemented qualification examination, all five departments are performing a generally similar role – monitoring streaming content, and controlling and enforcing rules against illegal platforms and streamers. However, there is no effective cooperation between these agencies, such as sharing work assignments to reduce workload. The majority of these agencies work together on censoring streaming content. It is not surprising, then, that supervisors from the CAC waste substantial time supervising content from unlicensed platforms, which supposedly have already been shut down by the license reviewer NRTA.

Moreover, despite the necessity of regulation emphasised by central government, local governments have to take any economic contribution into consideration when regulating live streaming. A study of Tomasic and Fu (2005) has shown that when there are connections between companies and local government, the rules from the state cannot easily be implemented because the local government tends to protect the companies when violations occur (Tomasic & Fu, 2005, p. 26). This could also be the case in the live streaming industry. According to one report, the revenue from live streaming reached 30.45 billion yuan (US$4.76 billion) in China in 2017, an increase of 39 per cent from 2016 (Lifang, 2018). For example, the report notes that Douyu made a significant contribution to the local economy of Wuhan, with tax amounting to 120 million RMB

(Yu, 2018) and through the creation of more than 1700 jobs directly, and a further 100,000 jobs indirectly (Yangyu, 2017). In Guangzhou, Huya is the first Chinese streaming platform listed on the New York Stock Exchange; its total net revenues for the first quarter of 2018 were

US$134.5 million, with average monthly active users being 92.9 million (Huya, 2018).

89

Rather than suppressing the media, the line set by Party leaders is to control social media within acceptable boundaries that do not harm the party’s core values and the stability of society

(Brady, 2017). It is common for central authorities not to specify the requirements, and instead let the local regulatory agencies formulate detailed terms for the regulation (E. Zhao, 2017, p. 119), and the local agencies are reluctant to enforce the regulation if it contradicts local job creation and economic development (2017, p. 114). On this premise, the media filtering approach is totally conducted by local authorities and organisations, which is regarded as a decentralised and semi- privatised operation (Wright, 2014). Thus, the local governments are glad to see any forms of content that could maximise job creation and the profits of their local media companies, within the line set by Party leaders. Local government agencies tend to take the economic value of live streaming more seriously than central government, and perhaps also take a more moderate approach to governance of local live stream platforms.

Because of the multiple stakeholders, the regulatory system in China is inefficient for daily supervision, and the law enforcement agencies only cooperate to launch massive campaigns when an issue becomes serious (Liu & Xu, 2018). As King et al. (2013, p. 3) argue, different agencies and their leaders, or different levels of government in China, work for different purposes, and

‘branches of the state, including administration, the courts, and the legislators, do not always act in consistent ways’ (Hong, 2019). As a result, they do not reach a consensus on how to regulate live streaming, rendering the policy-making and implementation of regulation even more frustrating from a number of perspectives.

90

Outdated legislation, vague policy mandates and inefficient penalties

In terms of legislation, the criteria for recognising what needs to be regulated are often vaguely defined for streaming. This vague policy mandate for regulating media content creates a large grey area between what is legal and what is not, making the criteria subjective and the implementation of regulation inconsistent. This may also be explained partly because of the diffuse cultural construction of ‘moral goodness’ promoted by the government is inherently ambiguous for the implementation of specific regulation. Moreover, as China currently has no elaborated classification system for video content, existing standards cannot be a reference for regulating streaming content.

As a result, ambiguous mandates generally hinder manual censorship. For example, Custer

(2016) notes that while previously many female gamers would dress sexily to attract viewers while playing games, as showing cleavage is not illegal and streaming game play is legitimate, the combination of these two can hardly be defined as ‘pornography and obscenity’ (Article 9 of

Internet Live Service Management Regulations proposed by CAC). Flirtatious conversations are sometimes interwoven with major streaming content, which also does not cross any regulatory line.

Indeed, these kinds of live shows did contain sexual elements which were deemed to attract audiences in a ‘inappropriate way’ and may not be ‘healthy for youngsters’ according to Article 3 of the Internet Live Service Management Regulations proposed by CAC (CAC, 2017). Yet according to the regulations, the CAC and other agencies may not censor this content (Custer,

2016).

In addition, outdated laws also create loopholes for illegal platforms and streamers to evade enforcement. Indeed, some platforms have tried to avoid punishment and minimise the penalty by themselves making use of loopholes in the legislation. While licences from NRTA, MIIT and the

91

Ministry of Culture 15 are necessary for running streaming websites, some notorious and unlicensed companies just focus on developing their mobile apps, which have allowed them to easily evade supervision. These illegal apps frequently changed app names to avoid getting caught.

If they are banned by government or the app store, they have been known to simply change an app name and icon before returning to the app store (Fang, 2017). In addition, the penalties for illegal acts are relatively small. According to Lu, ‘Some popular platforms can earn more than one million

Chinese Yuan per night’, but if they do not cause a severely detrimental effect, the maximum fine according to the law is just 30,000 RMB. Lu also makes the observation that the meaning of what is ‘severe’ is not clearly defined in the regulations (Lu, 2016). For individual streamers, some speculators realise the inefficiency of real time supervision, so they are willing to broadcast a few sections of risky content in a long streaming video so that they can lure viewers while minimising the chance of being caught by human monitors (Jingyan, 2016). A fraction of illegal content during a long-term streaming can hardly be detected by monitors, and even if the streamers are caught and fined, they can still benefit from the huge amount of revenue (Lu, 2016). Thus, the inefficiencies in applying penalties are making live streaming platform governance extremely difficult for regulators.

Although the regulations for live-streaming have been revised in recent years, little change has been made to reduce adult content and illegal platform operations (Liu, 2016; Wang & Linyuan, 2017).

An overwhelmingly high workload for monitors, conflict between government agencies and outdated legislation all work to hinder live stream governance, demonstrating the inefficiency of Chinese

15 Ministry of Culture and Bureau of Tourism have been merged into Ministry of Culture and Tourism (MCT) from 2018, see https://www.mct.gov.cn/whzx/zsdw/whbxxzx_bnsj

92

government regulatory interventions around live streaming. Does this inefficiency of governance mean the power of the state has been weakened and the top-down mode is no longer effective?

After studying the regulation on online video in China, Li (2019) provides an additional interpretation. She argues that we cannot merely regard the inefficiency of regulating digital media as a failure of the Chinese regulatory system; rather, the online video industry can be categorised as a grey zone of tolerance configured by the Chinese state. The state is not only incapable of imposing strict and comprehensive regulations, but also unwilling to do so, considering the economic contribution made by the industry (Li, 2019, pp. 41–43, 109). As Ho (2001) points out, the ambiguity of the regulation might be deliberate, as it can facilitate social and economic development and creates flexibility for policy-makers when dealing with fast-growing industries.

So does live streaming fall into the same context, or is it a peculiar industry for regulation? Before drawing any quick arbitrary conclusions, it is necessary to consider the changing strategies of the

Chinese government, and the major transitions in the live streaming industry that have been underway for several years.

The evolving regulatory strategies

According to Baidu Zhishu, 16 the interest in Wangluozhibo (live streaming) increased dramatically at the end of 2015 and peaked in 2016, which is when live streaming became a significant concern for regulatory agencies. In April 2016, the Ministry of Culture proposed a new agenda for streamer authorisation and content regulation, announcing an investigation into illegal behaviour on major streaming platforms in China. Then, in September 2016, SARFT started to

16 Baidu Zhishu is the Chinese version of Google Trends. See https://index.baidu.com/?tpl=trend&type=0&area=0&time=13&word=%CD%F8%C2%E7%D6%B1%B2%A5

93

release official documents on regulating live streaming content, issuing a Notice on Strengthening the Management of Web-based Live Program Service (关于加强网络视听节目直播服务管理有

关问题的通知) (SAPPRFT, 2016). On 4 November 2016, CAC enacted the Internet Live Service

Management Regulations (互联网直播服务管理规定), the first and the only legislative document that specifically outlined the regulation policy on the live-streaming industry (CAC, 2016). As the first systematic rules for live streaming, the Internet Live Service Management Regulations established processes for news control, real-time monitoring and real-name authentication as the three major methods to regulate live streaming in China.

Furthermore, on 12 December 2016, the Ministry of Culture released the Administrative

Measures for the Operation of Network Performance (网络表演经营活动管理办法) to strengthen the regulation of live performances, based on existing regulations specified by other departments

(Ministry of MCT, 2016b). In addition to these three legal documents, CAC released the revised version of the Internet News Information Service Management Regulations on 2 May 2017 (CAC,

2017), based on the current digital environment for the purpose of controlling news content. Then, in March, 2018, SAPPRFT (2018a) issued the Notice on Further Regulating the Transmission

Order of Internet Audio-Visual Programs, which requires platforms and broadcasters not to re-edit, distort, parody or denigrate classic works, radio and television programs, or original internet audio- visual programs without authorisation. Figure 18 provides a brief summary of these five major documents.

94

Notice on Strengthening The Management of Web-based Live Program Service (SARFT)

Requires all streaming platforms to apply the ‘License for Online Transmission of Audio/Video Programs’ from SARFT

Internet Live Service Management Regulations(CAC) Require qualification for news Platform should build up regulating Regulating live comments Real name authentication services system

Administrative Measures for The Operation of Network Performance (Ministry Of Culture)

Reaffirmed qualification application Reaffirmed real name authentication Archive streaming content for review Reinforce content filtering

Internet News Information Service Management Regulations (CAC)

Reaffirmed qualification for news services Define news services and speficy regulation

Notice on Further Regulating the Transmission Order of Internet Audio-Visual Programs (SAPPRFT)

Platforms and broadcasters should not re-edit, distort, parody or denigrate classic works without authorization

Figure 18: Evolving national policies on live streaming in China

The core intention of these policies is to force the live streaming platforms to take up self- regulation, and the role of government has changed from direct intervention to a subtle control in order to ensure that the self-regulation of the platforms is guided on the right track. Data collected in my interviews for this research also support the idea that the Chinese government has now become an inspector rather than a direct regulator, tending to let the industrial stakeholders conduct self-regulation. The next section will analyse the details of these changing policies, recent regulatory cases and the data collected from my interviews focusing on live streaming regulation, in order to examine the ways in which the current regulatory framework of live streaming features self-regulation, and how this self-regulation is then conducted by various stakeholders across different levels.

95

A central strategy: Forcing the platforms to take up self-regulation

In terms of digital media, applying distinct rules for different media ‘in respect of content policies, ownership, copyright and intellectual property’ is necessary for the state and key stakeholders

(Flew et al., 2016, p. 11). Indeed, governments globally have increasingly relied on responsible self-regulation as a complementary method to direct control, due to the substantial scale and speed of interactions on digital media (Flew, 2015, p. 2). As policies in China for live streaming shows have evolved, the state has made distinct rules for the normally uncontrollable live interactions: self-regulation has become the central measure, replacing direct state control.

Historically, self-regulation has been a requirement for all internet companies in China, obliging them to comply with government policies and to carefully scrutinise their own content in order to stay in business while avoiding enforcement of penalties (Liang & Lu, 2010; Weber &

Jia, 2007). The idea of self-regulation has become the central strategy for regulating live streaming

– that is, a system of self-regulation has been established, permitting key agencies in the live streaming industry to oversee these processes. In the live streaming sector, self-regulation has become quite structured, and now includes a range of key stakeholders, including industry associations, platforms, agencies and user communities. All these stakeholders are working at different levels to fulfil their regulatory missions. The data collected from the my interviews with stakeholders and associated documentation is now presented to show how this self-regulation works in detail.

Stakeholder interviews

To explore the complex relationships between different state agencies and the power structure in

China’s live streaming industries, stakeholders – including individual streamers, staff from

96

streaming platforms, talent agent companies and representatives from government agencies – were interviewed. These interviews were conducted over a period of approximately 20 months, from late 2018 to 2020. The interviews were conducted in person or by telephone, were semi-structured, and each lasted about 20–50 minutes. While this research planned to conduct face-to-face interviews for all participants, several issues occurred during the process: some participants who had accepted the interview chose to switch to phone interviews because of their schedule and their concerns of privacy; some interviews were conducted during the COVID-19 restrictions. Thus, the interviews of these participants could only be conducted via phone.

The interview questions focus mainly on the participants’ personal experience in live streaming activities, their relationship with the other stakeholders in the industry, the regulation of live streaming activities and their general understanding of the industry. All the information is de- identifiable, with the interviewees being tagged with unique identifying numbers in this thesis to protect their privacy, given the highly sensitive nature of the data. The selection criteria and characteristics of final respondents that accepted our interviews are summarised below:

• Streamers. This research used a combination of accessibility sampling and purposive

sampling. Accessibility and proximity of the participants to the researcher were the main

concerns for the accessibility sampling, so this kind of sampling has cost and administrative

benefits (Greenfield & Greener, 2016, p. 248). The major reason for adopting accessibility

sampling was the inherently sensitive nature of the research. Amateur streamers tend to

keep their identities secret and are typically reluctant to share any detailed information on

the publication even with the protection of anonymity. Accessibility sampling was

therefore the best way to generate enough participants and materials.

Bias can be a drawback of accessibility sampling, but purposive sampling can help

97

to reduce this bias (Greenfield & Greener, 2016, p. 248). Purposive sampling can be used

when the researcher believes there is inherent variations in the population of the research

objects, thereby controlling the variations by subjective judgement to select the most

‘representative’ samples, which can significantly reduce the bias of accessibility sampling

(Greenfield & Greener, 2016, pp. 248–49). In this research, I balanced the accessibility

sampling with purposive sampling by considering the categories of streaming content (all

three categories, including live shows, game streaming and e-commerce streaming, were

covered), and the popularity of the streaming channels; therefore the samples can be more

highly representative. I identified streamers through social media, agent companies and

live streaming platforms. Contact details of the streamers were acquired from their

webpages of streaming channels and social media profiles. Finally, eight streamers were

interviewed. Five of them were live show streamers, two were game streamers and one was

an e-commerce streamer. They are labelled S1–S8 in this thesis.

• Staff from talent agent companies and live streaming platforms. I obtained the contact

details via social media or the official websites of the companies. Two platform staff and

two staff from agent companies were interviewed. For the purpose of viewpoint diversity,

these four interviewees were drawn from different companies. They are labelled A1, A2

and P1, P2.

• Official regulators. To interview officials in regulation departments, I obtained their

contact details via the official websites, social media and personal networks. I contacted

several officials but most did not agree the interview. They said the information in response

to my questions might be confidential and it was too risky for them to expose it. Others

agreed to talk to me, but required that I keep the information secret and not use it in any

98

publication because of the confidentiality agreement they had signed as governmental

officers. Therefore, collecting information directly from government officials was

impossible. As an alternative, I contacted scholars who were working with the regulatory

officials in collaborative projects. Two scholars were interviewed who had previously

collaborated with regulatory officials, and with whom they maintained good relationships.

These researchers are conducting a live streaming platform regulation project with a local

regulatory agency in China. They provided substantial materials, which have been helpful

in understanding the regulatory policies from the perspective of the officials. These two

scholars are labelled R1 and R2 for anonymity. In addition, one platform employee (P2)

that I interviewed had also previously worked as a ‘content supervisor’ for the

governmental agency. This interviewee was able to provide additional information from

the perspective of a regulatory official.

First-tier self-regulation: Self-regulation through industry associations

Industry associations have been formed by streamers or platforms for the purposes of management of the self-regulatory ecosystem. These associations aim to provide legal and commercial consultancy services and to generally build a better streaming environment for the sector. On

13 April 2016, more than 20 live streaming companies from the Beijing Internet Culture-

Association gathered and issued a Self-discipline Convention (Culture-Association, 2017), with several requirements:

• All streamers shall authenticate their real name: providing name, ID number, phone number,

and personal bank account information. Moreover, streamers should provide a photo for

processing their ID certificates (including a clear digital ID image and number). In addition,

streamers are required to have a video chat with their supervisor in order to identify and

99

verify their real name ID.

• All streaming content should have watermarking with website logos and time-stamped for

recognition.

• All streaming content should be archived for 15 days for further review.

All these participating streaming platforms forged a consensus and agreed to commit to the convention. A year later, on 8 July 2017, 18 streaming platforms, including YY and Huajiao, founded the Live-streaming Industry Self-Discipline Union, which has built a united reporting platform to receive public scrutiny and to establish a jointly agreed blacklist system to filter illegal streamers (Wu, 2017).

Because of the prosperity of e-commerce live streaming, the China Advertising Association enacted a Code of Conduct for E-commerce Live Streaming specifically for regulating e- commerce live streaming content and trade on 24 June 2020. It clearly outlines the rights and responsibilities of merchants, streamers and platforms. Notably, it forbids traffic fraud and paid posters, and proposes to build a credit rating system for both streamers and the merchant, which is a step ahead of the existing law (CAA, 2020).

Second-tier self-regulation: Self-regulation from the live streaming platforms

A super regulator: The real-time management team with detailed rules

Self-regulation by the platforms is the main measure enforced by the Chinese state from 2016. As the amount of streaming content has been increasing dramatically, the central government has realised the pressing need to build a larger censorship team for live streaming. But, rather than expanding its own workforce of human censors, the Chinese government began in 2016 to outsource the supervision workload to platforms, forcing live streaming companies to build their

100

own censorship teams. To this end, on 4 November 2016, the CAC enacted the Internet Live

Service Management Regulations, requiring all platforms to build a real-time management team for content scrutiny (CAC, 2016). Rather than using ambiguous criteria such as moral goodness and social orderliness, platforms were required to categorise live streaming channels based on the details of content and user amount, and to develop specific a regulation code based on different categories under Article 7 of the Internet Live Service Management Regulations (CAC, 2016). The establishment of this regulatory frame was confirmed by R1, who noted that the platform code of regulation was very rich in detail, and listed all the obligations and potential punishment for both streamers and agents who were found in breach of the code requirements.

Following the issuing of these regulations, companies such as YY, Douyu and Inke built teams ranging from hundreds to thousands of staff, all sitting in front of their computers to monitor live channels on their platforms. Each supervisor is responsible for monitoring about 30 channels simultaneously, checking that no offending content is shown (BBC, 2018a). These supervisors have been dubbed ‘super regulators’, and they are employed by platforms specifically to filter streaming content and to catch any offending streamers and viewers. Under the regulations, all forms of content, including video, audio, photos and text, are to be monitored by supervisory teams.

P1 noted that the platform he was employed by had a real-time monitor team to supervise the streaming content 24/7:

There is a team, with hundreds of real-time monitors to watch over all the channels

24/7. Each monitor takes care of tens of channels, and they do not have to look at

all the channels all the time – with the assistance of image and voice recognition

technologies, the potentially inappropriate content can be highlighted and the

monitors can stay focus on these channels. (Interviewee, P1, 31/12/2018)

101

As a support for the supervision teams, and as required by the government, platforms also implement their own regulatory code of practice based on government policies, making the rules clearer and easier for supervisory team to enforce. For instance, there are two levels of offending content according to the Management Regulations for Live Stream Content on Douyu:

A. Serious offences relating to antigovernment or criminal content; and

B. Normal offences relating to violent and copyright infringement content.

For normal offences, streamers will be banned from the platform for a period and fined. For serious offences, streamers could be permanently banned and referred to the police (Han & Shengchen,

2017). These regulations are very concrete, specifying almost every possible form of content. For example, in Digital Device Section of the document, words such as ‘budget’, ‘purchase’ and

‘professional’ cannot be displayed in the title. For female streamers, the regulations specify the

‘exposed area of the chest should not exceed one-third of the chest, and the minimum size of the tops should not exceed one-third of the chest’ (Douyu, 2016). Douyu has also developed a credit point system for each streaming room and streamers, and super regulators can deduct points of offenders according to the rules (Douyu, 2016). Once the offenders’ credit goes below a certain point, they will face specific enforcement actions, including ‘no gift reception’, and their accounts will be frozen or even permanently banned.

Platforms have also made attempts to classify the content by themselves. Over the past two years, in an effort to protect youth from inappropriate content, many live streaming platforms including Douyu released a Youth Protection Model, which limits the duration of watching, displays selective stream content and disables live interaction functions to protect children (Laotou,

2019). This preliminary classification system is a typical case of how self-regulation from media

102

practitioners can be more advanced and more flexible than government agencies for regulating crowd-sourced live streaming with significant content variations.

Advanced technologies have been applied to real-time monitoring

To reduce the workload of the super regulators, digital tools are utilised. In addition to text identification system based on pre-set keywords, Peak Concurrent Users (PCU) are used to assist in live streaming content supervision (Jingyan, 2016). The PCU will increase dramatically over a brief period if one room detects the broadcasting of illegal or controversial content. Thus, manual supervision only needs to pay attention to those rooms with high and frequently changing PCU.

Image and audio recognition technology is also used by the platform supervisory team to identify illegal content for some platforms (Douyu, 2019, p. 154; Soo & Ho, 2019). According to the interviewees, the digital tools that are deployed by the platforms are very advanced. R1 noted that:

The live streaming platforms have developed their own monitor technologies,

which are very advanced. Particularly, the image recognition technology of the

live streaming platforms can recognise nude and bloody elements of a video.

And the monitor technologies that developed by YY is even more advanced than

those of Tuputech – a well-known tech companies specialised on content

monitoring and voice recognition technology. (Interviewee, R1, 28/06/2019)

Even so, current video and voice recognition technologies still need improvement, as streamers can always find a way to bypass the detection, as noted by R2.

It is worth noting that the Internet Live Service Management Regulations requires all the streaming content to be reviewed before being released, but it seems that none of the platforms is following this process. According to P1, P2 and R1, all the streaming content is reviewed when it

103

is streamed or after being streamed and archived – no platform actually reviews the video content before releasing it, because it does not fit the style of live streaming. However, the platforms do follow the requirement of archiving the streaming content. For example, the Internet Live Service

Management Regulations requires all the streaming content to be archived for at least 60 days for the post-review, and P1 said the platform archived the video for even more than 60 days. In his own words:

There are two ways for the platform to archive the videos: logical storage and

physical storage. The logical storage means that the videos are stored on the

platforms and are accessible for everyone in a period, which is usually three days or

more. The logical storage will last longer if it involves big streamers or events.

After this period, the video will be deleted from the platforms. However, all the

videos are also archived on the database of the platform for at least 180 days, which

is called physical storage. We can still review it when necessary or being required

by the government. The details about how long a video should be archived depends

on the current polices and the cost. (Interviewee, P1, 31/12/2018)

Clearly, advanced technologies have been applied for the real-time monitoring and post- publication reviews. The platforms did follow most requirements of the government, but chose to ignore the unreasonable ones such as pre-production review. Even so, none of the platforms has been punished or questioned by the government for not doing the job of pre-production review, showing the flexibility of the governmental policies.

104

Real-name registration system as the premise for self-regulation

On 15 April 2007, a real-name registration system was established by government departments including the Ministry of Education, the Ministry of Public Security and the Ministry of the

Information Industry (Douyu, 2019, p. 163). Regarding the live streaming, as a response to the requirement of CAC according to the Internet Live Service Management Regulations enforced in

2017 (CAC, 2016), and to better implement self-regulation, all the legitimate platforms require streamers to register with their real name before streaming (Figure 19).

Figure 19: Registration process of Douyu (Douyu, 2019, p. 147)

For viewers watching live streaming content, real-name registration is not necessary if they do not upload any content. However, they must register if they want to post any live comments.

All streamers must register and verity their identifies with ID issued by the government before streaming. For example, they can simply register through their mobile phone number, as their bank card, real-name and personal information have all been connected with their mobile phone number.

After completing real-name verification, the streamers can create their own room/channel, and they are responsible for the content of their room. Usually only one room is assigned to one streamer to enable better management, and when multiple streamers need to co-stream together they can connect with each other by using the co-stream functions on their channels provided by the platform (Douyu, 2019, p. 147).

105

The review process in leading platforms such as YY and Douyu is very restrictive, as claimed by R1 and all the streamers: new streamers must register with their real name, ID and phone number. Moreover, streamers need to pass a video chat test to confirm their real identity.

S8 noted that her platform also requires face recognition for the registration. Obviously, there is no loophole for the real-name registration system on these leading platforms, and the real identities of the streamers are linked to their channels. To protect the privacy of the streamers, personal information collected by the registration process is kept securely by platforms, as noted by R1:

‘The real name information is kept at the backstage, and the streamer can chose whether to reveal their identify or not by themselves.’

However, some smaller platforms do not have tough requirements for real-name registration. According to one agent:

The real name verification process of some platforms is not that restrictive, so the

streamers can just use another ID to register a new channel if being banned. As

long they do not commit a serious offence that catch the attention of the

government, streamers can use new ID to register a new channel again and again if

the original one was just banned by the platform (instead of the government).

(Interviewee, A1, 14/01/2019)

Copyright is not yet a major issue for live streaming in China

All the streamers and agents said they had not been involved in any copyright issues yet. One platform staff member (P1) also said that copyright was not a big issue. According to P1, there have been no copyright issues related to game streaming in the platform. However, there are some requirements for movie streaming and singing. The platform bought the copyright of some movies

106

and it is permitted for streamers to live stream those movies. However, if the streamers are streaming movies that have not been purchased by the platform, the regulatory staff would ‘move the channels to the bottom’ so they would not be listed in obvious sections. P1 noted that, ‘This is also case by case, depending on the policies at the moment.’

According to P1, it is acceptable for the streamers to sing any song when streaming, ‘as long as the lyrics are not displayed on the screen’ (Interviewee, P1, 31/12/2018). If the lyrics are displayed, it might lead to copyright infringement.

Third Tier self-regulation: Peer regulation from viewer communities

While the self-regulation of the companies is not seen as a satisfactory long-term solution by many stakeholders, co-governance models are regarded as more democratic and accountable approaches.

Peer regulation from community users, as one of the co-governance models, has been adopted by many platform companies (Gorwa, 2019b, pp. 864–65). On Chinese live streaming platforms, peer regulation approaches have been harnessed by platforms to strengthen their own self-regulation, as discussed below.

Room regulators for the moderation

Currently, there are a few room regulators for each streaming channel, who are nominated by the streamer in charge of a particular room. While one super regulator from the platform has to take care of about 30 channels, one channel can have about 10 room regulators, who are at the front tier. Room regulators are normal viewers, but once being nominated they become the supervisors of that streaming room, and they have the privilege of either muting or expelling violators. These room regulators have less power than super regulators, since they can only control the viewers but not the streamers. Instead, these room regulators obey the streamers. Generally speaking, viewers

107

who donate more money and maintain a good relationship with streamers are likely to become room regulators. They volunteer for these supervision jobs and receive no pay, but sometimes streamers will gift room regulators virtual currency (which can then be exchanged for RMB) as a reward (Han & Shengchen, 2017).

I asked the streamers how they nominated the room regulators. Three streamers said the agent companies provided an assistant as one of the room regulators to manage the viewers. Other streamers claimed that the room regulators were nominated by streamers only. For example, S5 told me that the room regulators were their best fans, and they were not decided by the agent companies. Another streamer S4 said the position of room regulator is highly sought after among viewers:

Room regulators are nominated by me only. Usually they are the top ranked

viewers17 of the channel and I know them very well. They also help me to edit the

streaming videos voluntarily. (Interviewee, S4, 27/01/2019)

On the question of how they deal with the hate speech and harassment in their streaming channel, most streamers noted that the moderation jobs are handled to the room regulators and they just ignored those examples of inappropriate speech – the room regulators would delete the inappropriate chat and might ban the posters. One streamer (S8) told me that some inappropriate content might be posted by the competitors of the streamers:

Sometimes they employ spammers to post hate speech in my channel. In this case I

will confront these spammers directly or let my room regulators to argue with them.

17 Those who donate most ‘tips’ to the streamers become the top-ranked viewers, and they are highlighted on the channel.

108

You know, sometimes arguments and debate can draw me more attention and

fans … So I turn their weapon into my own weapon. (Interviewee, S8, 24/07/2020)

S4 said she would get furious at the beginning of her streaming career towards those who said something inappropriate, but right now she has a ‘positive mind’: ‘It is quite funny to argue with those people for me right now, and the room regulator will do their jobs if ban them is necessary.’

Another streamer, S5, has a special way to treat those hate speech posters:

I negotiate with them and try to convince them or tough their mind with my love. In

this way, I can turn them into my own fans or friends. I did succuss in some cases.

If I really cannot change them, my room regulators will stand out and sanction them

– mute these viewers or ban them. (Interviewee, S5, 29/01/2019)

Flagging

Another form of peer surveillance is peer reporting, usually known as flagging. In the case of live streaming platform YY, which focuses mainly on live shows, it is remaining in ‘close touch and dialogue with government’ and using ‘relentless effort to crack down’ on undesirable content (Lu,

2014). One method promoted by YY is peer reporting, and YY is rewarding reporters with RMB.

As required by the Ministry of Culture’s Administrative Measures for the Operation of Network

Performance in 2016, platforms have to set a ‘Report’ button for viewers to report illegitimate content (Ministry of MCT, 2016b), but viewers have no obligation – and often no intention – to do so. In that case, YY gives cash rewards to users who report inappropriate content (Lu, 2014).

Once the content is reported, supervisory staff from the companies or governments make the final decision about whether the content should be deleted or removed, and whether the disseminator should be punished. Similarly, Douyu is striving to improve its peer reporting system, promising

109

to resolve user reports within 90 seconds and awarding reporters with virtual currency (Douyu,

2019, p. 154). The report system is also connected to the government. In fact, it has become a key means for the government to identify illegal streamers and platforms, and this is significantly reducing the workload of government censors (CAC, 2018b).

By empowering self-regulators and transferring responsibilities to other stakeholders, the burden of regulating live streaming industries is reduced for the government. In the wake of these policies, self-regulation has become more effective as industry policy-makers and everyday users are experienced practitioners in the live-streaming industry. They are the stakeholders who can keep up with the ever-changing technologies, cultural and economic demands, and adopt macro government policies in various stream content. This seems to verify the claim that ‘a significant amount of economic, political, and cultural power has actually moved from the state to the media system’ (Howard, 2011, p. 20). But the question of whether ‘the power residing in media networks is stronger than that residing in states’ still needs further exploration in the case of live stream regulation (Howard, 2011, p. 20). Reinforcement of self-regulation does not mean that the government loses its grip on control of live streaming; instead, the central government of China continues to use various regulatory techniques, including several subtle measures, to keep the industries under control.

The complementary measures of the state: More subtle forms of control

The bottom line: News content control

The Chinese government still maintains a high level of direct control over all forms of news transmission, and live stream content is no exception. The Internet Live Service Management

Regulations of 2016 specifically outline the requirement for news streaming (CAC, 2016):

110

• ‘Double qualifications’ on news services. For news services, both the platform and streamer

shall obtain qualifications (or licensing) specifically for Internet news information services,

and the streaming content should operate within the scope of the license (Article 5).

• Supervision. Platform operators shall establish a supervising team for content scrutiny. The

streaming content should be classified, and for all the news content streaming prior

censorship is necessary (Article 7).

With regard to news, CAC enacted the Internet News Information Service Management

Regulations on 2 May 2017 (CAC, 2017), based on the current digital environment. These regulations also require that all digital media including websites and mobile applications apply for

News Information Service Licenses from CAC if they are providing news services.

However, according to the Chapter 2 and 3 of these regulations, it is impossible for live streaming platforms to meet the standard to apply for a news licence, as it requires a professional standard of production and distribution. For example, Chapter 2, Article 6 requires service providers to have a professional editorial panel and reviewers. Chapter 2, Article 8 forbids the involvement of non-publicly owned capital in news production. These requirements are impossible for amateur individual streamers to comply with; as a result, the authorities will cut off any news program in live streaming platforms. This may be explained by the requirement, specified by the regulations in Chapters 1 and 2, that the news services do not just refer to original news production and distribution, but also need to refer to real-time video footage, news reposting and commentary

(CAC, 2017). In this sense, even reposting any news-related content on live streaming platforms is an illegal activity. These restrictive policies on live streaming show the consistent grip on news production and distribution, ensuring the dominant position of mainstream media agencies for cultural construction while eliminating the potential of citizen journalism in live streaming. For

111

example, one leader of the supervisory team of Inke commented that the government planned to build a refuse incineration plant, but local citizens who did not agree with the plan then gathered to protest. According to the policy, there is a requirement to prevent users from streaming the event.

The location of the protests were revealed by the supervisors, who drew a circle with a radius of

10 kilometres, and all users in that area were unable to use the live stream app (Jing, 2019). During important events such as the Communist Party Congress, the supervisory team is required to be more cautious in relation to sensitive information (Jing, 2019).

Admission to the market: Licensing

Using licensing to control the digital platforms is not a new strategy of central government.

Previously, each streaming platform only needed an Internet Content Provider (ICP) Licence from

MIIT and a Network Culture Business Licence (网络文化经营许可证) from the Ministry of

Culture. These two licences are necessary for all digital media platforms in China. From 2016,

SARFT began to take control of live streaming industries. When applying for a Radio and

Television Program Production Licence (广播电视节目制作经营许可证) from SARFT, it became a new requirement for the platforms that wanted to use the word ‘TV’ in their names and in order to make its own official program (SAPPRFT, 2016). As a result, many streaming platforms chose to cut off their own program production and remove the term ‘TV’ from their names (e.g. Douyu.TV became Douyu.com) to avoid the licensing issue18.

In addition, on 9 September 2016, SARFT issued a Notice on Strengthening the

Management of Web-based Live Program Service, which for the first time defined streaming

18 By July 2020, leading video platforms such as Huya, Douyu, Tiktok and Kuaishou had gained Radio and Television Program Production Licences, and had started to make their own video programs in addition to user-generated content.

112

content as audio-visual content that should be regulated directly by SARFT (SAPPRFT, 2016).

The Notice requires all streaming platforms to apply for a ‘Licence for Publication of Audio-Visual

Programs through an Information Network’ (信息网络传播视听节目许可证) issued by SARFT.

Figure 20: Licences and related government agencies

Radio and Television Licence for Publication of ICP Network Culture Program Production Audio-visual Programs Licence Licence Business Licence Licence (for producing through Information own program) Network

Agency MIIT Ministry of Culture SARFT (Now NRTA) SARFT (Now NRTA)

However, compared with other industries, some licences that are required to operate a live streaming platform are extremely difficult to obtain. For example, the License for Publication of

Audio-visual Programs through an Information Network, which is now required for live streaming platforms and for all audio-visual content on the internet in China, was easy to obtain before 2008.

However, after 2008 there were new requirements for applying for this licence (SAPPRFT, 2007).

According to the new Article 8.1, the applicant should “have the corporate capacity, is an exclusive state-owned entity or a state-controlled entity, and have not committed any violative act within the three years prior to the date of application”. Thus, in order to obtain this licence, a live streaming platform must have applied and acquired it before 2008, or have been state owned or state controlled to obtain this licence after 2008.

As most live streaming platforms in China started their businesses after 2008 and are not state controlled, they remain in an awkward legal situation, operating without a licence or seeking to acquire one. Acquiring a licence is not an easy task: the price was up to 35 million RMB (about

A$7 million) in 2017 (Chuanmeitoutiao, 2017). Once offending content is detected on these platforms, they would be found to be operating without a licence (SAPPRFT, 2017a, 2018b).

113

Leading platforms such as Douyu have gained this licence by acquiring licence-holders in recent years (Douyu, 2019, pp. 25, F-25; Shen, 2019). However, most platforms that do not have sufficient budgets can only run without the licence. This places these platforms in a high-risk situation as they can face a penalty at any time if government wants to pursue them. Platforms without licences face penalties ranging from fines to being forced to suspend services or even being shut down permanently; and licence-holders who offend against these regulations will also have their licences revoked in some cases (Wei, 2017). Moreover, with all licences requiring to be displayed at the bottom of the homepage of the websites, this renders it easy for authorities to filter out the unlicensed platforms.

In short, the requirement for licensing applications and maintaining regulatory standards has dramatically increased the costs of operating a streaming platform; consequently, there is a dampening effect placed on rampant growth of live streaming platforms. For one thing, this high entry barrier automatically filters out speculative small platforms that cannot afford licences. For another, it disciplines giants who care about their brands and the initial cost – either reinforcing self-regulation to censor offending content in a timely way or risking severe legal sanctions being pursued by the government if detected.

Becoming an inspector: The changing role of the government

Drawing from the interviews and related documents, we can see that the role of the Chinese government for live streaming has changed. The traditional command-and-control regulation is no longer applied to live streaming. As one agent (A1) noted, the government does not monitor streaming content directly. Even if the inappropriate content has been reported to the government, usually it is only the platforms that are punished, leaving other actors to negotiate with the

114

platforms about their final accountability. In a regulatory discipline sense, streamers and agent companies are relatively ‘safe’. The platform staff (P1) also mentioned that:

If there is a problematic content, as long as it has not been reported by the media,

the government would not care. There is not direct supervision from the

government right now; the supervision is mainly conducted by the live streaming

companies, and the government only punish the platform afterwards if there is a

problem. (Interviewee, P1, 31/12/2018)

Although the government has reduced consistent supervision on live streaming, it still acts as an inspector to ensure the self-regulation goes well. Different departments may work together to launch several spot checks per year, and to crack down on unlicensed or delinquent live streaming platforms. The spot checks, which are called ‘special operations’, are the main way to investigate suspect platforms and punish offenders (CAC, 2018b; SAPPRFT, 2017a; Wei, 2017).

As most live steaming platforms are known for hosting ‘inappropriate’ content, they have been placed on the Chinese Ministry of Culture’s investigation list. There are occasional investigations of these suspect platforms, and many streaming platforms and channels are banned by the Ministry of Culture because they contain inappropriate content (iiMedia, 2016). In June

2017, the Ministry of Culture specifically investigated over 10,000 live streaming mobile apps, closed down 11 apps and revoked the Network Culture Business Licence of one company (Wei,

2017). On 26 July 2017, a number of platforms and about 4000 channels were forced to close because of illegal content (CNNIC, 2017, p. 63). Six months later, more than 5000 streaming channels were closed down by CAC and MIIT on February 2018 because of illegal content (CAC,

2018b). As mentioned earlier in the chapter, the peer reporting system of the platforms is connected

115

to the government, and it provides key information for the government to identify offending platforms (CAC, 2018a). Most recently, according to the peer reporting of internet users, donations on the live streaming platforms – especially those on the small and medium live streaming platforms – has stimulated the growth of vulgar content, with a detrimental impact on mainstream values. Thus, all the relevant governmental departments worked together to conduct special operations from early 2020, focusing on investigating the platforms that had been reported by the internet users. Many delinquent small and medium live streaming companies have been disciplined or suspended (People.cn, 2020).

In most situations, the government only targets platforms that have failed to fulfil their self- regulation responsibilities (SAPPRFT, 2018b). However, streamers could also be identified and banned in extreme cases during special operations actions. For instance, a live streaming celebrity called Tianyou Li was permanently banned from broadcasting after he was shown singing songs that promoted drug use. Tianyou Li had huge fan communities, so his actions were deemed to be very consequential for the public interest, including their negative effects on youngsters’ mental health, according to CAC (2018b). Therefore, serious offences deemed to influence social stability or public interest are a reason for the government to crack down without hesitation, even as an inspector.

Discussion: Combining self-regulation with direct government control?

Given the evidence laid out in the preceding sections, examining past processes of Chinese regulation on platforms in a centralized and top-down model fails to comprehend the complex and dynamic power relationships between different stakeholders. In this complex and dynamic situation of live streaming, constituted by changing social relationships and communication

116

technologies, we can observe that the regulatory power residing in the media networks is stronger than was seen previously. While the traditional state interventions around the live streaming industry have turned out to be less effective, the strategy of passing the responsibility and workload of regulation to industrial actors, including associations, platforms and streamers, has turned out to be a very effective strategy. Rather than the continuing centrality of the party state in terms of digital media regulation in China, non-state actors have emerged as the core regulatory organs.

This clearly represents a shift, and the party state has changed its approach. On one hand, the Chinese state is not acting as a regulator but instead as a stakeholder in the market. It is acting like an ‘enabler or facilitator’ to achieve its political, economic and cultural outcomes (Flew, 2016, pp. 25–27; Z. Zhang, 2020). The agendas of building digital infrastructure, cultural empowerment and digital sovereignty have boosted the development of live streaming in China. However, live streaming content is difficult to regulate in a traditional command-and-control model. Moreover, live streaming companies are neither state owned nor state controlled; instead, they are non- ideologically oriented media platforms created mainly for entertainment, which means that they enjoy less strict regulation (Guo, 2014). These conditions, together with the rapid development facilitated by the state, have brought a number of issues to live streaming all at a once, as discussed previously.

To deal with this problematic situation, the state has created political and regulatory conditions that enable and facilitate the self-regulation of the actors in the media industries. For oversight of self-regulation, the state is acting both as an inspector in specific fields and as a complement to the self-regulation of the live streaming industrial actors. The complementary measures taken by the government include, but are not limited to, licensing, news information control and spot checks. Licensing control increases the initial cost and operating standards for

117

streaming platforms; together with special operations on unlicensed platforms, small and speculative live stream companies tend to be automatically filtered in the process. Peer reports from users, as part of the self-regulation, are also used as a reference for any special operations into suspect platforms and channels. Many platforms have been closed down by the government or acquired by the big platforms over the past few years. From 2015–16, there were more than

700 streaming platforms; however, by the end of 2017 only approximately 200 remained and the number is still decreasing (Laoyu, 2019). Following this shakedown, the remaining platforms and streamers have become increasingly professional and consequently have begun to take their brands more seriously after the crackdown, making self-regulation and government control easier and more efficient.

Issues arising from complementary government control

The combination of self-regulation and subtle governmental control is effective but not perfect.

There are still points of friction when it comes to implementation. This subtle control exerted by the government, in a complementary fashion, is something that is occasionally meted out to platforms when they step out of line. However, the line is not always clear, and it is sometimes unpredictable. In ‘special operations’, the categories of inappropriate content is broad: in addition to regulating sexually explicit, vulgar and politically sensitive content, some game-streaming sites are regarded as ‘abetting crime’ – for example, Grand Theft Auto 5 (GTA5) is seen as encouraging violence, and poker games as instigating gambling (Ministry of MCT, 2016a). Yet the ambiguity of these clauses with regard to regulating digital media has always been quite tricky both for regulators and corporations in China, and the actual implementation of regulation often falls to individual discretion and interpretation (Yan, 2014). As one regulator (R2) noted:

118

The regulation policies from the government lag behind the industrial development.

Moreover, there are too many regulatory departments for regulating the live

streaming industrials and these departments are not well coordinated. The specific

legal provisions of these department are not consistent and vague – no one knows

how to define vulgar content … Therefore, the platforms are relatively more

sensitive towards high profile content and they tend to censor it all, as the

companies may not be able to afford the cost of the punishment. For example, Huya

cut off all the sensitive and political content directly … Even so, sometimes the

government still blames the companies for not doing enough in the self-regulation,

and the platforms feel wronged as they have no idea how to do it better and faster.

(Interviewee, R2, 16/07/2019)

It can be seen that, although self-regulation criteria of the platforms are reasonably clear and concrete, the criteria for direct control from government are still unpredictable. If we examine recently released documents more closely, we can see that there is still a large grey area between what is legal and what is not, and this has resulted in poor policy and regulatory enforcement outcomes during special operations. First, the policies do not give a clear definition of what is

‘inappropriate content’ for the censors. There are only two Articles in the Internet Live Service

Management Regulations proposed by CAC that mention content management criteria: Article 3 requires live-streaming to ‘cultivate a positive online culture’, ‘maintain national and public benefit’ and ‘create a healthy online space for youngsters’; Article 9 forbids live-streaming from containing content that ‘threatens national security, damage social order, disturb social order, infringe others’ right, spread pornography and obscenity’. For the Ministry of Culture, according to Article 6 of the Administrative Measures for the Operation of Network Performance, streaming

119

content should not have ‘horrible, cruel, violent, vulgar content that damage performers mental and physical health’; nor can it contain gameplay without the approval of Ministry of Culture, as it might ‘damage mental and physical health of juveniles’. Moreover, Article 15 claims that ‘credit rating management systems and blacklist systems should be established by platforms for regulating streamers’, with the words ‘should be’ having a guiding purpose in that context. The criteria for regulation are not only concrete laws and regulations, but also more abstract concepts of the public interest and mainstream values that are required by current administrations (SAPPRFT, 2017b). It is common to see platforms or streamers being accused of publishing vulgar content and corrupting public morals without any reference to specific laws or regulations (CAC, 2018b; Lai, 2017;

SAPPRFT, 2018b). Moreover, the responsibility for content violation is also not clarified. As Zhao

(2019) notes, Chinese service providers can receive safe harbour protection if they follow notice- and-takedown procedures and have no actual or constructive knowledge of a user’s infringement activities. However, the provisions are ambiguous in relation to how service providers would be deemed to be constructive in users’ infringement activities (Zhao, 2019, p. 67).

These criteria, which lack details and standards for content filtering and accountability, are thus leaving regulators to make often-difficult decision about what is ‘inappropriate content’, and this can lead to poor policy understanding and disordered regulation. In the Western context, the personal discretion of the moderator rather than the rule of clear black-letter law is one of the biggest problems with self-regulation (Common, 2020). Similarly, in China this has rendered the regulations ‘rule by man’ rather than ‘rule by law’, making consistent regulation implementation difficult and leading to the possibility of corruption (Hao, 1999). In this situation, regulatory staff from the government are likely to be ‘giving priority to following orders and not making mistakes, even if it is considerably more inefficient to do so’ (King et al., 2014, p. 6). Censoring legitimate

120

content will tend not to place regulatory staff into a difficult situation, but letting inappropriate content pass may lead to accusations of dereliction of duty or even constitute a political mistake.

As a result, the regulation from government regulatory staff is usually more restrictive than it should be, thus making regulation and penalties unpredictable to streamers and platform operators.

As a response, self-regulation rules from the platforms may on occasions be excessive. One platform employee P1 confirmed this in an interview:

The self-regulation from the platform is tougher than the regulation from the

government. For content that is difficult to determine whether it is a violating one,

the platform regulators tend to censor it rather than keeping it as a threat.

(Interviewee, P1, 31/12/2018)

Zhou (2006) uses the metaphor of the Sword of Damocles to describe the effectivity of ambiguous policies on regulation: ‘Like the sword of Damocles hanging over editors’ heads, the ambiguity of government policy in this area has turned out to be very effective’ (Zhou, 2006, p. 179). Whether intentional or not, for the Chinese government this metaphor is still applicable to the disciplinary power on live streaming. On the one hand, the absence of a Licence for Publication of Audio- visual Programs Through an Information Network and the objective of cultural construction justify any accusation on live streaming platforms. On the other hand, we can observe that special operations on offenders and the ambiguous provisions can lead to excessive regulation, requiring platforms and streamers to steer clear of sensitive content in order to secure ideological and cultural compliance. Even so, platforms are still at risk according to Douyu: ‘under the uncertainties of interpretation and application of pertinent laws by the government authority, any operating measures from the platforms may lead to fines, confiscation of income, and suspension or revocation of the licenses’ (Douyu, 2019, pp. 25–26).

121

Hurdles for self-regulation

The risk of platform-led moderation

Although the self-regulation rules and policies made by platforms are clear and executable, the question of how to keep self-regulation compliant with the ambiguous and constantly changing government standards still surfaces for the platforms – especially so when the supervisory cost has become quite high. This issue has emerged as an obstacle for regulation, given the tendency to apply digital technologies for automatic supervision rather than manual supervision, as AI requires more standardised and consistent criteria for the operation (Jingyan, 2016; Soo & Ho, 2019).

Self-regulation-centric modes have led to the platforms becoming both the rule maker and rule enforcer; which means that certain forms of content may be moderated automatically by the ever-evolving ‘rules’ of the platforms, even if such content complies with the policies from the government – and platforms become ‘legal’ entities for their own interest. While self-regulation in a Western context is often justified as a process of ‘cleaning up’ the online environment, or providing better user experience (Gillespie, 2018), it is also becoming a tool for platforms to maximise their own interests (Cohen, 2017). In China, where the operation of media law is imperfect, the situation could be worse – for example, forcing companies to act totally as their own regulators, and thereby replacing the government, can create a problematic position in countries with a weak rule of law (Human Rights Watch, 2018). It is plausible that for the platforms in China may overreact and delete more content than necessary, all in the interests of improving efficiency and protecting the interest of the companies, even though these categories of content might be beneficial for society and would be regarded as appropriate for the government supervisors. Given the philosophy ‘kill by mistake rather than let it go’, for the government regulators it is not the government’s concern if the legitimate content was censored by the platform regulators.

122

Consequently, freedom of expression might be affected by the platforms in a new way. For example, White55, as a top streamer of Douyu in the period 2017–18, was reported cheating when he streamed gameplay in November 2017 (Shi, 2018). At first there were earlier reports and critics on Douyu about White55, and the platform regulator was able to negotiate with the reporters and

White55 to resolve the issue. However, after the reports and complaints were filtered by the self- regulation, it was found that these accusations were reasonable and evidence based. There were two reasons for this phenomenon. First, the room regulators, who were nominated by White55, were fanatical fans of the streamer, who were willing to do virtually anything to protect their idol; thus, any information in the channel found to be detrimental to the reputation of White55 was deleted soon after being posted. Second, the platform regulators – namely the super regulators – somehow acquiesced though this ‘self-censorship’ of the appropriate content for the interest of the platform. In other words, the platforms were unwilling to punish or suspend their top streamer,

White55, who was generating huge profits, but nor did they want this case to be noticed by the government supervisors. In addition, another streamer, Songshu dabuguo Cangshu, reported

White55 and streamed their analysis of the evidence of how White55 had actually cheated in the game. The result turned out to be a hoax, with Songshu dabuguo Cangshu instead of White55 being banned by the platform; his Douyu channel has since been closed down (Huaixiansen, 2017).

The grievances of the reporters then escalated quickly, creating a social conflict with the fans of

White55, both online and offline.

By the time the government regulators started interfering, the conflict had already become a huge problem. It can be seen that the trend for platform-led moderation is to be profit leading and short-sighted. Particularly in this case, because of the excessive and inappropriate self- regulation, a small issue like this grew rapidly, causing negative impacts on social stability, moral

123

goodness and the economy. The government agencies transferred/outsourced the regulatory responsibility and activities to digital platforms, thereby empowering the digital platforms and increasingly making them ‘gatekeepers’ for their own good (Flew, 2019). However, the resulting self-regulatory actions by platforms can easily be distorted and lead to new forms of capital accumulation and labour extraction.

An accountability conflict with the protagonist of self-regulation

It is arguable that although the platforms are the stakeholders who ought to be accountable for self- regulation, the direct regulators are not the platforms. Different stakeholders are conducting self- regulation together across different levels. At the bottom level, agents are the first entity to regulate streamers, and probably the only stakeholder to regulate streamers directly. During the interviews, all the streamers said that they dealt with agents and there was no interaction with the platform – that is, the agents handle all matters on behalf of streamers. When there are new rules, agents are the ones to inform them. Both the agents I interviewed told me that there was a training program before streamers were hired, and that they also coached the streamers during their employment.

According to one agent (A2):

The streamers only need to listen to our order … we do know how to display

relatively sexually implicit content with designed dressing and make up, while

avoid the risk of being caught and punished … If there is a problem with the

content, the government would not go to us. The government only punishes the

platforms, and the platforms are the ones to interrogate us later. (Interviewee, A2,

15/01/2019)

124

The real protagonists of self-regulation, and perhaps the most influential ones in this sense, are the agent companies. Yet the platforms usually take sole responsibility. All the regulators and platform employees informed me that the government was not fully aware of the existence of the agents, and usually penalised the platforms when rules were breached.

While platform employee (P1) noted that the platforms sometimes sign contracts with the agent companies to transfer the responsibility to the agent companies (Interviewee, P1,

31/12/2018), it seems that the government does not recognise these kinds of contractual arrangements, and continues to punish the platforms in the majority of cases. As one agent (A1) mentioned, the contract forces his company to take the financial responsibility for the inappropriate content, but it is not a significant issue in his opinion:

We pay the fine and fire the streamer – the fine is then deducted from the wage of

the streamers. Relatively, we suffer less loss. But it is a great loss for the platform

because it is not just about money for them. The streamers could also be blacklisted

by the government in addition to the fine. (Interviewee, A1, 14/01/2019)

From 2016, outsourcing regulation responsibility and workload to agencies has become a common strategy for the platforms, enabling them to be as lean as possible (Jingyan, 2016), but from the interviews above and recent penalties imposed on the platforms, we can see that agents are downplayed or omitted, despite being the protagonists of self-regulation. While the platforms are the second entities to regulate, they have to take all the responsibility. This conflict puts platforms in an awkward situation while giving a chance to the speculative agents to profit from ‘illegal’ content.

125

The flipside of compulsory self-regulation: Speculative small platforms and the domination of industrial giants

The cost of the real-time monitoring team is substantial, and the companies tend to reduce the cost by moving the regulatory department to the regions where cheaper labour is available. As R1 noted:

While Huya is located in a first-tier city (Guangzhou), it moves the monitor team to

second-tier or third-tier cities such as Zhuhai and Changsha. The monitors are still

full-time employee of the company, but their labour in those cities is cheaper. In

addition, the company can also enjoy the tax deduction in those cities according to

local policies. In short, moving the monitor team to these cities can significantly

reduce the cost. (Interviewee, R1, 28/06/2019)

Imposing self-regulation and liability on the platforms could increase the cost of the platform, and these industrial giants would gain more advantages compared with smaller business competitors if this self-regulation became a requirement – when they have already built up an existing regulatory framework and teams. In addition, the industrial leaders have an inherent impulse to carry out the self-regulation – they value their brand, and take seriously the costs of applying the licences and other long-term efforts and investments.

In contrast, the smaller platforms can barely afford the costs of having real- time monitoring.

These platforms, which often have insufficient budgets and expectations of long-term development, tend to be more speculative. As my interviews indicate, some of the smaller platforms do not even have strict real-name authentication systems in place, nor do they have a strict real-time supervisory team. They even tend to make use of vulgar and sensitive content in order to attract the attention of viewers, and of course to gain profit in the ecosystem that is dominated by the big

126

platforms. Most live streaming platforms that were recently cracked down on by the Chinese government were smaller platforms (People.cn, 2020).

This might be good news, as it shows that the self-regulation of big companies is going well. Yet, from another perspective, it indicates the potential of monopolisation in the near future.

The live streaming market in China is becoming oligopolistic, but given historical trends in the media and communications industries, it is likely to evolve to become monopolistic. Live streaming platforms have in fact shrunk from about 700 in 2019 to 200 in 2019, and top streamers have been acquired by top platforms. Among the top live streaming platforms, Tencent is the largest shareholder of both Huya and Douyu. On 10 August 2020, Tencent issued a preliminary proposal to merge Douyu and Huya (She, 2020) – a strong sign that the live streaming industry is very likely to become dominated by Tencent in the near future.

Towards a more nuanced analysis

This chapter has analysed the changing role of the Chinese state in platform regulation and particularly in the platformisation of live streaming. While the official regulators have changed their strategies, issues such as excessive moderation, conflict of accountability and monopoly remain. Zhao points out that the evolving state-market dynamics in China complicate the regulatory pathways to the state, and ‘governance operates not only from the top down, but also from the bottom up and sideways, involving multiple actors negotiating in the space where collaborations, contentions and compromises coexist’ (Zhao, 2019, p. 141). While it can be seen that the traditional command-and-control regulation from the state is ineffective for live streaming, this does not mean that state regulation is weakened. Rather, state-centric regulation could still remain or become even stronger if it successfully recognises and manages the complex relationships between different stakeholders in the industry. As examined earlier, the ways the

127

state steers self-regulation with licensing, and the plan of taking direct control of the peer- regulation system from the platform companies, have shown the increasing capacity of the evolving state-centric regulation. Therefore, rather than simply providing suggestions to resolve the regulatory issues merely from the perspective of the state, a more detailed analysis of the integrated ecosystem of live streaming is required. This chapter has focused mainly on the regulation from a political perspective; the next chapter will draw on economic and cultural perspectives to further analyse the dynamic power relationship between different stakeholders in live streaming, to draw out a more holistic picture of the live streaming platformisation in China.

128

Chapter 4

The complexity of platformisation

Chapter 3 analysed the dilemma of platform regulation and the contentious regulatory issues of live streaming platforms in China, along with the transition of policies in recent years. While these discussions were drawn from institutional perspectives and the focus is limited to the moderation of inappropriate content, this chapter will take a more holistic view of the complexity of the platformisation of live streaming in China.

As Gorwa (2019b) notes, when analysing the regulation of the platform, we need to further capture the relationships between layers of regulation and its structuring interactions with key actors. Rather than seeing the Chinese media as a uniform, singular entity, it is better to see it as a multi-stakeholder media system (Wang, 2018), characterised by negotiation and conflicts between different actors, including users, platform providers, fragmented regulatory authorities, and even non-human actors such as algorithms (Hong & Xu, 2019). In this sense, an examination of the politics of live streaming should not just focus on narrow factors such as copyright, privacy issues and the moderation of illegal content, but also needs to consider the underlying mechanisms that may influence live streaming activities, foster capital accumulation or exert social and political influence over all aspects. In particular, we need to consider the impacts of these mechanisms associated with the fundamental origin of the regulatory issues. Drawn from this perspective, this chapter uses interpretive and comparative approaches to analyse the complex relationships between multiple stakeholders and the unique live streaming practices that exist in China. Three

129

key words lead the discussions in this chapter: empowerment, commodification and professionalisation.

Filling the gap: Live streaming for the empowerment and gratification of marginalised social groups

Live streaming as a new production tool for the grassroots

The regional variation in China is huge, and it is characterised by imbalanced economic development (Mai et al., 2017). According to one report, while the average wage of people in

Shanghai reached 9500 RMB per month (Shanghai Bureau of Statistics, 2020), 600 million people in China, or about 40 per cent of the overall population, still had a monthly income of less than

1000 RMB (Q. Li, 2020). Residents of remote or undeveloped regions, or those migrant workers who travel from the hinterland to find a job in coastal cities, are rendered marginalised and become the voiceless grassroots workers. Yet with the evolution of mobile devices and the enhancement of ICT infrastructure in rural areas, these grassroots workers have also gained access to digital technologies. However, rather than empowering them, these digital technologies seem to have intensified the digital divide. In Qiu’s opinion, we cannot simplify the social division in China as a binary mode of ‘information haves’ versus ‘information have nots’ (Qiu, 2009, p. 1). Instead, the marginal social groups – who are indeed capable of using forms of digital media and integrating them into their daily lives – are in a disadvantaged position as they are unable to fully access or utilise the fast-evolving ICTs, or to invest and financially benefit from ICTs’ entrepreneurship compared with the privileged groups (Fong, 2009; Qiu, 2009). Therefore, the grassroots are still voiceless and became the ‘information have-less’ (Qiu, 2009, pp. 13–15). Moreover, the development of ICTs has made the grassroots even more passive in a globalised market, and the

130

grassroots have also been turned into network labour and are exploited by multinational corporations – the workers experiencing extremely poor working condition at Foxconn are a typical case in point (Qiu, 2009, pp. 192–96; 2018). While these grassroots users previously relied upon agriculture and derivative industries in their hometowns, their farming life has also been influenced significantly by increasing urbanisation and informatisation. A substantial number of farmers have turned to migrant workers and services-related industries because of the shrinking amount of arable land (Chan, 2010) and imbalanced working opportunities (Sun et al., 2020). In this sense, the appropriation of ICTs by different social groups has increased polarisation and further marginalised these already marginal social groups (Sun et al., 2020).

However, social media, based on the creative content of individuals, seems to provide opportunities for marginalised social groups to regain an advantage and relieve the digital gap.

Existing studies have shown that social media can empower grassroots users in China by offering alternative communication spaces for them (D. Huang, 2017), and that they can help them to become online celebrities and gain social recognition (Jia & Zhou, 2017; Lin & de Kloet, 2019).

As (Zhao, 2019, p. 66) claims, ‘the low entry barrier in terms of budget, equipment and expertise indeed attracted many grassroots amateurs’ to produce UGC on video websites. Inspired by this, many remote, impoverished towns or even undeveloped hinterland cities have turned their focus to the development of an ‘online celebrity’ economy; this transition has been successful, and some cities such as Xi’an and Chongqing have become so-called ‘online celebrity cities’ (Z. Zhang,

2020). Nonetheless, for the individuals involved, some research indicates that only a limited number of video uploaders have successfully turned from amateur to professional, attracting capital investment in creative entrepreneurship. Many more are suffering from precarity and inequality (Zhao, 2019, p. 134). It is still too early to say whether or not the social media platforms

131

have been able to empower grassroots users, and thus reduce inequality. Rather than engaging in continuous debate about the empowerment of grassroots users, this section is directed at analysing how live streaming, compared with other forms of social media, has influenced the lives of marginal social groups, and integrated these people into a broader digital media ecosystem. This engagement renders the relationships between different actors in live streaming industries in complex ways, providing us with a new understanding of platformisation and the politics of live streaming in Chinese contexts.

With the rise of smartphones and 4G networks, marginalised social groups have increasingly gained access to smartphone apps, and the advancement of ICT infrastructure has made streaming and watching streaming – which was very demanding for data transmission rate

– technically possible in rural areas (M. Shen, 2020). In such contexts, live streaming – a highly interactive form of social media – provides us with a new lens through which to see the affordances of social media in terms of empowering grassroots users. First, live streaming provides a lean and authentic production system for their creative content, lowering entry barriers for informal, creative entrepreneurs. While becoming a successful video uploader on traditional video sites still requires content planning, recording devices and post-production tools and skills, live streaming removes these requirements. With a smartphone and a selfie stick, everyone can live stream immediately without the limitation of space, expensive recording devices or even post-production tools and processes. In my interviews with the streamers and agents, all pointed out that now streamers prefer to use mobile phones to stream, except when they need to use a computer for the streaming activities such as game streaming. In this sense, raw video shot live on a phone camera is usually the final streaming content seen on screens. Moreover, the interactions between streamers and viewers are instant and ephemeral, based on the real-time streaming content, making

132

these ‘more like a slow phone chat than a well-thought-out exchange of letters’ (Powazek, 2002, p. 13). All these features contribute to making the live streaming activity less deliberate or rehearsed, and more casual, emotional and authentic, in contrast to the usually staged and deliberate production system of traditional video making. As noted, it is this authentic and unpolished lifestyle presented by the streamers that makes the content enjoyable for viewers in

China (Lu, 2019). Therefore, the disadvantages of the information have-less, such as a lack of skills and limited access to equipment, have been reduced in this new production system.

Second, Duffy notes that the goal of being a successful social media influencer might constitute ‘aspirational labour’, which refers to uncompensated creative work provided by users who hope to do what they love, and accompanied by the expectation of being fully paid when becoming a celebrity in the near future (Duffy, 2017). Yet this expectation is usually confronted by a cruel truth: most will never be celebrities. As with other forms of social media, successful streamers who emerge as online celebrities in the minority, with the majority of streamers making little money and exist in a life of precarity (Cunningham et al., 2019; iiMedia, 2020; Xinhua, 2019).

Unlike other forms of social media that require a substantial viewership to be ‘successful’, a live streamer can be both mentally gratified and financially compensated even with a small number of viewers. Although some creative workers on social media platforms need to devote labour without compensation for the goal of becoming an online celebrity (Duffy, 2017), streamers do not have the need and desire to achieve ‘celebrity status’ in order to earn profit and gain social recognition.

Less popular streamers will generally always find a position in this industry, gaining sponsorship from companies and forming a community of viewers around them in order to gain recognition and donations (Woodcock & Johnson, 2019a).

133

Research indicates that many rural and uneducated streamers moved from lower-class positions to middle-class roles through their live streaming activities. The relatively undeveloped regions such as the North-East Provinces of China have the highest proportion of full-time streamers, and their streaming work gets the highest recognition from the people of these regions

(Xinhua, 2019). The interviews undertaken for my research indicate that although some streamers from economically underdeveloped areas such as Jingzhou and Nanning put a great deal of effort into their live streaming, but still did not become influencers, yet they would often remain mentally satisfied with their live streaming work. Some streamers – also known as nano-influencers or micro-influencers 19 – gain both mental gratification and financial benefits from their casual streaming. In order to fully understand the nuanced practices of live streaming, I have highlighted several examples in the following sections.

Gratification for the non-celebrity streamers

The joy of streaming and interactivity

S2 is a dance teacher in a college and she also teaches a training class after hours. Therefore, she usually gets off work after 10.00 pm. She engages in live streaming from when she gets home until midnight. When I asked about her motives for live streaming, she explained:

My dream is to become a professional dancer, and I started learning dance in my

childhood. However, I was too short to be a professional dancer when I was

growing up. I was pretty sad at the beginning, then I found live streaming as an

19 According to an article in the New York Times, nano-influencers have less than 1000 followers on social media, and micro-influencers have 1000–100,000 followers. See https://www.nytimes.com/2018/11/11/business/media/nanoinfluencers-instagram-influencers.html

134

alternative. I can fulfil my dream on live streaming platforms – to be a dancer,

showing myself and earn the attention and admiration of the others, without worry

about the defect of my height. (Interviewee, S2, 15/01/2019)

At the beginning of streaming, she only had hundreds of viewers, but now has about 2000 viewers.

In addition to showing dancing or talking with her viewers on the platform, she also maintains a good relationship with her fans outside of the live streaming platforms:

We have a WeChat Group organised by the agent company. Some fans also add my

personal WeChat account. We have a good relationship online, and we also play

popular computer games together after streaming hours. However, I have to limit

my activities outside of the live streaming platforms with fans because it might

expose my privacy. In addition, as the agent told me, if I had too many activities

with fans outside of the live streaming platform, the fans might see it as an

alternative to approach me, and thus come to my live streaming channel less. This

might reduce my popularity and profit. (Interviewee, S2, 15/01/2019)

This streamer noted that she really enjoys her live streaming, even though it takes up a lot of time and energy – she has to stream at midnight and sleep late. The revenue she gains is considerable – now she can earn more than 5000 RMB (US$760) per month from live streaming (Interviewee, S2, 15/01/2019).

According to the report, the average salary of private companies in her region was just 3300 RMB

(US$502) per month (Finance Sina, 2019). Clearly, the revenue from live streaming can significantly improve her financial situation in addition to satisfying her intellectual needs.

135

Another streamer (S1), a college student, explained that for her live streaming was just a way to amuse herself. She only has hundreds of viewers and her monthly income is relatively small.

Despite this, she live streams two to four hours every day in her free time:

As a student, I am free and have plenty of spare time. I really enjoy live streaming –

I can talk to my fans and sing songs … It is like entertainment rather than a job.

The streaming activity is very casual to me. I start new conversations with my fans

whenever an idea comes up to me. Sometimes my fans bring about topics and start

discussions, too. (Interviewee, S1, 15/01/2019)

Live streaming activities bring her amusement, even though she does not earn much money from them. S7, as an economic commentator, also expressed that he live streamed to interact with his

‘buddies’ from the community and to answer their questions: ‘It is enough for me if everyone is happy … I do not care about the donations and I have no desire to be an online celebrity’

(Interviewee, S7, 24/01/2019). While these streamers were part-time only, other full-time streamers highlighted their changing careers because of live streaming.

A decent job with considerable income for full-time streamers

S5 was a performer and director behind the scenes, and he notes that it is a precarious job without stability and insurance. After being introduced to the agent company by a celebrity streamer, he quit his previous job and began live streaming as a full-time streamer from 2018. When I asked him why he chose to become a full-time live streamer, he said:

It is mainly for showing my talents, sharing my happiness and humour with others.

Making money is just the secondary purpose. If we do this only for making money,

then it becomes meaningless and soulless. I was an uploader on both traditional

136

video site such as Youku and short video sites such as TikTok. However, live

streaming is a totally different form. The live interactions on live streaming

platforms are synchronous, so I can interact with the audience live, creating a sense

of intimacy. (Interviewee, S5, 29/01/2019)

The main content of S5’s stream is singing and playing games. He has hundreds of concurrent viewers and 7000 followers (this is not many in the Chinese context). He revealed that he streams five to seven hours per day as a full-time streamer, and that he had only taken one day off in the last 40 days. Although the ‘workload’ is extremely heavy, he sees live streaming as ‘leisure’ rather than serious ‘work’:

Now I earn 10,000 RMB per month. To be honest, my previous job got me more

wage, but I enjoy live streaming more. When my fans are happy, I feel happy too. I

feel satisfied when I have a group of fans to support me … We have our own

WeChat group and QQ group, as the length of regular live streaming is even not

enough for me. After streaming I still continue chatting with my fans and organise

group game play. (Interviewee, S5, 29/01/2019)

Another full-time streamer (S4) had been a micro-influencer with 58,000 followers, but had about

150,000 followers in 2020. She started streaming in 2015 on different platforms and joined Douyu in

2017. In the beginning, she live streamed as a part-time job, for just two hours per day. Now she is a full- time streamer and streams five to six hours per day, six days a week, only taking a break on Mondays:

I used to be a live show streamer featuring on a singing and talent show. In 2017, I

joined a new agent company and start game streaming. Now I focus on streaming

playing PC games, including H1Z1, PUBG, League of Legends and Overwatch.

(Interviewee, S4, 27/01/2019)

137

She is very satisfied with her full-time streaming career:

In the beginning, I did it just for money. However, now I enjoy live streaming and

the interactivity enabled by it … I’ve got fans, and even after the live streaming we

still play games together. We also organise game competitions among the

communities for fun. (Interviewee, S4, 27/01/2019)

Among all the interviewees, S8 was the only one who expressed dissatisfaction about the workload.

S8 graduated from a decent university with a Bachelor’s degree in Broadcasting and Hosting. While she wanted to be an anchor, she was too short to meet the minimum height requirement and was rejected by all the employers, which is similar to the situation of S2. S8 changed her jobs frequently in the past few years, with those jobs having nothing to do with anchoring – they were low paid and stressful. Live streaming has given her another chance by minimising the disadvantage of her height and maximising her advantage – eloquence and affinity – partly because of the professional training she has received (Interviewee, S8, 24/07/2020). She became a full-time streamer in October 2019, and now works six days per week and nine hours per day, with three hours’ preparation and six hours live streaming. In her words, ‘It is too exhausting for me to enjoy, given the workload … but I still love doing it, and the pay is really good … my talent and skills fit live streaming’ (Interviewee, S8,

24/07/2020). According to this interviewee, the money she has earned from live streaming, even in the worst period,20 is still higher than she earned in her previous jobs.

It can be seen that some streamers who come from marginal social groups gain gratification from live streaming activities. To find an ideal job in China’s highly competitive society is difficult for

20 S8 is an e-commerce live streamer and she lives and streams at Wuhan. According to her, e-commerce live streaming in her region has been influenced dramatically by COVID-19 epidemic in early 2020. Although it is getting better now, her monthly income is still less than it was in 2019.

138

them, particularly in undeveloped regions that are often characterised by shrinking amounts of arable land and decreasing working opportunities due to urbanisation (Chan, 2010). Live streaming minimises their disadvantages: their educational background does not matter; their physical location does not matter; their height does not matter, as the cam only shows their upper body; their appearance does not matter, as the filter can minimise any defects from their appearance or sometimes even perfect their look during the whole streaming exercise (see Figure 21). Amateur streamers do not need to become celebrities or compete with the celebrities and professional performers; everyone can become an idol for their fan community and enjoy the feeling of being embraced.

Figure 21: Actual appearance and filtered appearance21

In addition, monetisation channels for live streaming are not like the revenue models of other social media, which are dependent on advertising. The revenue of non-celebrity streamers comes mostly from donations rather than advertisements, which of course means these streamers

21 Your Highness Qiao Biluo’s real appearance (left) and one of her filtered faces (right) on Douyu, retrieved from https://www.bbc.com/news/blogs-trending-49151042

139

do not really need a large viewership to be sponsored by advertisements – a group of paying viewers is sufficient. As this thesis indicates, live streaming is the most profitable digital media for users (SCMP, 2019). S2 mentioned that she only had 2000 viewers, perhaps only hundreds of which were paying users. She has not been sponsored by any company, but those hundreds of donors can continuously enable her to earn 5000 RMB per month (Interviewee, S2, 15/01/2019).

Because of the significant regional variation in China, people from remote areas get much lower wages (Finance Sina, 2019), but that also comes with a lower living cost. Therefore, the financial gain from streaming, mostly from donations, could be much higher than the average wage in these undeveloped areas.

It can be seen that the financial pressure of casual streamers can be significantly relieved by what is in effect pocket money received from live streaming. Rather than a winner takes all logic, grassroots streamers can always find a position in the live streaming industry, earning recognition and often extra money.

‘Down to earth’: Streamers and the social needs of viewers

Live streaming does not just benefit grassroots streamers; it also fits the social needs of the viewers in China. The sociability of live streaming helps viewers to feel recognised and empowered.

Viewers are often ignored in their daily lives, but they are valued by streamers and other viewers of the channel. Viewers’ real-time comments can get instant responses from the streamers and other viewers. In this way, the ‘voice’ of the viewers is heard by their idols – streamers – and their buddies – community members.

Compared with an uploaded video that has been through a pre-recording process and post- production, the immediacy of live streaming video lets viewers experience a strong sense of

140

presence and the embrace of the community (Spilker et al., 2020, p. 616). Moreover, compared with professional performers and celebrities who consider themselves superior or aloof, the streamers are jiediqi (down to earth) and therefore tangible for the audience. In this way, streamers as the amateur broadcasters can gain advantages compared with professional broadcasters who provide a staged performance.

P2 was a part-time streamer and then moved to working in the operations department of a live streaming platform. According to P2, her platform invited professional broadcasters to live stream, and the show was good at the beginning. However, after a period these professional broadcasters were reluctant to follow the requests of the audience, which – unsurprisingly – upset their audiences. As the interviewee mentioned, these streamers rejected an audience member

‘ordering’ a song because they were ‘too concerned about their own image’:

They are afraid if they did not sing an unfamiliar song well. They are very cautious,

afraid of making any mistakes or making fools of themselves if they become

imperfect in front of the viewers. They wished to follow the existing narrative

rather than embracing the unpredictability of live streaming. Actually, perfection is

not what audience of live streaming wants. It is the casualness and unpredictability

making the live streaming attractive and ‘down to the earth’. (Interviewee, P2,

04/02/2019)

What makes live streamers appealing compared with other social media influencers is that they are not lofty and thus give viewers a sense of intimacy (Interviewee, P2, 04/02/2019). This intimate relationship appeals to all, as recounted by the experience of P2 when she was a streamer:

141

I feel good even when the viewers are making fun of me – a viewer once posted a

screenshot of me having dinner, he wanted to embarrass me but I argued back. It looks

like flirting to me, and flirting like this is entertaining for me and all the audience; we

do not have to be too serious when streaming. (Interviewee, P2, 04/02/2019)

The gratification arising from this kind of close relationship is probably magnified by the gender imbalance in the Chinese context. This is a concerning issue in China, where a huge number of single males from disadvantaged social groups exist, working alone in the first- and second-tier cities. For them, watching and interacting with female streamers becomes a way to deal with their loneliness and sense of loss (Cunningham et al., 2019). While pretty girls in real life tend to ignore these single, less- advantaged males, in streaming contexts female live streamers are willing to talk to them directly, dance, sing for them or even flirt with them. This gives these grassroots viewers a high level of satisfaction. This can be amplified by donation when viewers tip the streamers, so that they can receive extra attention and responses from streamers, as well as gaining attention and admiration from other viewers. For instance, when the donation from a viewer ranks highly, they will then be identified and highlighted on the channel ranking board; this draws audience attention (Chen & Chen, 2020). Studies show that the young, low-income population enjoys the projection of their donating and texting on the bullet screens, providing them with a sense of control and overcoming feelings of inferiority in their everyday lives (Song, cited in Cao, 2019; Fan & Zhang, 2018).

Grassroots viewers also often find that watching live streaming content resonates because of the sense of proximity. In China, these grassroots viewers’ daily lives are not usually included in the mainstream mediascape (Sun et al., 2020). However, because of its authenticity, live streaming provides images of their lifestyle and particular ways of entertainment (Lu, 2019) that usually would be ignored by the mainstream media and other forms of social media (G.C. Li, 2020,

142

p. 23). For example, Lin and de Kloet (2019) point out that grassroots streamers are mobilised on live streaming platforms to create a form of rap, called Hanmai (shouting with microphone). As a category of live show, Hanmai uses lyrics and high volume to shout out anger at discrimination and social inequality (Xu & Zhang, 2020). Tianyou, one of the most well-known Hanmai performers, grew up in a small city from north-eastern China that has not benefited much from the rapid economic growth that has occurred in many other places. He dropped out of middle school and struggled for many years until he started streaming on YY in 2014 (Xu & Zhang, 2020). The lyrics of his Hanmai describe the betrayal of his girlfriend, the ‘reality’ of building a relationship based on money, defeating the emperor, conquering the battlefield – all topics that resonate with his fan base, who are also mostly from the grassroots. This proximity to a huge audience base has made him famous and rich – as the data show, he can earn more than US$2 million a year (Lin & de Kloet, 2019).

It should be noted that Hanmai, and other forms of live streaming content including hunting, rapping, complaining and flirting, still have a reputation for being ‘lowbrow’ culture and

‘vulgarity’; neither of which is favoured by the official media of China, and which don’t appeal to mainstream popular culture (Chen & Chen, 2020; G. Zhang, 2019; Zhang & Hjorth, 2017).

Nonetheless, this content indeed fits the needs of the grassroots, partly because it is sensational and entertaining, and also partly because of the proximity – the streamers and viewers have similar lifestyles, working conditions, frustrations and ways of amusement. This authentic representation of streamers’ ordinary lives or daily thoughts, which may be regarded as lowbrow content from an elitist perspective, can often gain the attention of disadvantaged social groups, striking a chord with these viewer communities. As noted by P2:

143

There are many streamers providing daily makeup tutorials and random chat on our

platform, these forms of content can gain supports from viewers. A celebrity

streamer has tried to stream something elegant such as poetry reading, but the

audience stopped watching her soon … You must have self-mockery and be

‘jiediqi’ to fit the needs of the mass. The professional broadcaster who cannot get

rid of their pride and sense of superiority cannot be a successful streamer … Well,

the newly graduated professional broadcasters may have a chance, but it is

impossible for the experienced broadcasters to get used to the style of live

streaming. (Interviewee, P2, 04/02/2019)

As studies show, viewers from live streaming channels see themselves as fandoms from the subculture and members of marginalised communities (Chen & Chen, 2020, p. 731). To some extent, particular live streaming channels that are formed and organised by grassroots streamers have become a ‘self-marked space that allows connection to all fellows of the same taste in the same camp’(Chen & Chen, 2020, p. 736). These channels become individual domains for gathering the like-minded grassroots, rather than just a stage of performance for streamers. In short, grassroots viewers find resonance with streamers, streaming content and other viewers, particularly when the scene and living style of the live streaming content is similar to the everyday life of viewers. This resonance and synchronised daily interaction make streamers and viewers feel as if they are being recognised, which is why some streamers and viewers tend to refer to their communities as families.

144

The state as a key stakeholder for poverty reduction

It has been pointed out that Chinese state is not only a regulator but also a facilitator of the digital economy; it supports the development of digital media when they are in alignment with the goals of economic development and cultural empowerment (Hong, 2017b; Wang, 2017; Z. Zhang, 2020).

In the earlier stage of live streaming, the Chinese state adopted a wait-and-see attitude towards it, partly because of the proliferation of the vulgar content and also partly because of the unpredictability of live streaming content. One of the interviewees, P2, previously worked as a content regulator in an official broadcasting system. According to P2, while other forms of social media have been adopted by the official broadcasting systems for political, economic and cultural purposes, not many governmental agencies have adopted live streaming. As she explained, live streaming is inherently problematic for the official broadcasting system to govern:

Are you going to preview all the comments before releasing them? If so, numerous

monitors are needed, and there would be an inevitable delay. If we release the

comments before reviewing them, who is going to take the responsibility for those

illegal comments? It is just so unpredictable. Our TV station has planned to conduct

some news live streaming, as it obviously can draw more attention especially from

youngsters. For example, after arriving at the spot, we can listen to the proposal of

the live viewers: what is your focus? Who do you want to interview? What

questions do you want us to ask the interviewee? It could be so interactive.

However, we still need a photographer, a journalist, and also a director for the

viewership management and communication. Another question arises, we cannot

ask prepared questions to the interviewee like we were in the traditional interview,

because the viewers might say, ‘It is scripted – the journalist taught them to say so.’

145

And it would also be problematic if an official broadcasting programme is led by

viewers in the whole process. (Interviewee, P2, 04/02/2019)

Another reason for the hesitation of officials when considering the adoption of live streaming is that donations from live streaming are difficult to ‘manage’. P2 has indicated that some official

TV stations have live streaming channels, but none of them has opened up the donation system, and R2 mentioned that receiving donations was risky in the existing financial system. As she knows, an anti-corruption campaign has been implemented for curbing rampant corruption, and any form of revenue in the official system has to be calculated strictly. While there is an existing, comprehensive calculating and auditing system for other forms of revenue, streaming donations are too new to be managed through the old system. This means that the revenue brought in from donations could easily lead to corruption or tax evasion (Interviewee, R2, 16/07/2019). Moreover, if donations cannot be calculated precisely, it could also become a new method of money laundering; there is already a case that has outlawed the use of a commercial live streaming platform due to money laundering (Huai, 2020).

From suspicion to support: ‘Peppa Pig’, e-commerce streaming and a shift in attitude

After years of the evolution of regulatory policies towards live streaming, and particularly after the COVID-19 pandemic, the Chinese state’s attitude to live streaming has changed. A remarkable turning point was the ‘Peppa Pig’ e-commerce live streaming event facilitated by CCTV. In 2020, because of the coronavirus outbreak and two months of lockdown, Hubei Province suffered a strong economic hit. Then, on 18 March 2020, the Central Cyberspace Administration, the

National Development and Reform Commission, the State Council Poverty Alleviation and

Development Office and MIIT cooperatively published Key Points of Online Poverty Alleviation.

This document set up a plan to further improve ICT infrastructure in rural areas, building up and

146

promoting e-commerce services, online education and online medical diagnosis for rural communities by the end of 2020 (CAC, 2020). In order to boost the economic recovery, CCTV launched a charitable e-commerce live streaming on 6 April, two days before the lifting of the lockdown of Wuhan (Global Times, 2020). Two famous anchors were invited: the first was Zhu

Guangquan, an official news anchor who has worked at CCTV for 17 years; the other was Li Jiaqi, one of the top e-commerce live streaming celebrities. Li Jiaqi is known as the ‘King of Lipstick’, and has the best sales of cosmetic products in China. The combination of the pronunciation of their names, ‘Zhu and Qi’, is similar to the Chinese pronunciation of ‘Peppa Pig’. Therefore, Chinese netizens gave the name ‘Peppa Pig’ to this team to express their love. Their streaming went viral online and 40.14 million yuan ($5.7 million) worth of Hubei local products were sold in just two hours. Some Chinese netizens expressed concern that they were unable to fight the virus on the frontline, so purchasing local products was seen as the best way for them to make a contribution to the recovery of Hubei Province (Global Times, 2020).

After this initial move from the central media, many local official media and peasants started to promote their local products by live streaming. Chinese peasants suffered from the lockdown during the COVID-19 period, and faced challenges to sell their agricultural products.

But the adoption of live streaming significantly relieved this issue, and in some cases even helped peasants from rural areas escape poverty and become better off than before (Z. Li, 2020; Ma, 2020;

M. Shen, 2020). According to the report, Taobao launched the ‘Help Farmers with Love’ event to help the farmers in the beginning of 2020. Some 60,000 peasants had started their e-commerce live streaming on Taobao from this event by the end of March 2020, turning their unsaleable agricultural products into bestsellers and providing some relief to their financial pressure (iiMedia,

2020). De Song, a peasant from a village of Guangxi Province, live streamed the process of

147

harvesting his sweet potatoes. He introduced his product and interacted in a lively way with viewers. With the cooperative promotion from local government and local agent companies, he gathered tens of thousands of viewers. He claimed that more than 50 per cent of his agricultural product was sold via e-commerce live streaming; much of it was sold to the economic developed zone in China (M. Shen, 2020).

In short, after years of attempts to steer live streaming, the state is now trying to integrate these previously unpredictable and uncontrollable practices into its media regulation ecosystem to achieve political and economic goals. As studies note (Z. Zhang, 2020; E. Zhao, 2019, p. 12), the

Chinese state is continuously playing as a stakeholder in the digital media market. And in the case of live streaming, state involvement encourages convergences between live streaming platforms, online sales, e-commerce platforms and portable devices. By the end of 2020, reports indicate that e-commerce live streaming has become the fastest-growing genre in the industry, with about one- third of the internet population in China as users, despite widespread dissatisfaction with the goods purchased through live streaming channels (Deng, 2020). Yet it is clear that the poverty reduction project has usefully been facilitated by e-commerce live streaming, and that people in rural areas with saleable products can benefit from this change. But the long-term consequences of the popularity of e-commerce live streaming and the potential regulatory risk remain in question.

Summary

Arguably, live streaming in China provides us with new insights into the wider discussion of the affordances of social media platforms and technologies. Rather than enabling pure entertainment and commercial activities, live streaming is making a contribution to the rebalancing of regional development in China. As David et al. (2017) note, live streaming enables a new politics among

148

those marginalised groups subject to dominant networks of power. Live streamers have access to a new mode of economic counterpower. In the Chinese context, this new mode of economic counterpower is manifested by the ability to create an alternative production system to challenge existing network capitalism. This easy-to-access, flexible and authentic means of video production has provided the grassroots with a chance to let their voices be heard, to help them fulfil their dreams, build fan communities around them and earn extra money to improve their social and economic conditions. In short, live streaming has the potential to fill a gap and reduce existing polarisation in China, focusing on the mental and material demand of both streamers and viewers who typically are marginalised during China’s rapid and unbalanced development. This process is driven by various stakeholders, including platform operators, streamers and government agencies.

Although the state finds live streaming risky for social stability and holds a cautious attitude towards it, it nonetheless embraces and even fully supports the use of live streaming for solving economic issues arising from the COVID-19 pandemic. However, it has to be noted that this alternative production system is very likely to be harnessed by digital capitalism in the long term.

Current state support has been based on a sense of entrepreneurial solutionism (Keane & Chen,

2019) and the premise that live streaming fits existing social needs and the wider goals of maintaining stability. Therefore, to say the grassroots would be continuously empowered by live streaming is an overstatement. Another point to make in this context is that, as e-commerce live streaming becomes the fastest-growing genre as facilitated by multi-stakeholders, it therefore calls out for new questions about the impacts of commodification and consumerism, particularly as selling and promoting goods has emerged as the major content of live streaming.

149

Rethinking the audience commodity: An audience-led narrative and the unconscious ‘broadcasters’

Audience commodity theory and the change of monetisation channel

In 1977, Dallas Smythe’s (1977) famous theory of ‘audience-commodity’ provided a theoretical framework to account for how media users are commodified in the capitalist economy. The primary product of the media, he claimed, was not the media content but the commodified audience.

Within this view, media produce the content – also known as a free lunch – to then recruit viewers.

And the viewers – or, more specifically, the attention and purchasing power of the viewers – become the real commodity that is sold by media to advertisers (Smythe, 1977).

While this theory seemed to be not fully applicable to digital media, new theories have been proposed to describe the process of the commodification of digital media users in contemporary society. For example, the terms ‘digital labour’ or ‘free labour’ have been used to describe the spontaneous, voluntary activities of internet users. These pleasurable, productive activities and the consciousness of users are used by media and advertisers to advertise their products, thereby the users of digital media are working and being exploited all the time (Cohen,

2008; Fuchs, 2014a, pp. 34–36; 2014b; Terranova, 2000). In a similar vein, immaterial labour from users is embedded in the human interaction and communication, which can be both intellectual and affective labour, creating immaterial products such as knowledge, feeling, relationships and emotional responses (Hardt, 1999; Hardt & Negri, 2000; Lazzarato & Virno, 1996).

Further studies also claim that play, labour, production, consumption and usage are all intertwined when users are using digital media. For example, Julian Kücklich (20005) notes that users are working and devoting labour to capitalist production even if they are enjoying the usage of digital media. Therefore, play and labour are entangled in the process, thereby contributing to

150

the concept of “playbour”. Axel Bruns (2008) combines the words ‘production’ and ‘usage’ into

‘produsage’ to frame the spontaneous work of participatory audience on digital media. His conceptualisation was triggered by the word ‘prosumption’, used to describe the increasing convergence of production and consumption (Toffler & Alvin, 1980), particularly on social media

(Fuchs, 2014b; Tapscott & Williams, 2008).

Most of these studies critique the commodification process and the exploitation of user labour on digital media, so live streaming activities may also fall into these categories of commodification. One strand of research has pointed out the distinctness of the commodification of users on live streaming platforms, and its potential for empowerment. Van Ditmarsch (2013) claims live streaming has the possibility of empowering audiences and advocating a sense of participatory democracy, while Walker (2014) believes the participatory qualities of live streaming do not prevent it from being exploitative. The design of streaming platforms and the marketing strategy of their owners can influence the utilisation of streaming technologies, which makes it exploitive, as Walker (2014) warns. Some studies have looked specifically at the commercialisation of live broadcasting. Shaw et al. (2014) and Smith et al. (2013) examine the ways live streaming platforms insert advertisements into broadcasting content. The role of the streamer on live streaming platforms is similar to that of social media influencers, although streamers are regarded as more trustworthy (Gräve, 2017) and are more real (Khamis et al., 2017) than traditional celebrities when promoting products. The process of commodification and labour exploitation on Western live streaming platforms has been examined (Bingham, 2020; Taylor,

2016; Woodcock & Johnson, 2019a), showing a similarity with the exploitation that is prevalent on social media: users create economic value for the platforms but only get a small portion of compensation as a return, because the platforms control the means of production and distribution.

151

There are similar findings about the commodification on Chinese live streaming platforms.

In a Chinese context, Ding (2016) points out that some live-streaming platforms in China are using deliberate commercial exploitation to advertise themselves. Most recent case studies on Chinese live streaming platforms show how the framework of Chinese live streaming is designed specifically for promoting advertising and donations (Cunningham et al., 2019; Zou, 2018).

Moreover, to deal with advertisements and constant policy changes of platforms and government, some streaming celebrities in China even employ brokers to handle their daily affairs with streaming websites (He, 2016), which further accelerates the professionalisation and commodification of live streaming (Zhang et al., 2019).

Drawing from these studies, there is no doubt that user data, attention, activities and their purchasing power are sold to the advertisers in the commodification process on social media

(Andrejevic, 2007; Gräve, 2017; Hearn & Schoenhoff, 2016), and live streaming is no exception.

It is also clear that although both streamers and viewers can gain gratification in streaming activities, they constitute a sense of playbour that is being exploited by platforms. However, merely using these existing frameworks to understand Chinese live streaming overlooks the fact that the role of advertisers is absent in China. In general, the advertisers should be the main buyers who purchase the commodities, but what if advertisers are less engaged in the process? It is known that live streaming communities can be extremely hostile to the traditional forms of advertising shown in streaming channels (Woodcock & Johnson, 2019b, p. 328). Similarly, in China, some platforms even cut off advertisements: by 2019, according to P1 (Interviewee, P1, 31/12/2018), YY had prohibited any forms of advertisement and decreed that its main revenue should come from donations. We can then ask, ‘If the monetisation is not through advertising, and the content is free to view without paying, then who is the buyer of the commodity and who becomes the commodity?’

152

As Chapter 2 of this thesis examined, while donations are only just emerging as a way of monetisation on Western live streaming platforms, Chinese live streaming platforms are highly dependent on donations. If donations rather than advertisements are the key point for the commodification of live streaming in China, can we still use the existing frameworks of audience commodity or digital labour to understand the live streaming activities? Notably, the decreasing significance of advertising is not only a phenomenon that limited in the scope of live streaming, as Winseck (2017) notes, but subscription fees and the pay-per model are increasingly becoming the main source of monetisation for media companies in the West when compared with advertising.

This change in the mode of monetisation has enormous implications for political economic studies, so we need to rethink the concept of ‘audience commodity’ in current contexts.

From these points, this section argues that, as one of the distinct features of live streaming, donations might constitute a new form of commodification and labour exploitation, particularly when the role of traditional advertising is less significant. The role of donations in the Western context has been examined by scholars (Johnson & Woodcock, 2019a, 2019b; Taylor, 2016).

Woodcock and Johnson, 2019a), who claim that both streamers and viewers are devoting affective labour, as a form of immaterial labour, to produce ‘a feeling of ease, well-being, satisfaction, excitement, passion-even a sense of connectedness or community’ (Hardt, 1999, pp. 95–96) for the platforms. In a similar vein, donations on live streaming are regarded as paid emotional labour to satisfy viewers (Guarriello, 2019). Partin (2020) takes a holistic view of platformisation and live streaming, particularly how Twitch facilitates the practice of donation in its own ecosystem. He notes that on Twitch the change from donating through PayPal to donating with Twitch’s Bit (the virtual currency that can only be purchased and used on Twitch), and the wide adoption of third- party donation management services, mark the digital enclosure facilitated by Twitch (Andrejevic,

153

2004; Partin, 2020). Compared with the abundant research on Twitch in a Western context, a more nuanced, updated and comparative analysis is required of how China’s political, social and economic circumstances are impacting the commodification process of donation.

The enclosure of virtual currency in China

While viewers on Twitch can still donate real money through Paypal in addition to virtual gifts

(Partin, 2020), in China only virtual currency transactions are allowed on the platform. From 2007 to 2009, about 10 Chinese regulatory authorities, including the Ministry of Culture, the Ministry of Public Security, the Ministry of Information Industry and General Administration of Press and

Publications, jointly issued several regulations to curb money laundering, illegal trade and online gambling. In particular, the transaction between real currency and virtual currency is forbidden privately among players/users, which means that it is illegal for a player/user to convert virtual currency or virtual item into real currency with another player/user and vice versa (Douyu, 2019, pp. 39–40,163–64). Therefore, according to current policies, direct donations of RMB to streamers are prohibited on Chinese live streaming platforms, so the only way is to purchase virtual currency with RMB is to donate virtual currency to streamers. Similarly, when streamers get a substantial amount of virtual currency, they cannot trade their virtual currency into real currency with other streamers. The only way is to redeem the donations from the platform, as the platform is specifically licensed to handle the transaction (Douyu, 2019, pp. 39–40). Under these circumstances, although the initial purpose of the Chinese government was to regulate illegal transaction activities, and the platforms were under the scrutiny of the government to make sure the rate and means of transactions were reasonable (MCT, 2009), there is no doubt that the power asymmetry between users and live streaming platformed has been extended. Once all the transactions are controlled by the platforms, streamers will be unable to gain the full price of their

154

donation, as it will only be the platforms that can receive the real currency and decide the distribution. According to interview data, streamers often do not even have a clear idea about the exact amount of virtual gifts they have received. When affiliating with the agents, the live streaming platforms will only pay the agent company real currency, depending on the amount of virtual gifts, instead of paying the streamers directly. In this circumstance, the money from donation has been transacted twice before it reaches the hand of streamers. As interviewee A1 claimed, platforms distribute 30 per cent of the donations and his agent company gets 70 per cent.

Streamers then share between 50 per cent and 70 per cent of what agent company receives, which amounts to approximately 50 per cent of the overall donations (Interviewee, A1, 14/01/2019). In addition, other forms of income, including bonuses for good performance and commission fees, are paid to agents, with agents being the final arbiters to decide on the cut to streamers (Interviewee,

A1, 14/01/2019). In this way, platform companies and agents form an opaque, enclosed ecosystem that absorbs real currency, as well as distributing the revenue from donations to streamers.

Figure 22: Flows of virtual gifts

155

Two paradigms of commodification in live streaming

As examined in the first section of this chapter, grassroots streamers still maintain a high level of autonomy of production in this enclosed system. This section, however, argues that the distribution of their live streaming videos, and the interactions between them and the viewers, are in the hands of other stakeholders. The conflicts between independent production and dependent distribution of the content result in complexity in the commodification of live streaming activities. Analysing the commodification of Chinese live streaming in a multi-stakeholder system helps to avoid a one- size-fits-all approach. Instead of seeing live streaming as a stationary and unified entity, here I propose understanding live streaming activities in China through two dynamic and intertwined paradigms, depending on the level of commodification and the involvement of stakeholders other than streamers and viewers.

A content-based paradigm: Live streaming as enhanced street performance; donation as reward

The first paradigm delineates the majority of live show activities, which are usually conducted by grassroots, or non-celebrity, streamers for live show performances. While in the Western context these smaller streamers would not earn much money, in China the situation is different. As examined in Chapter 2, the genre of live shows in China has a shorter history but has gained much more popularity than its Western counterparts. Compared with other genres of live streaming content, live shows are more engaging and can earn significantly higher revenue in terms of gaining virtual gifting in the same period (Lan, 2019). For example, while game streaming has been dominating the number of channels and viewers, on Douyu non-game live streaming – mostly live shows – earns 2.5 times the revenue (Douyu, 2019, p. 127). The key to understanding this difference is the intimate relationship between community members and the high frequency of live

156

interactions, which tends to improve the viewing experience of live streaming, resulting in viewers being more willing to donate.

Studies of Western platforms have shown that the smaller the community is, the closer the relationship is between streamers and viewers: streamers have a closer relationship with their fans and maintain a high level of trustworthiness among the fan communities (Gräve, 2017; Khamis et al., 2017; Senft, 2013). On Twitch, the relationship between viewers and streamers in smaller channels has been proved to be more intimate, and small channel viewers were more socially motivated than those watching celebrities’ channels (Hilvert-Bruce et al., 2018). While it looks like this phenomenon contradicts the network effect, the reason is not difficult to understand: the paucity of attention from an individual streamer. As examined in Chapter 3, live interaction is the main component, in addition to the originality of the show, and in a small community of organised and maintained micro and nano live show streamers, almost all the individual viewers maintain a close relationship with streamers and are able to interact with them. However, this could be totally different for a celebrity streamer. As Recktenwald (2016) notes, while the attention from viewers to the streamers is abundant, the attention and response from streamers to viewers is a scarce resource: numerous viewers can post chats, send emojis or donation tips to streamer at the same time, but streamers can only respond to individuals one at a time. It is therefore impossible for celebrity streamers to give sufficient attention to every viewer if the number of viewers is high.

The disadvantage of micro and nano streamers, with far fewer viewers, thus becomes an advantage.

In a small community maintained by micro or nano streamers, their live viewers are usually only around one thousand or even less in number, and the active viewers who actually interact with streamers are even less. Therefore, the attention from streamers can be distributed to almost every active viewer.

157

In a similar vein, game streamers have to concentrate on playing games instead of giving sparing attention to viewers. Interviewees S4 and S5 are game streamers, and as S5 noted, sometimes game streamers might be too immersed in the game, so they inevitably overlook some viewers:

I do not have a chance to look at the comments when I am engaging in combat in

the game, so I might overlook some viewers. However, I will respond to the

viewers when I have a break … or when I am streaming singing. (Interviewee, S5,

29/01/2019)

As S4 noted, in order to cater for all viewers while she is using her computer for gaming and streaming, she has to have her iPad with her to look at the responses from viewers and to capture all their comments (Interviewee, S4, 27/01/2019). Interviewees S4 and S5 told me that although they did not provide enough attention to viewers, they had a focused response: they appreciated donators every time they received comments and gift. For example, live show streamers such as

S1 and S3, who received less than one thousand viewers, said they responded to all comments and donations from viewers unless the comments involved harassment (Interviewee, S1, 15/01/2019;

Interviewee, S3, 23/01/2019).

I therefore argue that this paradigm of live streaming is content based rather than platform based or bandwidth dependent. The live streaming activities of these non-celebrity streamers can be seen as a form of street performance: streamers are the buskers, performing their skills and talent in front of the viewers. Viewers watch the show, cheer and tip streamers, and spontaneously reward them when the show hits the mark. In addition, every viewer has a chance to interact with streamers and to be part of the show: they can interact with streamers and gain instant responses;

158

donating even the smallest number of tips can prompt appreciation from the streamers. In this sense, the platform is the pitch – the place for the show – and also the container – the box collecting the tips – while sharing a cut from the tips as a commission for the usage of the pitch.

In short, for these smaller streamers in the earlier stage of a live streaming performance career, advertisements and platforms are not so much involved in the production of live streaming activities – streamers enjoy a high degree of content production. Although this looks as if it constitutes a sense of immaterial labour or playbour, as the UGC is created by users but owned by the platform, using the framework of traditional labour exploitation to understand this practice is not so effective in these circumstances because this content-based paradigm is less capital intensive and less profitable for the platforms (Winseck, 2017, p. 98). In addition, to argue that UGC created by viewers and streamers is sold to the donators is an oversimplification that ignores the engagement of other stakeholders such as platforms and agents. Therefore, we need to look more closely at another paradigm of live streaming in order to understand the role of other stakeholders and other mechanisms of commodification in Chinese live streaming.

A platform-based paradigm

Platforms as overpowered distributors

With the increasing popularity of live streaming platforms and a growing number of streamers, the pitches for performance on live streaming platforms became overcrowded. It is impossible for viewers to have a quick sample of all the channels before selecting one to watch. In this sense, channels that are listed on the front page or highlighted on the loading page gain a significant advantage in catching the attention of viewers. In this way, platforms as content distribution mechanisms have outsized power to assign streamers to these popular pitches.

159

The highlighted channels on the front page of the platforms – something deemed to be dependent on the real ‘popularity’ of the channels – are easily manipulated by the platforms in order to control streamers or charge a higher commission fee for these popular ‘pitches’. According to interviewee A2, although the recommended channels on the front page are highly dependent on the number of current viewers and the amount of donations, these can be influenced dramatically by certain payments:

The platform offers us (agent company) a ‘flow package’, which can be a boost of

one thousand viewer numbers to the target channel. In addition, the hotspot – front

page channels can be purchased. Streamers who have a high amount of viewers and

donations can take the front page positions, but only for a few minutes if they did

not pay extra money to the platform. But if the position is purchased by agent

companies like us, the streamers can occupy the hotspot for a very long period.

(Interviewee, A2, 15/01/2019)

A1 also confirmed that the daily recommended channels on the front page are decided by the platforms, and usually sold to agents (Interviewee, A1, 14/01/2019). This form of revenue from the ‘flow package’ becomes part of the ‘donation revenue’, which partly explains why the amount of donations in Chinese platforms is so much higher than for their Western counterparts.

The front page of Chinese live streaming websites is also displayed differently compared with Western platforms. If we open the front page of Twitch, we can see the thumbnails of some recommend channels and content. However, if we open Chinese live streaming websites such as

Douyu.com and YY.com, a ‘popular’ live streaming channel will be opened and shown to us directly, together with a few other recommended channel thumbnails (see Figures 23 and 24).

160

Figure 23: Frontpage of Douyu22

Figure 24: Frontpage of Twitch23

Consequently, viewers have to take a look at the recommend channels every time they enter the website. By directing the attention of viewers in this coercive way, platforms further strengthen

22 Retrieved 4 November 2020 from Douyu.com. 23 Retrieved 4 November 2020 from Twitch.tv.

161

the advantage and the monetisation opportunities of their popular pitches. As a report from Douyu indicates, the user clickthrough rate that follows content recommendation on the home page increased from 18 per cent in June 2016 to 49 per cent in March 2019 (Douyu, 2019, p. 152).

Through highlighted or recommended channels, platform companies maintain strict control over content distribution. In this way, platforms have emerged as an overpowered allocator in this

‘street performance’. Streamers, like street performers, have to pay to obtain popular pitches in order to gain more attention and donations from the audience. For smaller streamers, the agents might be willing to pay for a ‘flow package’ to boost the popularity of streamers (Interviewee, A2,

15/01/2019; Interviewee, A1, 14/01/2019), but for celebrity streamers the cost of boosting their popularity is set extremely high, so streamers themselves need to pay part of it, which often creates considerable financial pressure for streamers (The Paper, 2020).

In short, rather than selling the attention of viewers to advertisers like other social media platforms, live streaming platforms sell popular pitches, and these present viewer attention – and the possibility of donations – to streamers. Viewers are commodified in this process, and streamers rather than advertisers are the purchasers of this commodity.

Platform and agents as ‘the bottler’

In a street performance, the street performer usually has a container to collect the donations, and in some cases a collector or so-called ‘bottler’ is included as an assistant to help the performer to collect them. A good bottler can encourage audience members to donate more money, and as a return the bottler can split the donations they receive (Reeve & Reeve, 2011). To some extent, a live streaming platform is not only a pitch distributer but also a successful bottler, using a range of measures to please donors and thus stimulate donation.

162

The latest subscribers and donators, and the daily and monthly top donations, are always highlighted on live streaming platforms and channels (Reeve & Reeve, 2011). This shows that the underlying design of platforms is far from neutral, and can more accurately be seen as an instrument for its developers (Bucher, 2012; Gillespie, 2014). Essentially, it is devised to stimulate donation by ranking and promoting paying viewers, highlighting the donors and tips, and using gamification to make the practice of donation more immersive (Johnson & Woodcock, 2019a;

Partin, 2020; Zhang et al., 2019; Zou, 2018). Interviewee S2 commented that it is an interactive game for her to lure the donation:

There is a prize wheel in my channel, and viewers need to become a member (pay a

monthly fee) to participate in the game. I spin the wheel and check … the ‘prize’

for us depends on the result. It could be dancing or singing, so I have to dance or

sing. When the result is a donation, participants have to donate the virtual gift to

me. (Interviewee, S2, 15/01/2019)

From the interview data, the majority of streamers claim they paid more attention to the donors and responded to them more often, and these frequent donors would then be highlighted in the channel with announcements whenever they entered the channel (Interviewee, S1, 15/01/2019).

So donors are placed under a spotlight and can enjoy both the admiration of viewers and appreciation from streamers. However, it is becoming increasingly difficult to be put under a spotlight, given the scarcity of streamer attention and the growing numbers of viewers. This leads to a kind of donation competition among viewers – a behaviour that can commonly be witnessed on the live streaming channels (Sjöblom et al., 2019; The Paper, 2020). This donation competition can be regarded as a battle between viewers for scarce attention from the celebrity streamers

(Recktenwald, 2016), whereby viewers use an increasing amount of donations to be a top donator

163

in order to win the limited amount of available attention from the streamer. This form of competition can be more fierce on celebrity streamers’ channels, and the agent companies are a peculiar actor in the ecosystem, further intensifying this competition. Zhang et al. (2019) notes that ‘fake gifting’ is used by agent companies to promote streamers and to provoke the audience to donate more virtual gifts. Agent companies purchase virtual gifts from the platforms at a discounted price, then create multiple fake viewer accounts. These accounts keep donating a huge number of virtual gifts to streamers so that the real viewers are provoked into donating further in order to gain attention from the streamers. This mechanism of ‘fake gifting’ and the underlying motivation has been further revealed in my interviews. As interviewee S2 noted, in one competition launched by the platform between streamers, the streamers performed the show, and the amount of donations, numbers of viewers and posts were the final evaluation criteria for selecting the winner. When she entered the top 70 nationally, all of the agent company staff, and other streamers afflicted with the agents joined the channel to boost her popularity and increase donations for her by cheering and making donations. So the real viewers of interviewee S2 donated more to express their loyalty and support (Interviewee, S2, 15/01/2019). Interviewee A1 also noted the underlying logic behind the fake gifting:

The agent companies usually donate ‘fake gift’ to streamers. We can get a discount

when purchasing these gifts from the platforms, and a high proportion of the virtual

gift would be returned to us later on. These gifts are used to lure the real

donators … approximately 95 per cent of the value of the fake gifts would be sent

back to us after the being donated to the platform, and companies can use these

gifts to donate again. The popularity of the streamers is boosted so it is easier for

them to be the recommended channel on the front page; this can also draw more

164

audience and provoke real donations from the audience. Moreover, it also makes

the traffic and the financial reporting of platforms look good, so the platform can

use this good-looking data to seek more investments. Everyone benefits from it.

(Interviewee, A1, 14/01/2019)

It is clear that the donors are not included in this category of ‘everyone benefits’, as it is they who pay. The same amount of donations is getting less and less influential because of the manipulation of the agents and platform companies, so the donors have to donate more to gain approval from streamers.

In short, the infrastructure of Chinese platforms works in a similar way to Twitch, which aims to ‘maximise content monetization through virtual gifting’ by commodifying the virtual relations among the community (Lin & de Kloet, 2019; Zhang et al., 2019). This echoes the earlier statement about the nature of platformisation, whereby platforms tend to turn all traditional non- tradable services into tradable services for monetisation (Srnicek, 2017, p. 42). Together with the affiliated agents, platforms develop their own evaluation mechanisms and algorithmic logic to commodify the intimate relationships. Deciding the amount of donation becomes a key metric for relationships and a prerequisite for the pleasure. In this way, the power of the distribution is embedded in the digital objects of the infrastructure of the platforms (Fuchs, 2014b; Gehl, 2014); pushing viewers and streamers to follow platform values, and further intensifying unequal and exploitative relations.

The unconscious performativity of audiences

In the street performance, a viewer could be invited to volunteer in part of the show, and live streaming inherits and carries forward this tradition. The channel’s audience is not only the

165

spectator but also the performer. This can be understood from two aspects. First, viewers are not passive participants who follow the existing narrative of the show; one important change in live streaming is that the real-time interactions can in turn affect streaming narrative. Broadcasters’ performance, including activities, voice, dressing, facial expression and their responses to the comments, are changing over time, based on the interactivities with audiences (Smith et al., 2013).

In this sense, the active audience that is engaged in the interaction can significantly influence the streaming narrative and thus the viewing experience of the whole audience. As Smith et al. (2013) claim, ‘It is common to hear the performers answering questions from viewers and reacting to their suggestions.’ In this sense, streamer–viewer interactions are not just an affiliation or a bonus of the streaming content that is performed by streamers, but also a way for viewers to influence the streaming narrative. This synchronous communication connects viewers and streamers in virtual places simultaneously (Powazek, 2002, p. 11), while putting the viewers on the front stage as well.

The ‘performance’ of viewers includes donating virtual gifts and being highlighted, ordering a song and encouraging the streamer to sing it, or discussing gaming strategies with streamers for winning the game, all of which make the live streaming more vivid. Moreover, when streamers express their appreciation to donors, other viewers also tend to express their admiration by posting comments such as ‘so generous’ or emojis of appreciation (Interviewee, S2, 15/01/2019;

Interviewee, S4, 27/01/2019), further satisfying the vanity of donors and stimulating continuing donations.

Second, viewer-generated content is part of the streaming content for viewers. While the streamer-generated content can be regarded as the first screen content, viewer-generated content such as bullet screen comments, donations and notifications can be seen as the second screen content that viewers are watching simultaneously. To some extent, the streaming content is itself

166

a catalyst of the live interaction; the streaming channel can be considered as a space for gathering like-minded people who have similar daily routines and habits, while the viewer-generated content might be seen as the main event. The point has been made that viewers claim they are excited when they see the live comments echoing their own thoughts or sentiments, and they are then motivated to further express themselves through live interaction (Cao, 2019). It is also notable that viewers’ interactions can be widely seen in Chinese live streaming platforms, even without the presence of the streamers. Unlike Western live streaming, the live comments on Chinese platforms are usually displayed in full on the screen, and can even block the original streamer-created content.

The comments can be more interesting than the performance of the streamers, and some viewers claim they are there not to watch the streamer, but for the comments and socialising (Cao, 2019).

There are reported cases where millions of viewers have watched a streamer sleeping (Yuan, 2020), and even cases where the streamer will turn their camera off and just chat with the viewers

(Y. Zhang, 2020). In these cases, the role of streamers is downplayed, and the audience members themselves become the real ‘broadcasters’ who are actually performing.

In these ways, audiences have become the unconscious performers of live streaming, creating essential components for the show. This form of UGC can boost the popularity of the channels, facilitate virtual gift donation and increase traffic on the platform. However, this viewer- generated content is not paid for in this form of immaterial labour, and indeed sometimes viewers even donate for the gratification they gain from these interactions (Yuan, 2020). In this sense, viewers create value for the platforms while still paying for the content they have created themselves.

167

Summary: The commodity audience and donation inflation

To more fully capture the politics of Chinese live streaming, it is essential to realise that there are two different paradigms of live streaming operating simultaneously. In the content-based paradigm, the activities of streaming are like an enhanced street performance, whereby smaller streamers can respond to every viewer and all viewers have a relatively equal status to enjoy the live interaction, donating voluntarily as a reward for good content. This paradigm of streaming is interfered with less by platforms, which is understandable given its lower profitability and reduced capital investment. Although it entails a sense of playbour, platforms are mostly operating as pitch providers, without relying on donation revenue in this paradigm of live streaming. As examined earlier in this chapter, grassroots participants are indeed empowered in this paradigm of live streaming.

The second paradigm of live streaming is characterised by the activities of promising streamers and celebrity streamers, who are seen as the golden goose for the platforms and agents.

Platforms exert their power over the distribution of the attention from viewers and streamers in order to maximise the profitability of donations. Under these conditions, only the streamers and donors who follow the values and rules designed by the platforms can fully engage in the live interaction. Freelance streamers are moved to less prominent spots, and freeriding viewers and less generous donors are ignored. This ignorance constitutes the logic of the commodification of donation: the amount of donation is directly linked to the degree of satisfaction – no donation equals no gratification in this regard. Specifically, in Chinese live streaming platforms where the role of advertisers is not yet significant, the practice of donation constitutes a new mode of audience commodification. The real buyers, rather than advertisers, are the donors. As part of the audience, donors are driven by the platform to ‘purchase’ satisfaction with the virtual gift by

168

donation. This satisfaction, including the feeling of companionship, admiration and recognition, is not created by the platforms or agents, but rather comes from the affective labour of streamers and viewers

(Guarriello, 2019; Johnson & Woodcock, 2019a). In a donation contest, the leading donors gain the attention and appreciation of the streamers in front of the community members, and therefore the donors are put under the spotlight and enjoy the feeling of intimacy and appreciation from streamers, as well as a sense of superiority and admiration from other viewers. By contrast, those donors who don’t donate sufficient virtual gifts are excluded from this emotional reward. Although they devoted their labour to the production of a sense of fulfilment from the purchased virtual gifts for the streamers, they gain little or nothing in return. This tendency has been exacerbated by increasing commercialisation and the feeling of vanity promoted by the platforms, whereby agents and streamers use fake donations to gain higher profile spots, and viewers are provoked to make more donations. As a consequence, each single donation becomes less and less worthy. Clearly, live interactions are a scarce resource: the response from streamers is scarce and it does not increase with more donations; the attention from the viewer is also a scarce resource in a large community, as the flows of comments scrolls by quickly, and only the paid viewers are highlighted and can remain that way for longer periods

(Cao, 2019). Accordingly, a larger community and increasing amounts of donations do not necessarily mean increasing gratification. Rather, the more expensive the same amount of gratification is – that is, the more the donors donate – the less valuable the donation itself is, and this leads to donation inflation.

In addition, as unconscious performers, viewers are also creating value for the platforms: they create viewer-generated content as part of the show without being paid, while paying via donation for what they have observed.

Live streaming platforms seem to fill the digital gap in Chinese contexts and empower both grassroots streamers and viewers in the content-based paradigm of live streaming. However, I

169

don’t think we should overestimate their merit, or overlook the risks. The second paradigm implies a new mode of commodification, whereby the overpowered platforms control all the actors through distribution, and turn users into commodities, who then sell themselves to users. In this situation, the opportunity of ‘empowerment’ is at risk. The grassroots streamers who do not have a popular pitch lack visibility to the viewers and are hardly able to be noticed by large numbers of viewers.

Viewers who cannot afford high amounts of donations are excluded from participation, not to mention that the democratic means of ‘participating’ highlights the right of users to govern and control the structures that affect them (Fuchs, 2013, p. 57). While discussions exist about whether

‘content is the king’, platformisation of the live streaming in China is calling into question whether it is content or the platform that is the king. Essentially, these two paradigms cannot be separated from each other; instead, they need to be seen to coexist and to be intertwined in live streaming in

China. We can see how live streaming practices have increasingly transformed from a content- based paradigm to a platform-based paradigm. We still need to scrutinise the nature and the most recent development of platforms so we might better understand the politics of audience commodification in China. In the next section, we will look at the dynamics between these two paradigms and consider how stakeholders push the transformation of live streaming practice in the process of platformisation, which may lead to irreconcilable dilemmas.

Beyond precarity: The inherent dilemma for professional ‘amateur streamers’

The commodification of UGC and the gig economy have led to widespread concern about the precarity of labour on the platforms. While ‘working’ or ‘playing’ on digital platforms is characterised by flexibility, casualisation or freelance work (Friedman, 2014; Mosco, 2009;

Srnicek, 2017), relying on these jobs is precarious because of the lack of insurance, stability and

170

retirement savings. While the first paradigm of live streaming in China features independent

‘lifecasting’ of leisure activities, its transition to the second paradigm of highly commodified live streaming has shown that the leisure activities on live streaming platforms have become professionalised, and streamers have become increasingly dependent on an income from live streaming. This change, along with the process of continuing convergence of digital platforms, has created a very high probability of intensifying the career precarity of streamers (Johnson &

Woodcock, 2019b; Kim, 2012; Sjöblom et al., 2019).

Digital platforms feature UGC, but the increasing proportion and influence of professionally generated content (PGC) facilitated by the media conglomerates have the chance to

‘overshadow UGC, marginalizing individuals’ own creations’, and this process can be accelerated by the concentration of media ownership (Kim, 2012, pp. 61–62). In online screen entertainment in China, it has been noted that amateur production content has created cultural and social values

(Yang & Jiang, 2015), but the profit-driving online platforms have tended to professionalise these

UGC, converting them to PGC in order to maximise the economic profit instead, while minimising the regulatory risk (E. Zhao, 2016). Live streaming also shows a similar tendency as well as distinctive particularities. On Chinese live streaming platforms, more professional and higher qualified PGC is becoming a clear trend for the development of live streaming. By 2020, leading video platforms such as Huya, Douyu, TikTok and Kuaishou all gained Radio and Television

Program Production Licences, which means that they have the right to make their own video programs in addition to UGC. Instead of employing professional video-making teams and streamers, platforms use existing contractual streamers to create PGC for branding and more profit

(iiMedia, 2020). For example, Douyu started to launch variety shows including Werewolf and

Music Talent Show with its celebrity streamers from 2016, and Douyu Carnival, as a fan meet-

171

and-greet offline event, also gathered prominent streamers drawn from more than half a million offline viewers in 2018; it has been viewed online over 230 million times (Douyu, 2019, p. 155).

In addition, e-sports tournaments are also frequently organised by Douyu to promote competition among streamers and viewers, in order to further boost its popularity and consolidate its branding

(Douyu, 2019, p. 136). In such circumstances, these contractual streamers, who are usually amateur streamers, are led by the platforms to create professionalised content. In the meantime, freelance streamers who are not affiliated with platforms are becoming less competitive and marginalised.

The Chinese state also seems to favour the professionalisation of live streaming from 2020.

A few weeks after CCTV began using e-commerce live streaming and online celebrity for poverty reduction and economic recovery, the Ministry of Human Resources and Social Security for the first time acknowledged live-streaming sellers as a new profession (X. Shen, 2020a). This formal approval from the authorities further accelerates the growth of live streaming in China at a time when most industries have shrunk because of the COVID-19 pandemic. Live streaming, and especially e-commerce live streaming, has seen dramatic growth in 2020 (X. Shen, 2020b). The booming of e-commerce live streaming in China, can be seen as an extreme case of ‘sponsorship’ live streaming, which implicates the trend of commercialisation and professionalisation of live streaming industries in China.

Sponsorship and e-commerce live streaming as unique forms of advertising

In addition to putting advertisements in pop-up windows or forcing viewers to watch ad footage before watching the streaming content, live streaming enables a new form of ‘advertising’ on these platforms, which is becoming more prevalent nowadays. It is generally referred to as sponsorship

172

(Johnson & Woodcock, 2019a). In this context, streamers cooperate with a company to set up live streaming videos for the promotion. It may involve playing the game developed by the company, participating in an event organised by the company or introducing a new product for the company.

There is usually a link on the streaming channel for viewers to access the sponsor sites, generally accompanied by a discount for viewers if they purchase the goods through the link (Johnson &

Woodcock, 2019a).

This live streaming sponsorship may look like traditional TV advertising, but the detailed explanation from streamers and the live interaction make it less intimidating, and more like a product demonstration and information provision to inform viewers (Cai & Wohn, 2019).

Moreover, streamers who take the job of sponsorship are usually live show streamers or game streamers who already have an existing viewership, so these streamers are viewed as more trustworthy by viewers in the community (Gräve, 2017). The sponsorship streaming can also be subtle, as it only takes up a small portion of the overall streaming content. Therefore, I would argue that sponsorship is a component of a commercialised live show rather than an enhanced TV advertising with live interaction; the advertising components are integrated into the overall streaming content and generally manifested in a moderate way. It is common to see that platforms recruit streamers to promote certain games or event; according to their agents, platforms also assign the mission of recruitment to agents. Platforms ask agents to look for new streamers or request existing streamers in their companies to integrate the content proposed by sponsors into their original streaming content (Interviewee, A1, 14/01/2019; Interviewee, A2, 15/01/2019). For the company of Interviewee A2, commission fees to provide streamers for sponsorship constitute a major revenue stream for agents.

173

It is also interesting that some agents are using sponsorship to bypass the rule of the platforms. Streamers cannot contact advertisers or display any advertisements without the approval of the platform (Douyu, 2019, p. 146; Interviewee, R1, 28/06/2019; Interviewee, P1, 31/12/2018).

Yet the components of advertising are less obvious in the form of sponsorship, and they can hardly be identified by the platform regulator; as a result, the agents tend to look for ways to evade these rules. As interviewee A2 mentioned:

The streamers are not allowed to accept advertisements and post them on their

channel, but we have a solution. We cooperate with some companies and ask

streamers to go to the spots to participate and promote the event of the sponsors,

such as an auto show or tourism festival. We call it commerce streaming, and in

this way, we and the streamers can share the money from the sponsors without

being noticed platforms. (Interviewee, A2, 15/01/2019)

Interviewee A1 also confirmed that this sponsor live streaming is a typical way for agents to earn extra money without intervention from the platforms (Interviewee, A1, 14/01/2019). Agents approach the sponsors directly without the engagement of the platforms, thus avoiding having to share these sponsorship fees.

While the narrative of game streaming and live show only contains a portion of sponsorship content, e-commerce live streaming changes this – it is all about sales. For streamer–viewer interactions, viewers ask streamers to show different perspectives of the product, the process of making this product and the functionality of the product. For the viewer–viewer interaction, viewers ask other viewers about the experience of using the product, and the trustworthiness of the streamer. Live stream viewers also expressed that the live interactions between viewers could help

174

them get to know other consumers. They said they preferred to trust other consumers rather than streamers (Lu et al., 2018, pp. 6, 7). E-commerce live streaming, as an enhanced sponsorship version of live streaming, takes the commodification of live interactions in live steaming to a new level, as the whole streaming narrative and the interactions are focused on advertising and selling the product.

Interventions into production spaces

The professionalisation of leisure activities and homogenisation

As analysed in the previous sections, in China sponsorship and advertising management are handled by platforms or agent companies on behalf of streamers. We can see, therefore, that the leisure activities of streamers are subject to the interest of the agents and platforms, and are increasingly connected to commercial activities. On Western platforms, equipment including microphones and soundboards is prepared by streamers, and the streamers are self-trained to develop their skills (Johnson & Woodcock, 2019b). However, it is common to see that the streaming facilities and activities in China are prepared, and also highly normalised, by the agents.

Agent companies usually guide streamers to use deliberatively affective performances to meet the expectations of the viewers, and consequently this process will blur the personality and creativity of the streamers (Hu & Yu, 2019). According to the streamers interviewed, a training session is required by the agent companies before they can start live streaming:

The company provide us with a big training room. And all of us need to go through

a training session – we are taught how to sing, how to dance, how to interact with

viewers, and how to provoke viewers to donate more virtual gifts … Before

streaming, there would be a prepared script about today’s streaming content. The

175

theme of the script is usually connected to recent event or festivals, such as

Christmas. We can have right of improvisation under the outline of the script.

(Interviewee, A1, 14/01/2019)

As interviewee A1 noted, his company has rented big villas, and it makes each rooms of the villa a streaming room for each affiliated streamer, so the agents can handle the facilities and manage streamers easily: ‘It is easier for us to manage them if we let them stream at the “office” at a fixed time’.

Some streamers also claimed that they got real-time coaches from the agent company to monitor and help them when they were streaming. While this form of assistance and preparation from the agent companies can relieve the workload of streamers and professionalise the live streaming activities, it is also very likely that the agent companies, as a profit-seeking actor, would erode the originality or the creative activities of streamers as they tend to make the live streaming a professional live performance. As a consequence, live streamers who affiliated with the agents tended to use a more standardised and patterned way to interact with viewers (Hu & Yu, 2019).

The creative production of streamers has been influenced by agents and platforms, so it is common to find that streamers have no choice but to stream content they do not like. For example, interviewee S4 found that she could not switch to stream the game she wanted to stream because the agents had signed a contract with the platform to stream the game Overwatch:

The popularity of this game decreased and now no one wants to watch it, [and] I do

not want to play it, either. I planned to switch to H1Z1, but the agent company do

not allow me to do it. When the company signed a contract, I have to stream this

game regardless of the viewership and my willingness. I have tried to negotiate

176

with the agent company, but the manager said we have to follow the contract.

(Interviewee, S4, 27/01/2019)

After a few weeks, she found that the popularity of this game kept decreasing, as did her passion for this game, and the number of viewers on her channel. She then started playing another game,

H1Z1, and successfully attracted a larger audience. However, the platform decided to move her down from the front page as a punishment, saying that she had breached the contract (Interviewee,

S4, 27/01/2019).

Even celebrity streamers cannot avoid this control situation. In 2018, the popularity of the game PUBG grew rapidly and dominated the streaming platforms. Prioritising the needs of audiences, many streamers were forced to switch to live stream PUBG instead of their own content.

A celebrity streamer of the game DOTA, whose nickname is PIS, was one of the first to attempt to resist this coercion. PIS had played the game DOTA since 2008, and was one of the most famous video uploaders on Youku (Dexter, 2011). In 2018, he started to live stream DOTA on YY, and reached an agreement with the platform to mainly stream DOTA. However, after streaming

11 times, the managers of YY found PUBG could make more profit, so from 27 February they forced PIS to stream PUBG; otherwise the only choice PIS was given was to ‘leave and go find other platforms’. Rather than compromise, PIS finally chose to quit the platform (UUU9, 2018), but very few streamers have such courage and determination.

According to Douyu’s terms, the affiliated streamers have to follow the suggestions of the platforms to adjust their content:

Article 3: We have the right to make suggestions and advice on your live streaming

content. You should make corresponding changes within three working days after

177

receiving our suggestions and comments, otherwise we have the right to restrict and

close, recycle, or terminate your use of the Douyu live streaming channels. There

would be a corresponding deduction of the service fee (if there are any), which may

cause you a certain loss, and you should take the responsibility of the loss, We do

not take any responsibility. (Douyu, 2020c)

Douyu also claims it can arrange any commercial activities and sign any commercial contract on behalf of the streamers, and streamers should follow the arrangement and participate in the commercial activities (Douyu, 2020c).

It is clear that the profit-seeking platforms and agencies tend to control the production of streamers, standardising streaming activities and forcing streamers to produce commercial content that has higher profitability. This issue has been exacerbated in the case of sponsorship and e- commerce streaming, in which streamers have to give up their authentic personality and become salespersons. Under these conditions, the authenticity of content is damaged, and therefore the tension between presenting an ‘authentic lifestyle’ and selling out (Duffy, 2017) becomes a growing issue. As Woodcock and Johnson (2019b, p. 332) note:

Yet it is the presentation of authenticity and genuine opinions that streamers

promise to their audiences, and numerous streamers (and also creators on

YouTube) have found themselves having to negotiate challenging relationships

between the financial and cultural requirements of their work, which do not

necessarily align.

There is an inherent issue when the authentic performance of an amateur streamer is professionalised under the requirements of agents and platforms, whereby the streamer has to

178

produce highly commodified content regardless of the preferences of streamers and original viewers. The contradictions between content-based and platform-based, authentic and staged, professional and amateur, casual and formal are further intensified by interventions from various actors in the Chinese live streaming industries.

In short, in the development of live streaming, agents together with the platforms focus more on short-term profit maximisation than on pursuing possibilities for communitarian and reciprocal relationships among the communities – these profit-seeking actors tend to exploit existing social relationships instead of maintaining the social relationship and keeping the divergence of the content. This process does not empower streamers, but it does offer new possibilities to platforms for exploitation, leading to ongoing inequality (Zhang et al., 2019). These outcomes erode the creativity of streaming and, at the same time, result in performances that are homogeneous and less pluralistic.

Standardised and opaque evaluation

On Twitch, the evaluation of a streamer’s performance, in addition to the number of donations, involves a calculation of the number of viewers. The number of viewers is displayed clearly on the thumbnails and inside the channel. The viewership is directly related to the bonuses in addition to the donations, as streamers who have a consistent viewing number can gain the opportunity to be sponsored (Johnson & Woodcock, 2019a).

On Chinese live streaming platforms, there is another number shown at the same position on the thumbnail of the channel. However, this number is not the number of viewers, but rather a figure called Renqi: it relates to the number of viewers, comments, donations and other data (Xu

& Zhang, 2019). While this Renqi figure is calculated and monitored by streamers and agent companies, this new form of algorithmic evaluation is too complex and opaque to be handled by

179

streamers. As Xu and Zhang (2019) have noted, there are cases that have demonstrated that Renqi is not an objective figure, but rather a subjective one that is often manipulated by the platforms for promoting or punishing streamers (Xu & Zhang, 2019). Recently, this metric has been updated to an even more opaque one: the Redu. For example, Douyu has used a Redu as the sole criterion to evaluate streamers from November 2017 (Douyuservice, 2017). Douyu claims it has “built a comprehensive streamer evaluation system that utilizes our strong AI and machine learning capability to predict and identify streamers with potential through over 20 quantitative metrics, including characteristics of streams and interaction frequency’ (Douyu, 2019, p. 152). Interviewee

S8, who commenced her live streaming in 2019, told me that she never knew her real viewership

(Interviewee, S8, 24/07/2020) but only this Redu metric. The Redu is used for the purposes of streamer evaluation and is usually associated with signing bonuses and tracking payments from sponsorships; it is handled by the platforms using opaque and ever-changing algorithms. This opaque evaluation system is usually supported by the platform’s terms of service. For example,

Chapter 2, Article 2 of the Douyu Agents Streamers Service Agreement states that Douyu has the right to evaluate the streamers, and ‘the evaluation standard is decided by Douyu solely without the agreement of you’ (Douyu, 2020c).

Using this subjective, ever-changing, algorithmic evaluation, platforms can easily control and exploit streamers – there are many cases where streamers have not been paid because they do not fulfil the evaluation requirements (Xu & Zhang, 2019). Streamers have to work long hours and conform to the requirements of the platforms. These demands tend to distort their own creative desires, leading to exploitation and the homogenisation of their content.

180

Unfair contracts and power asymmetries

Although some agent companies are affiliated with the platforms, the relationship between streamers and agents is not straightforward: there is no employment relationship, but only a contractual relationship between agents and streamers, without insurance, welfare or the protection of labour law (Douyu, 2020c). Therefore, agents are cooperators with streamers rather than employers (Pan, 2018; Post, 2018). Some agent companies are informal organisations, and they are a category of dispatch agencies in the Chinese context. In the practice of labour dispatch, companies do not hire workers directly but rather do so through dispatch agencies; thus there is no labour relationship between companies and workers. This started from the 1990s and companies have been exploiting informal labour through this unfair relationship for decades (P.C. Huang,

2017; Zhao, 2019, p. 111). In the context of live streaming, the existence of agent companies is reducing the responsibility of the media platforms and transferring the risk to the creative workers, which constitutes a form of labour dispatch that characterises the informal labour practices on

Chinese social media. Through the cooperation of dispatch agencies, live streaming platforms can recruit streamers indirectly. In this way, platforms outsource labour and assets in order to reduce their own employment responsibility and accountability for the UGC, thus making themselves as lean as possible to gain advantage for market expansion, as Srnicek (2017) notes. While the precarity has been a common issue for the social media ‘workers’, in China the existence of agents and their complicated relationship with streamers have led to new forms of exploitation and labour issues (Pan, 2018).

From 2015 to 2017, there were 28 civil cases about the disputes between agents and streamers (Pan, 2018), and the overall number of civil cases before 2019 was 180. However, from

2019 to 2020, the number of cases increased to 630, indicating the increasing conflict between

181

agents and streamers. 24 Two central issues prevail in these cases: the first concerns unfair contractual obligation, while the second concerns the invalidation of informal agreement.

Unfair contractual obligations

As a consequence of the cooperative relationships between streamers and agents, there is usually no or less base salary for the streamers, and donation revenue constitutes the main share

(Interviewee, S8, 24/07/2020; Xu, 2019). However, the share of donations is deemed to be a form of bonus and is subject to certain contractual obligations. Streamers are required to fulfil certain requirements before being paid from the share of donations. Yet, rather than only requiring streamers to stream for minimum hours, agents also set targets for streamers to reach. In most cases, there are minimum requirements for subscriptions, viewership, donations and streaming hours (Pan, 2018). Streamers can only get paid when they finish these targets, and if they fail to do so they cannot get fully paid according to the contract, no matter how hard and how long they have streamed. In short, in this form of formal contract, instead of paying for the labour of streaming (Post, 2018), agents only pay streamers when broader goals are accomplished by the streamers. Moreover, as noted in the earlier section, the requirement could be something like Redu, which is evaluated and manipulated by the platform companies (Xu & Zhang, 2019). Therefore, it is reasonable to argue that the platforms have become the decisive actor when evaluating whether the streamers are eligible to be paid.

This asymmetric relationship could be worse, given the amateur status and low priority given to legal matters by the grassroots streamers. My interviews suggest that many streamers do

24 Data from China Judgement Online, https://wenshu.court.gov.cn/website/wenshu/181217BMTKHNT2W0/index.html?pageId=28eaf8a217977fe910 b63ce67b655319&s21=%E7%9B%B4%E6%92%AD%20%E4%B8%BB%E6%92%AD

182

not fully read their contracts, not to mention understanding them. Recent cases have shown that a proportion of streamers have failed to claim their payments, seemingly unaware of their entitlements (Xu & Zhang, 2019). This unequal position further distorts existing unfair labour arrangements and compensation for the social media components of the work. In order to pursue these rightful entitlements, streamers often need to follow the requests from the platforms and agents, working unnecessarily long hours. In this sense, although the affective labour created by streamers in the same period of time is fixed in light of the scarce attention, platforms and agents can exploit streamers by extending their working hours with unfair contracts in order to increase surplus value production.

Ever-changing agreement

Although some agent companies in China are registered organisations, many remain informal organisations and have not even signed contracts with streamers. According to Interviewee S8, the overall income of e-commerce streamers consists of a base salary and commissions from donations and sales:

There are usually no clearly written clauses about the number of commissions, and

only the base salary is clear. Sometimes the agent companies have an oral

agreement with us about the commissions, but they can also change it. Small agent

companies do not follow the rules – as I know – more than 80 per cent of streamers

are affiliated with small agent companies and they have not signed a formal

contract. For them, both the base salary and the commissions can change without

any guarantee. It is like an unregulated field and every company has their own rules

in their domain. (Interviewee, S8, 24/07/2020)

183

By changing the ratio between these two forms of revenue, agents find a justifiable way to maximise their own profit while harming the interests of streamers. Interviewee S4 also mentioned that she has changed agents many times to obtain higher revenue, but in most cases there is no formal contract with the agent, so there is no insurance for streamers, and whether agreements are binding or not is totally dependent on the morality and reputation of the companies (Interviewee,

S4, 27/01/2019).

This informality of agreements creates a large degree of uncertainty for streamers: the base salary is low and the commission fee is not stable, as it is decided or even manipulated by agents and platforms. Agents and platforms are always the winners if they keep the commissions and bargain hard with streamers. Therefore, platforms and agents tend to absorb and incorporate as many streamers into their network as possible without caring about the performance and future career development of these streamers. In addition, the option of opting out has also been blocked by agents when the streamers are already locked in: streamers have to pay huge amounts of compensation to agents if they switch platforms or agents – such clauses are clearly written into contracts (Guangxi Zhang, 2018). According to Douyu, affiliated streamers cannot participate in any other commercial activities or live streaming activities without approval, and fines apply

(Douyu, 2020c). According to interviewees A1 and A2, the penalties can range from half to millions of RMB, depending on the popularity of the streamer and the requirements of the agents.

Interviewee A1 said the companies have to add this clause to protect themselves:

Many streamers just disappear when they do not feel like keeping streaming,

without notifying us. And we have to look for new streamers, and train them again

so as to complete the targets set by the platforms. Some streamers quit the job and

notify us, this is okay and we would not penalise them. However, we then found

184

some of them were working at other agent companies, which is unacceptable for

us … we trained them and promoted them, but then they switched to other

companies for a higher salary when they become famous, which is unfair.

Therefore, we set a fine for breaching the contract, which is up to millions of RMB.

(Interviewee, A1, 14/01/2019)

While it is reasonable for agents to add such clauses to protect their rights, a lawyer points out that it is illegal to set such a high penalty for this kind of breach of contract (Zhang, 2018). However, it seems that the regulatory agencies and streamers themselves have been slow to realise this situation.

Essentially, the underlying logic of blocking the opt-out option is to control the copyright ownership of production. As analysed previously, live interaction is the key component of live streaming, so viewers are very unlikely to watch the archived videos because of the absence of live interaction. As the viewing metrics indicate, a top streamer with 14 million subscribers and millions of live viewers may only have been watched hundreds of times.25 Therefore, owning the copyright of the recorded videos is pointless for the platform, while the brand of streamers and their future productions remains the valuable asset – the copyright ownership of the future production matters. In order to possess these assets, platforms and agents seek to control the streamers; by limiting their commercial and streaming activities, they can assert their companies as the only owners of future production.

To sum up, the power asymmetry between streamers, agents and platforms has put streamers at a clear disadvantage. Unequal contracts and opaque evaluation systems coerce the

25 See https://v.douyu.com/author/aBADGb0yodXm

185

streamers into submitting themselves to platforms and agents, and to following those rules when producing content. The professionalisation and commercialisation of individual live streaming are pushing this informal UGC practice towards this form of structuration (Zhao, 2019, p. 80).

Arguably this professionalisation has been driven by the commercial logic of the platforms (Zhao,

2019, p. 77) and the specifically skewed algorithms. The end state may well lead to an inevitable consequence: the homogenisation of content production and the end of pluralism. This process can be seen to be well underway for many Chinese game streamers as they stream similar games within a mainly e-commerce milieu. According to one report, e-commerce live streaming is increasingly dominating streaming content, and the homogenisation of the content is a key issue, despite increasing profitability (iiMedia, 2020).

Platformisation and the concentration of ownership

The previous section analysed the issues arising from professionalisation and an unequal power relationship between different actors from the perspectives of datafication and commodification.

This section sets out to analyse the relationship between actors from a big picture standpoint, and to reveal its connection with the concentration of ownership in a uniquely Chinese context. Over the past decade, live streaming in China has undergone rapid change. The rise and fall of the live streaming platforms, along with the dramatic changes in ownership, call for closer analysis of media ownership – which has always been a key focus for the political economy of media studies

(Mosco, 2009). I will begin with an analysis of the changes in these markets and platform ownership, with a view to further understanding how platformisation could fundamentally transform culture production and distribution in Chinese live streaming industries.

186

From the war between the thousands to oligopoly

From 2012, YY shifted its main business from video chat to live streaming, and in 2014 – marked by AcFun changing its name to Douyu – hundreds of live streaming platforms began to spring up.

From 2015 to 2016, there were more than 700 new live streaming platforms (Laoyu, 2019). The phrase ‘the war between the thousands’ has been used commonly in Chinese live streaming industries to describe the fierce competition between platforms in this period (Laoyu, 2019;

TechSina, 2017). These platforms absorbed huge capital investments and competed fiercely for viewers and influential streamers. However, most of them were quickly wiped out, with the war between the thousands lasting only for approximately one year. Two factors led to the drastic attrition of these platforms. First, the unregulated content quickly attracted the attention of the government in 2016, so new restrictions were enforced by the authorities, and many platforms were closed down by the regulatory agencies from 2016, as Chapter 3 examined. The high entry barriers – specifically the licence requirements – set by the government also automatically filtered many small platforms out. Second, platforms were required to use high sign-up fees to lure top streamers and to bring them on board. The sign-up fee for leading streamers could be up to tens of millions of US dollars (PUBGAssistance, 2017), which further weakened the ability of smaller platform to compete. These small platforms, instead of obtaining more streamers, could only watch their top streamers being grabbed by bigger platforms. Because of this fierce competition and increasingly high entry barriers, by the end of 2017 only about 200 platforms remained (Laoyu,

2019).

In this process, while normal streamers suffered from precarity, the celebrity streamers were the winners. It was common to see celebrity streamers being paid millions of US dollars per year in signup fees, and they would frequently switch to other platforms for even higher fees

187

(PUBGAssistance, 2017). Losing a celebrity could be fatal to a small platform, as the top celebrity streamers brought in millions of viewers their departure could result in three to four million uninstalling live streaming apps (Xinhua, 2018b). Obviously these celebrity streamers are overpaid, but the underlying logic behind this was a ‘growth before profits model’ that prevails with platformisation, and that aims to defeat the competitors and then establish a monopoly position

(Srnicek, 2017).

The sign-up fee constituted the major cost for leading platforms such as Douyu (Douyu,

2019, pp. 1, 14, 97), and it could only be covered by continuous venture capital investment on the expectation that high profits would follow once fewer players came to dominate the sector. Douyu attracted huge investment, but its profit margin was quite small from 2018 (Douyu, 2019, p. 14).

It is possible that the high sign-up fees, as a characteristic of leading platforms in China, were only feasible for the platforms that were underwritten by venture capital investment. This ‘requirement’ led to unfair competition, and contributed to the sector’s current oligopoly structure.

The good times for these leading streamers enjoying a high sign-up fee did not last long, as the market structure quickly changed. Since 2018, hundreds of live streaming platforms have closed down as a result of a combination of strong regulatory action from the government, failing operations and the collapse of investment arrangements. Even the leading platforms, such as Panda

TV, which was founded by the son of a Chinese tycoon, faced capital chain rupture and closed down (Yeung, 2020, p. 104), leaving both their streamers and employees unpaid (Fu, 2020). On the other hand, concentration of ownership began in the sector from 2017 with Tencent investing in Kuaishou, Huya and Douyu (Laoyu, 2019). There have been a number of mergers and acquisitions since then. Hong (2017a) notes that the joint efforts by the state, internet companies and global capital all gave rise to the tech giants in China. Baidu, Alibaba and Tencent (BAT) are

188

aggregating economic and political power to achieve platform convergence. Moreover, like

Western counterparts such as Google and Amazon, BAT also adopt lock-in strategies and an enclosed business loop to seize and exploit users (Su & Flew, 2020). Not surprisingly, live streaming in China has been increasingly dominated by the BAT tech giants. Alibaba directly benefits from e-commerce live streaming as it owns leading online shopping site Taobao; Taobao

Live now is the biggest e-commerce live streaming platform. Baidu is a step behind, but has begun serious investment into live show streaming by acquiring YY – which is the biggest live show streaming platform (D. Li, 2020). According to Interviewee P1, who was an employee of YY, this acquisition happened very quickly and he suddenly became a Baidu employee (Interviewee, P1,

30/11/2020). In 2020, Tencent became the largest shareholder of both Douyu and Huya, and in

October Tencent announced that Douyu and Huya had entered into a merger agreement (Douyu,

2020e). According to a report, Huya and Douyu had 40.25 per cent and 35 per cent market share of game streaming respectively in 2019 (Finance Sina, 2020). Accordingly, game streaming in

China is now dominated by Tencent. In addition, Tencent’s own streaming platform, Egame, also began to sign leading streamers, and continues to gain in popularity.26

It is notable that Tencent’s influence is not limited to gaming and live streaming industries.

The dominant place of the Tencent cyber ecosystem in China has been consolidated through its global mergers, acquisitions, investments and growing domestic political influence in recent years

(Birkinbine et al., 2016, pp. 347–48). When leading platforms aggregate users, advertisers, data and other third parties, they are able to benefit from the network effects. The result is further consolidation of their dominant position, which then automatically makes it increasingly difficult

26 See report from http://www.iresearch.tv/research/platform/data

189

for new enterprises to break into the market (Poell, 2020; Srnicek, 2017, p. 26), and leads to the further enclosure of the market.

Oligopoly, platformisation and labour

The tensions between oligopoly and labour, especially the extraction of surplus value and datafication, have always been a central issue of platformisation. As Srnicek (2017, p. 42) argues, the platform model is inherently precarious: platforms simply turn those traditional non-tradable services into tradable services and incorporate them into their own ecosystems. Most notably, the platforms make independent contractors and freelancers, rather than employees, into their workers

(Srnicek, 2017, p. 40). These workers may enjoy some job flexibility and acceptable pay in the beginning, during the expansion of the platform, but these conditions then decline following the concentration of ownership. This can be explained in part because the platform model relies on future profitability and adopting a ‘growth before profits’ model, which is funded by venture capital investment to realise a future monopoly (Srnicek, 2017, p. 12). For platforms, the wellbeing of the creative workers and viewers during the expansion process is underwritten by venture capital investment, and this is neither profitable nor sustainable. Therefore, after gaining a dominant position, it is to be anticipated that platforms may dramatically change their strategies towards these creative workers.

Global venture capital has become an emerging form of investment in China’s internet industries (Tang, 2020). Existing studies have observed how venture capital has invested in platforms in China and accelerated its surplus value extraction, creating significant labour issues after gaining a monopolistic position in the market, despite the platforms’ promises of universal access and democratic participation (Guan & Zhang, 2020; L. Zhang, 2020; Zhao, 2019). For example, the venture investment companies took more than half of the shares of Douyu (Douyu,

190

2020a, p. 104), so it is arguable that after reaching a monopoly status, live streaming platforms that depended on venture capital will change strategies to make up the losses in the expansion process. Two typical strategies have been used widely by platforms: cross-subsidisation and data extraction.

Cross-subsidisation

The cross-subsidisation strategy describes a process whereby one arm of the firm that reduces the price of the services (or even makes them free to enrol more and more users), then another arm raises prices for existing users (Srnicek, 2017, p. 26) or reduces costs by lowering the pay given to the labourer (Srnicek, 2017, p. 45). In short, a platform may reduce their costs by increasingly exploiting the ‘workers’, which can result in a significantly degraded working situation for those workers – the working conditions of Uber drivers is a typical case in point. Drawing from this, the impact can be that all the interest and reward that platforms grant to their users in the early stages of investment turns out to be ephemeral.

In the context of Chinese live streaming, streamers are becoming increasingly dependent on the agents and platforms. Rather than continuing to be a freelancer, streamers have increasingly been incorporated into the ecosystem – affiliated to the agents and bonded to platforms in order to

‘survive’ in the industries, even at the cost of losing their independence and freedom. There are three reasons behind this. First, many functions and monetisation channels on the platforms are only available to contract streamers. In a similar fashion to the situation at Twitch, in China only the partners or affiliates can earn revenue from subscribers, advertisements and sponsors.

Moreover, these partners or affiliates enjoy other privileges, including but not limited to more video storage and the right to launch polls and priority customer services (Help.Twitch, 2020b;

Zhang et al., 2019). In terms of self-regulation, only the affiliate streamers can use some automatic

191

moderation functions, while agent companies are doing real-time content moderation of streamers to reduce their risk of being punished, particularly when the streaming content is playing edge ball

(Interviewee, A2, 15/01/2019).

Second, and probably most importantly, streamers gain advantages when cooperating with agents, as agents provide the script, advice and training for the streamers. Agents also cooperate with platforms to boost streamers with promotion to a slot on the front page, as examined in earlier sections. All these make the affiliated streamers more competitive compared with freelancers.

Interviewee S2 said she has streamed for a year and made little money. After cooperating with agents and listening to their advice, S2 claims she now earns 5000 RMB per month. She also noted that she no longer needed too much time to prepare for the show as the platforms provided her with a script (Interviewee, S2, 15/01/2019).

Third, agents are always looking for streamers to try to incorporate them into the ecosystem.

They release live stream jobs on recruiting websites for potential streamers, and scout existing streamers on live streaming platforms. “We have a professional team to look for and talk to

[potential] streamers, asking them to join the agent companies. If they have already got one agent, we may offer better salary or services to get them,” said A1 (Interviewee, A1, 14/01/2019).

Interpersonal relationships are also utilised by the agents, who recruit streamers through the personal network of their affiliated streamers: ‘We let streamers advertise the job to their friends and classmates; the successful introducer can earn hundreds to thousands of RMB.’

Therefore, with these forms of short-term benefits, streamers are lured into enrolling in a system that is controlled by platforms and affiliated agents. When the ownership of the live streaming platforms has become concentrated, and agents have increasingly become suppliers to their targeted platforms, streamers have had no choice but to submit themselves to the rules of

192

BAT-owned platforms. The streamers can only choose ‘either submitting themselves as data commodities to the platform, or being denied the possibility of benefiting from online interactions while staying away from the platform’ (Zou, 2018). One the one hand, dominant live streaming platforms can exploit smaller streamers by forcing them to accept an unfair proportion of donations and subscription fee; on the other hand, influential streamers are driven by platforms to participate in rounds after rounds of carnivals promoting products and events, and competitions for garnering donations (The Paper, 2020). While participating in these activities, which benefit agents and platforms, streamers are forced to work overwhelming hours and are thus rendered exhausted (The

Paper, 2020). From the perspective of viewers, they have to pay higher and higher prices for donations because of the inflation of donations facilitated by the platform. Platforms cut down the free services of providing an intermediary for live interactions and make it more and more expensive, which is all part of the cross-subsidisation strategy.

Data extraction

The second strategy that platforms use is data extraction, which is associated with datafication and algorithmic culture afforded by network platforms (Srnicek, 2017; Van Dijck & Poell, 2013). For example, in 2000 Google unveiled AdWords using extracted data in October 2000 and began its transformation into a revenue-generating company (Srnicek, 2017, p. 29). Twitch, after being acquired by Amazon in 2014, not only uses user data to facilitate target advertising, but shares these data with other Amazon subsidiaries to gain a market advantage (Panneton, 2019). In the case of Douyu, in addition to advertising, user data has been extracted and used for content recommendations by sponsorship streaming and e-commerce live streaming (Douyu, 2019, p. 3).

The overall commercialisation and increasing need for profit has increased the coverage of advertisements on Chinese live streaming platforms. Compared with a cleaner interface in the

193

earlier stages, nowadays the streaming channels have been filled with pop-up advertisements, clickbait and thumbnail links for donations. A notorious case happened during the pandemic in

2020, when teachers and students needed to use live streaming platforms for online education.

While Douyu and Huya have commonly been used, the viewers’ information has been collected and it appears that the channels were filled with advertisements. As the viewers are the students, a lot of game advertisements appeared during the online teaching sessions targeted at students, luring them to click on the games and to play them (Xinhua, 2020). Online teaching, as parents criticised, has become online gaming (Xinhua, 2020). Moreover, many kids were using devices belonging to their parents for online learning, and sometimes their Ali Pay or WeChat Pay accounts were linked to the platforms when users were using these devices. Understandably, it seems many kids were easily lured to pay for a game or donate to the platforms using the accounts of their parents.

According to a CCTV news report, one student donated more than 30,000 RMB with her mother’s phone to Huya live streaming while they were meant to be using the phone for online learning on

Huya at the time (CCTV, 2020). In this sense, by utilising user data, platforms are taking commercial advantage of online teaching, not only for promoting games to children but also for harvesting the purchasing power of their family in an unethical way.

Although the authorities have called for the online advertising to stop being shown on live streaming teaching channels, a regulatory issue remains: existing laws and regulations have forbidden the harmful apps, including games and commercial advertisements, to enter primary and middle school in China (Ministry of Education, 2018). Speculatively, however, the question might be asked: What if the students are the ones that have to enter the domain of platforms with the existing commercial logic of promoting these advertisements and games? At its heart, this question is about the conflict between social responsibilities and profitability. Live streaming platforms are

194

owned by commercial organisations for entertaining users and earning a profit, but they are now being used for other purposes, such as online education or poverty reduction, which require a sense of social responsibility. So who can regulate them, how can they be regulated and what kinds of regulation policies can justify the regulatory measures? If the advertisements are forced to be removed to protect the children, is it reasonable or fair to prevent the platforms from accessing this source of revenue when they provide numerous channels and bandwidth freely for educational purposes or other social-cultural benefits?

Conclusion

This chapter set out to understand the multi-stakeholder dynamic of Chinese live streaming within the context of empowerment, commodification and professionalisation. While digital media tend to extend the digital gap, the practices of live streaming have the potential to minimise the disadvantages of the grassroots through new ways of creative content production and providing other financial benefits and satisfaction. Together with the poverty-reduction policies launched by the state, live streaming can empower the grassroots and reduce the digital gap.

The creative, pluralistic and democratic practices characterise the content-based paradigm of live streaming practices. Yet the platforms did not always act responsibly as passive intermediaries. In a few short years of evolution, platforms have come to exert power over the distribution of the attention in order to maximise their profitability through donations and sponsorship mechanisms. As a result, these changes have produced a highly commodified digital space of live interactions and social relationships between users; this amounts to the beginning of another paradigm of live streaming practices – the platform-based paradigm. In this paradigm of live streaming, platforms utilise TOS, architectural design and user data to make both the streamers

195

and viewers the real (and at times) unconscious performers, creating affective labour for the bottom line of these streaming platforms. Inflation in the donations facilitated by the platforms is extending the divide between goldmine owners and free-riders, exploiting average viewers in their live interactions.

In recent years, the commodification of live interactions, the concentration of ownership, the rise of streaming agents and the accumulation of platform power have increasingly turned the content-based paradigm into a platform-based paradigm. On the one hand, in order to harness the streamers and make the live streaming content ‘manageable’ and profitable, platforms and agents have struggled to make live streaming a legitimate platform-based career. This has occurred as a result of accelerating the standardisation and professionalisation of individual live streaming practice, previously supposed to hinge on individuality and authenticity. On the other hand, platforms and agents take advantage of the informality of this work by building cooperative relationships instead of employment relationships, and by exploiting the grassroots streamers who are unfamiliar with these unfair agreements. In short, platform companies and agents work together to take benefits from both sides: they enjoy the stability and profitability of the formality of individual live streaming, while avoiding accountability through informal means.

At this stage, although the two paradigms of live streaming coexist and are intertwined, the platforms are increasingly in control of the production and distribution of the live streaming content and we are witnessing the dramatic expansion of the platform-based live streaming model.

The platformisation of live streaming is self-contradictory: the conflicts between content-based and platform-based streaming, authentic and staged performance, professional and amateur careership, informal and formal agreement, precarity and stability remain resolved, and are further

196

intensified by the increasing interference from various actors in the Chinese live streaming industries.

Studying the politics and regulation of live streaming should not be limited to narrow scopes such as privacy issues and the moderation of unlawful content; new research also needs to consider other mechanisms that may influence the live streaming activities and the changing dynamics between multi-stakeholders, to bring out issues such as inequality, precarity, illegal contracting and vicious competition. Live streaming practices and platform companies should be held accountable for fostering a pluralistic, democratic and more competitively well-balanced digital media ecosystem. To achieve this goal, regulatory research should highlight the complex processes and interactions when the state is ruling multi-stakeholders in the multi-jurisdictional, decentralised networks (Bevir, 2011, cited in Flew, 2021), which frames the discussion in the next chapter. Studying the politics and regulation of live streaming should not be limited to narrow scopes such as privacy issues and the moderation of unlawful content; new research also needs to consider other mechanisms that may influence the live streaming activities and the changing dynamics between multi-stakeholders, to bring out issues such as inequality, precarity, illegal contracting and vicious competition.

197

Chapter 5

Conclusion

Supervised cultural production: Paternalism over ‘lowbrow’ endeavours

‘Yangchun-baixue’ and ‘Xiali-baren’ are two ancient Chinese melodies. The former, ‘Yangchun- baixue’, is a masterpiece composed by a famous virtuoso, which circulated among the upper class in the Spring and Autumn Period and is considered a superior, elegant melody. The latter, ‘Xiali- baren’, is a series of popular folk songs that were enjoyed by underclass in the same period (Zang,

2007). Nowadays, ‘Yangchun-baixue’ is often used to refer to elitist cultural production and consumption that is considered to be superior, proper, polished and elegant. ‘Xiali-baren’ refers to cultural production and consumption of the rustic poor country people, and it is related to popular, common ‘or even vulgar and boorish’ content (Wu, 2004, p. 255). These two different forms of cultural production are generally described as highbrow and lowbrow in Western contexts (G.C. Li,

2020; Lu, 2015).

Anti-vulgarity campaigns and stereotypes

Live streaming platforms, as analysed in previous chapters, appeal to grassroots streamers and viewers, while live streaming practices themselves can be characterised as lowbrow cultural production and consumption. At the beginning of live streaming around 2012, Chinese live streaming featured game streaming and live show performances (Lu et al., 2018; Zhou, 2017); these were typically lowbrow, yet also legitimate and far from vulgarity. Not long after, live streaming content became difficult to regulate for the government within any traditional regulatory

198

framework, thereby providing an opportunity for voiceless grassroots groups to assert self- expression with few limitations (Ruberg, 2020). Moreover, cultural production could be shaped largely by the infrastructures, business models and governance of the platform, as several studies have noted (Bucher, 2012; Gillespie, 2014; Van Dijck & Poell, 2013). Lowbrow live streaming content was making its mark. In this situation, it was inevitable that platforms and their affiliated agents were inadvertently promoting vulgar and sensational content, and often also deliberately condoned violations with inaction, in order to draw more viewers and donations (Jingyan, 2016;

People.cn, 2020). As a consequence, live streaming content from 2012 to 2017 contained a significant proportion of ‘inappropriate content’. In general, it has been tagged with the labels

‘low’, vulgarity, violence and pornography, while viewers have been criticised in public discourses for their low or ‘vulgar’ tastes (Chen & Chen, 2020; Zeng & He, 2017; G. Zhang, 2019; Zhang &

Hjorth, 2017). Although such ‘inappropriate content’ is not explicitly illegal, the government tended to side with the elite groups against this content with lowbrow tastes (G.C. Li, 2020, p. 23) and has launched operations to ban vulgar and obscene content on platforms (CAC, 2018b; Lai,

2017; SAPPRFT, 2018b).

As examined in Chapter 3, the regulatory policies of the state are not only for economic or political purposes, but also for cultural governance, with the goal of promoting moral goodness and socialist values (Cui & Wu, 2016; Yang, 2014). Arguably, this diverges from Western digital media regulation, which focuses on child safety, terrorism and privacy violations (Flew & Suzor,

2018, pp. 168–69). Anti-vulgarity campaigns on social media have shown the state’s will to promote moral goodness and restrict ‘low’ cultural content. In 2000, culture minister Cai Wu criticised vulgar and kitsch media production, urging a fight against gossip and sensational and consumeristic media content (Chinadaily, 2010) by launching the so-called Anti-vulgarity

199

Campaign. Intended mainly for the protection of the mental health of the youth and promotion of positive social values, this campaign indicates the tradition of strong paternalism in Chinese regulation, whereby the state acts as the protector of the people, controlling their behaviour in order to promote their own good (Cui & Wu, 2016). The Anti-vulgarity Campaign and its paternalistic interference also widely penetrated live streaming platform governance. The Anti- vulgarity Campaign on live streaming platforms began in 2017 (Sina, 2017) when the CAC (2016,) stated that all the live streaming content should promote positive and healthy values. For example, some poker games and most console games were forbidden from being shown on live streaming platforms (Ministry of MCT, 2016a). As outlined by the policies of the CAC, the point of this

Anti-vulgarity Campaign was similar to the restrictions applied to TV game programming in 2004: mainly to protect the mental health of minors, as games continued to be deemed addictive and potentially detrimental (Wang & Chen, 2019). Game-related TV content was thus restricted from broadcast from 2004, and this restriction directly led to a stagnation of TV shows and live stream content related to games in the decade that followed (Shu, 2004). This form of cultural governance and crackdown on ‘vulgar’ content shows the tradition of regulating ‘legal but unethical’ content in a paternalistic way.

Promoting goodness with positive energy

The suppression of lowbrow content on live streaming subsequently changed because of the poverty-reduction agenda, as Chapter 4 noted. The Chinese state proposed ‘entrepreneurial solutionism’, including the development of digital technology and entrepreneurship, as a solution to China’s social, economic and cultural problems (Keane & Chen, 2019), with many other agendas originating from this proclivity. This idea of ‘entrepreneurial solutionism’ was applied to social media in 2019. Accompanying the ‘Internet +’ agenda, ‘Mass Entrepreneurship and

200

Innovation’ (大众创业万众创新) was a newly proposed policy agenda that aimed to encourage grassroots individuals to become digital entrepreneurs and realise their value. The underlying logic was to encourage grassroots users’ appropriations of digital platforms to launch their own businesses as a means to achieve the goal of reducing poverty and polarisation (Keane & Chen,

2019, pp. 746–47). The advancement of ICTs infrastructure in rural areas was facilitated by the

‘Key Points of Online Poverty Alleviation’, and also built up a solid foundation for the digital entrepreneurship of individuals in peripherally related areas (CAC, 2020). Nonetheless, these agendas have greatly contributed to the platformisation of grassroots cultural production, therefore leading to a growing platformed creative economy in China and also resulting in ‘grassroots individuals’ from diverse backgrounds becoming creative workers of the platforms, despite the content being deemed ‘low’ or vulgar (Lin & de Kloet, 2019).

Although vulgar and illegal content still exists on Chinese live streaming platforms, the overall environment has been improved after several rounds of special operations and the implementation of specific rules on live streaming (Huai, 2020; People.cn, 2020). Promoted by the state policy, live streaming platforms have rolled out a genre of streaming to promote moral goodness through cultural artefacts, which is generally referred to as ‘positive energy’ live streaming (Douyu, 2020d). As one genre of live streaming that belongs within the live show format, such ‘positive energy’ streaming includes Chinese traditional literature and art, traditional craft making and charitable activities. Online celebrities including influential streamers, who are believed to have ‘used their influence to do charity work’ and ‘spread traditional Chinese culture’, have been supported and promoted by the official policy (W. Zhang, 2020). In addition, the category includes folk artists who use live streaming to share knowledge and promote intangible cultural heritage. Benefiting from the interactivity, unpredictability and authenticity of live

201

streaming, folk artists in remote areas can show their country lifestyle and handicrafts to the world

(Lu, 2019).

These positive energy streamers promote Chinese culture in a way that fits the state’s agenda, and they also often become successful digital entrepreneurs. Although it should be noted that the popularity of positive energy is still not comparable to the other live streaming content,27 the stereotype that surrounds live streaming has been reduced. The state has shown increased tolerance and openness towards ‘lowbrow’ live streaming content that is not explicitly illegal and politically sensitive, and users and platforms do not always evade government regulation when there is a chance to be incorporated: those platform owners wishing to be recognised are able to be incorporated into the regulatory scope for the purpose of gaining legitimacy and sustainability

(Zhao, 2019, p. 138).

A paternalistic, principles-based model of regulation

The state’s regulation of live streaming over the past five years, then, has been less about the regulation of platforms, companies and capitalism than about the regulation of illegal content and

‘lowbrow’ cultural production by grassroots groups. Despite the explicitly illegal and politically sensitive content, the Chinese state has adopted an overtly paternalistic model of regulating all other forms of content that have not been clearly defined as illegitimate. The economic substructure of China is very distinct in comparison with Western countries – despite the rapid development of first-tier cities and a digital platform economy, the GDP per capita of China is still slightly lower than the world average, and less than 16 per cent of that of the United States

27 The Redu of positive energy is less than 1 per cent of the game LOL, and about 2.4 per cent of the Outdoor Live Streaming. Data retrieved 22 January 2021 from https://www.douyu.com/directory

202

(Worldbank, 2019). There are still huge socioeconomic disparities between rural and urban regions

(Knight et al., 2006), and central and the peripheral areas (Shen & Xiao, 2020), as well as discursive power between different social classes (Sun et al., 2020). It is in this context that

Chinese live streaming presents a different landscape compared with mainstream media and other social media. It is mostly ‘lowbrow’ cultural production created by the grassroots streamers to boost income and entertainment, and viewed by grassroots viewers for psychological satisfaction.

In this sense, the platformisation of live streaming in the Chinese context is largely a narrative concerning how the platform companies, along with other stakeholders, are transforming grassroots culture production, distribution and circulation through a platform logic that is building a new ecosystem.

In terms of governance by platforms, platform companies and their affiliated commercial entities, including agent companies, tend to commercialise live streaming and exploit the labour of users in controversial ways, as noted in Chapter 4. These commercial entities govern the streamers and viewers through TOS, architectural design and data analytics, exerting power over distribution in order to maximise their profitability. In terms of content regulation by the state, the sector remains underpinned by paternalistic ideologies (Wang & Lobato, 2019): from one perspective, the state aims for political and economic objectives, and thus it becomes a stakeholder that seeks to boost the platform economy. From another perspective, the Chinese state is still very much publicly accountable and plays a paternalistic role in platform governance. Although China is building a rule-based legal system (Tomasic & Fu, 2005, p. 26), and platform companies also prefer a rule-based system that provides certainty and stability, some commentators believe it is still too rigid and therefore not particularly suitable for fast-changing industries (House of Lords

Select Committee on Communications, 2019, p. 14). Moreover, the application of rule-based

203

systems in China faces challenges from huge regional variations, severe wealth inequality and fast- changing industries, and resistance from local government (Tomasic & Fu, 2005, p. 26). Yet the paternalistic ideologies and constantly changing attitudes of the Chinese state towards live streaming indicate the capacity for compromise, since it represents a more principle-based focus on future outcomes instead of singularity in media formats. This principles-based approach is consistent with the Chinese platform ecosystem, as an overly detailed regulatory approach is more suitable for static markets while a more functional approach and principles-based regulation are more applicable to dynamic platform markets (Gorwa, 2019b, p. 865).

Calling the community: An absence of protagonists for self-management

Previous chapters have examined the regulatory power of the state, local government, platform companies and related commercial entities in shaping the communication practices of live streaming platforms. It seems that platform users are rendered passive under the regulatory frameworks set out by these stakeholders, even if some of them become ‘room regulators’. This section points out that the ordinary users in Chinese live streaming platforms are not only passive producers, consumers or distributors, but also have the potential to constitute a promising form of emerging regulatory power for the platform governance.

Gorwa (2019b) concludes that there are three major models for platform governance. The first is self-regulation by the firms, including platform companies and other related commercial entities. The second is external regulation, namely the classic government or state intervention, including for privacy and data protection regulation, intermediary liability and the application of competition and monopoly laws (Gorwa, 2019b, p. 863). The third model is co-regulation, which features the involvement of civil society organisations such as NGOs and community self-

204

management from users. As examined in previous chapters, multi-stakeholder negotiation and conflict characterise live streaming regulation in China, and government intervention, together with platform self-regulation, has evolved over the past decade. However, the potential for co- regulation of civil society organisations has yet to be discovered. Given that most civil society organisations are relatively less influential in the Chinese context (Jiang, 2014), peer regulation from the user community has emerged as an essential and potential way ahead for co-regulation.

For both Western and Chinese live streaming platforms, regulation from the community has become another way to help other regulators to identify inappropriate content (CAC, 2018b;

Douyu, 2019, p. 154; Ruberg, 2020), and bringing community support into governance practice is a way to help both external regulation and self-regulation to gain legitimacy and trust from ordinary users (Gillespie, 2018, p. 87). In this sense, community regulation has been identified as a key alternative form of informal regulation to guide difficult content issues and user behaviour

(Gorwa, 2019a).

Community regulation in Chinese live streaming

Abbott and Snidal (2009, pp. 3–4) identify four essential competencies for the actors in platform regulation: independence, representativeness, expertise and operational capacity. Independence and representativeness guarantee the openness and democracy of regulatory decision-making, while expertise and operational capacity are the key competencies for the effective implementation throughout the regulatory process. Arguably, community regulation on live streaming platforms potentially fulfils the representativeness criterion in terms of self-government, but its independence, expertise and operational capacity are still far from competent for platform regulation. The most well-known form of community regulation is flagging, which has been widely adopted by social media platforms, as well as some live streaming platforms in China (CAC,

205

2018b). Clearly, the flagging mechanism grants users a democratic right to participate in the governance and to seek help when being offended, but it is still a relatively ‘narrow vocabulary of complaint’ that can hardly contextualize the complaint, with the flagging remaining constrained by the structure of the reporting system that was devised by the platform (Gillespie, 2018, p. 91).

On Chinese live streaming platforms, flagging is connected to the database of the government agencies and the actual enforcement actions are dependent on the state’s final implementation (CAC, 2018b). In addition to the platforms’ willingness to take action and their overall algorithmic design shaping the forms of community governance, reports from the community need to fit the categories set out by the platforms and to follow the structural logic of the platform. If the platform regulation rules and categories are vague and represent double standards, then peer-to-peer community regulation has little or no choice but to follow the platform logic for pursuing the platform’s commercial goals. In this regard, Twitch adopts a gendered double standard in definitions of sexual content, and this weakens the operational capacity and independence of the community regulation (Ruberg, 2020, pp. 13–14). Clearly, if community regulation lacks independence and operational capacity, it will inevitably lose its competency and become dispensable in the service of the commercial entities and governmental agencies.

The expertise of users themselves in terms of regulation is another issue. As Gillespie notes,

‘Flagging systems can also be gamed, weaponized to accomplish social and political ends’, and users are very likely to flag content that offends them personally or with which they disagree, even if the content is not a violation according to the rules (Gillespie, 2018, p. 92). The lack of expertise of community viewers and regulators renders peer regulation subjective, and this may influence the objectivity and competency of the community regulation. In the case of live streaming, not only flagging but also room regulation is compromised. For example, in the case of White55 on

206

Douyu, both the community members and room regulators reported the streamers’ comments with which they disagreed. The issue can be further intensified by community moderators who are granted privileges, but they may have a strong connection with and affection for the streamers, so their judgement may become clouded.

In summary, two issues need to be resolved to improve co-regulation from the community.

The first relates to how to increase the independence and operational capacity of the community regulation, connecting it with other actors in order to achieve a better level of co-governance, instead of being a referee or reporter without independent regulatory power. The second concerns how to improve the expertise of the community regulator. In particular, how can the correctness and objectivity of judgement made by affective users be improved?

Co-regulation models

In order to figure out a model that can deal with these two issues, we need to recap the existing models of community regulation. Live streaming platforms in China have made some initial moves towards integrating community regulation into their existing regulation systems. As discussed in

Chapter 3, peer report and community regulation have been widely used for the content moderation of live streaming. Huya also rolled out a positive energy rating system to allow viewers to rate the morality and goodness of the streamers after it had been criticised by the government for its vulgar content (Han, 2020). In the next section, I will highlight some innovative cases from other social media platforms, which provide useful evidence to assist in guiding us towards considering the features needed to develop a new community regulation framework for live streaming.

207

Trusted flaggers and the Facebook Oversight Board (FOB)

Given the abundant cases of wrongful content moderation conducted by the platform regulators

(Article 19, 2020), improvement to the correctness of moderation conducted by users is now at stake. Many platforms have tried to evaluate users and hire trained or trusted users for community regulation. For example, the YouTube Trusted Flagger Program was developed by YouTube to help the company to identify violating content. Any individual users, government agencies and

NGOs can become trusted flaggers if they have expertise in one field of the regulatory realm, such as ‘child safety’ or ‘suicide and self-injury’, and have been flagging high-profile content frequently and accurately (Facebook, 2020a). After the training session, successful flaggers can then be promoted to trusted flaggers. The flagged content by trusted flaggers is reviewed by platform moderators as a priority compared with other flagged content, thereby increasing the efficiency and reducing the workload of platform’s self-regulation. However, the flagged video is not automatically moderated; instead, it is reviewed by YouTube’s official content moderators. The reporting system is based the YouTube’s Community Guidelines instead of the law, making this co-regulation model part of the self-regulation of the platform (YouTube, 2020). Even so, the

Trusted Flagger Program appears to be a good way to improve the democratic participation of ordinary users; and it has significantly improved the effectiveness of platform regulation according to a recent report. The report notes that YouTube had a four-times bigger network of trusted flaggers in hate speech in 2019 compared with 2016. It also notes that notifications by trusted flaggers have doubled in the first quarter of 2019 compared with the fourth quarter of 2017, and that actions taken for hate speech violations have been almost at the same level (European

Commission, 2019).

208

The Facebook Oversight Board (FOB) was another trial intervention released in May 2020, with a view to rebalance the unequal regulatory power of the platform, and to deal with the ineffectiveness of existing moderation measures (Facebook, 2020c). The board has 40 independent members from all over the world, with different background knowledge. FOB can make content moderation decisions on large-scale, contentious cases. In addition, the board may also scrutinise the policy-making of Facebook, and it has the power to overrule moderation decisions or policies made by Facebook (Facebook, 2020c). While the FOB has only been operating for a short period, it is known for its independence, openness and expertise in terms of regulation, and it has dealt with the problematic content moderation to some extent. However, its implementation is still lacking the participation of ordinary users to govern themselves. Moreover, the platform company can still maintain a self-regulatory governance framework and integrate FOB into its own regulatory system. Therefore, conflicts of interest, under-enforcement of rules and fragmentation of regulation could be the main issues of concern (Flew & Gillett, 2020; Haggart, 2020). Despite these defects, the FOB and Trusted Flagger Program provide promising concepts to bring civil society organisations and daily users into the cooperative regulation of platform companies.

The Discipline Committee of Bilibili: Signalling an alternative approach?

Bilibili is a Chinese video-sharing website that has adopted a community regulation model similar to the Trusted Flagger Program of Facebook. In June 2017, Bilibili released a function called the

Discipline Committee (风纪委员会), which operates like a crowd referee to promote community regulation. Any Bilibili users that are Level 4 or above, who have passed the real-name authentication and have not committed any violation in 90 days, can apply to be a referee as part of the Discipline Committee. Each term as a referee is 30 days, and after that there will be an evaluation on referee performance – a referee who gets a high evaluation for their performance

209

can continue to be a referee for another 30 days as well as winning virtual rewards (Bilibili, 2017).

The duty of a referee is to review reported content: high-profile content is reported by ordinary users in the first place, then sent to the Discipline Committee for further evaluation. As a member of the Discipline Committee, referees can choose from four options: recommend to ban the account; recommend to delete the content; no violation; and abstain. In order to limit the power of individual referees, each referee has only twenty opportunities to vote for a decision per day (Bilibili, 2017).

In addition to making decisions, referees can also write referee arguments (众议) when they are voting, which enables them to provide further information, including their own analysis and discussions with other referees. This referee comments can also be liked and disliked by other referees (Bilibili, 2017).

The establishment of the Discipline Committee marks the setting up of a trial of a high level of independent community regulation. It grants high-level users operational capacity to regulate their own community without interference from platforms or even the government. So far, it is probably the most democratic and participative way for users to govern their own community; and users have been very active in volunteering to join the committee. By July 2018, 36,000 users had been recruited as referees (Sina, 2018). Nonetheless, there are still significant drawbacks at the early stage: the voting does not need to be justified by any rules, including community guidelines or laws. Therefore, the judgement has been criticised because of the arbitrariness and subjectivity at this early stage, and some referees’ actions have be shown to be bureaucratic and to lack procedural justice, as their voting and comments can be highly influential

(Wangnitianxinweijixin, 2018).

Training is necessary prior to becoming a referee on Bilibili, but many users believe that examination before inauguration should also be a requirement, on the basis that it could improve

210

the expertise and filter out the incompetent and unethical candidates (Bilibili, 2020). Another key issue that remains is the lack of a comprehensive notification and appeal system. If an account has been suspended or a piece of content is deleted, the user should be notified of the detailed reasons instead of merely announcing that they have been ‘violating the Community Guidelines’. Such an approach is sound: the best way to educate users is to let them know what kind of content is deemed inappropriate and why. Otherwise, merely censoring content is just a temporary measure and cannot guarantee the decrease of such activities, while punitive measures – including account suspension – seem too harsh for general users, whose conduct is not a serious offence. In this context platforms should also develop appeal guidelines and systems for users to contest the decision, as YouTube (Facebook Help, 2020) and Facebook have already done. In YY there is a way for users to contest the decision made by room regulators, but there is still no way to contest the moderation conducted by the platform moderators themselves (YY Operation Team, 2018).

After years of updates, the co-regulation model of Bilibili has been improved continuously through community effort, and two notable changes have been made. The first is a weekly report that was introduced from the end of 2017. In the weekly report, selected cases – good judgement, bad judgement and controversial cases – are highlighted on the front page of the Discipline

Committee for further reference, which represents a trial ‘case law’ spirit (Bilibili, 2020). The second is the contest system: users who have been punished by the Discipline Committee can make arguments within seven days of punishment taking effect. The platform regulator will review the appealed case and make the final decision (Bilibili Housekeeper, 2018).

To sum up, community regulation is relatively insignificant in today’s platform regulation in China (Z. Zhang, 2020), so there is an imperative to increase the participative rights of users in platform governance. The FOB and Trusted Flagger Program have provided good examples of

211

improving the effectiveness of the co-regulation model, and Bilibili’s Discipline Committee also marks an initial attempt to build up a community regulation system that enables an independent oversight of the content moderation. Despite these alternative mechanisms, the platform’s algorithmic evaluation system remains decisive for the nomination of the referee after one term of services. Bilibili’s Discipline Committee provides us with a good example of how to utilise community members for the co-regulation, particularly for live streaming platforms with tricky video content and numerous bullet-screens. For live streaming platforms in China, community regulation is capable not only of improving the efficiency of regulation and reducing the workload of platform moderators, but also of building up a more ‘equitable and just’ environment for platform users in the long term (Gorwa, 2019b, p. 865). Independence, expertise and operational capacity are the three key elements when developing new community regulatory models. Detailed directions include, but are not limited to, providing training sessions and examinations for room regulators before they commence; selecting trusted regulators and granting them more regulatory privileges; building up a comprehensive notification and appeal systems; and requiring the publication of weekly or fortnightly regulation reports to highlight the cases for further reference, in a spirit of openness and transparency.

Like the FOB, Bilibili’s Discipline Committee is still subject to the algorithmic categorisation and design of the platform company. To deal with this, Winseck (2020) proposes to learn from Germany’s new ‘platform neutrality’ rules and the ranking and sorting algorithms of the big commercial audio-visual platforms. The ‘user choice’ type ‘customizable algorithms’ that underpin this platform neutrality requirement could also be principles for Chinese platforms to develop in the long term, in the spirit of opening up the development of platform architectures to communities.

212

Building up a consistent, cooperative regulatory system

As discussed in Chapter 3, government regulation of social media is still highly dependent on continuous special operation rather than consistently implemented laws. Nowadays, ambiguous regulatory rules can have a chilling effect and to the proliferation of violating content, thereby endangering certain categories of content and making the streaming content more homogeneous.

From another perspective, the vagueness of definitions in relation to what constitutes a violation is giving commercial organisations considerable leeway to filter content based on their own benefit instead of the public interest (Ruberg, 2020), and this is particularly so for the live streaming platforms in which the majority of regulatory work is done through the self-regulation of industrial practitioners. Ruberg (2020) refers to the double standards of Twitch’s content moderation on pornographic content, and similar issues are also prevalent on Chinese platforms. Sexual and vulgar content is not allowed to be displayed on Chinese platforms, so after few years of purifying by special operation and platform moderation, streamers themselves now do not dress ‘sexually’ or speak in ‘vulgar’ ways. However, game streaming’s in-game content still displays violating content. To take one example, Cyberpunk 2077 contains a great deal of bloody action and vulgar language, so it has not yet been approved by Chinese authorities (McGlynn, 2020). However, there are a number of channels streaming Cyberpunk 2077, and there is even an individual category on

Douyu.28 In this situation, there is an urgent need to build a consistent and cooperative content regulatory system.

What could such a regulatory system look like? Consistency here does not mean that a uniform rule should be applicable to all platform sectors all the time, but rather refers to a defined

28 Cyberpunk Channels, from Douyu, https://www.douyu.com/g_Cyberpunk

213

and clarified bottom line of legitimate content. In a Chinese context, the kinds of activity that are illegal, and the kinds of content deemed to be detrimental to political stability and social security, should be defined more clearly without any leeway, changes or space of interpretation. A consistent regulatory system can regulate the explicitly illegal activities, thereby granting users more democratic rights by reducing the chance of changes in moderation. Standards could also be set up for the content moderation rules for the platform’s self-regulation, which could then significantly reduce the workload of real-time manual moderation on live streaming content. This is not to say that a principle-based approach is not applicable and only the detailed regulation policies provide a solution. The point is that a principles-based approach is only a temporary remedy to deal with implicitly inappropriate content in the contemporary situation; in the long term, only nationally uniform, consistent regulatory system can bring fundamental changes to the platform ecosystem and continuously contribute to the building a rule-based legal system.

A cooperative regulatory system would also need to deal with the implicitly inappropriate practices conducted by platform companies or users. The regulation on these ‘legal but unethical’ practices has always been tricky in a global context when the regulatory agencies want to implement publicity beneficial regulatory policies (Haggart, 2020). In the Chinese context, if the aim is to moderate ‘legal but unethical’ practices while maintaining the moderation in a justifiable and reasonable way, then a better option than recruiting more moderators is to facilitate cooperation between different government agencies, commercial entities and user communities to create a framework that automatically defends the public interest in the long term. A mandatory weekly or fortnightly moderation report, such as the Discipline Committee Weekly Report of

Bilibili, may be a way to facilitate the cooperation. By highlighting new cases and contesting notices, the report could serve as a reference for the government and NGOs to keep up with the

214

industrial change and enact new regulations and laws to increase the consistency of rules and reduce the vagueness. The highlighted cases could also be selected by the government to serve as precedents, establishing legal principles for the platforms (Common, 2020). Only through constant ongoing negotiation can the categories and definitions of the content violation be refined in the long term, to handle the ‘unethical’ activities occurring in the grey area.

In conclusion, the self-regulation of the firms and external regulation from the state have always been the key concern of platform regulation in China, but it is also necessary to empower and incorporate community regulation from the point of view of building up a comprehensive co- regulation framework. A consistent and cooperative regulatory system that considers firm–state– user relations is perhaps the best approach to deal with emerging regulatory issues of live streaming.

Nonetheless, another blind spot exists: the absence of regulations around donation practices.

To date, specific rules, laws or regulations that directly define and regulate donations as a form of virtual currency remain absent. The uncertainties in legal interpretation and implementation on virtual currency, the activities of transaction and donations, and especially practices in relation to lucky draws and raffles organised by streamers, might all yet turn out to be illegal activities that involve online gambling. This also means the platforms may potentially be subject to fines and penalties (Douyu, 2019, pp. 39–40). In addition to these issues, donation inflation and fake traffic caused by the collusive practices of platform companies and agent companies discussed in

Chapter 4 constitute a sense of commercial fraud or cartel behaviour, which might violate competition policies; however, this has not yet been investigated fully by government agencies.

Therefore, an updated supervision framework on the flow of virtual currency, and stricter regulation for restraining the power of platform companies and agents, is an issue that may yet emerge on the agenda.

215

Engaging with the world: International and cross-cultural negotiation via live streaming governance

An anti-monopoly campaign on platform economy: The real beginning for platform governance?

In November 2020, in order to curb the rampant acquisition and the increasingly monopolistic trend of the tech giants in China, and specifically to prevent and stop monopolistic behaviours of the platform companies, the State Administration for Market Regulation (SAMR) drafted the Anti- monopoly Guidelines on the Platform Economy Field (SAMR, 2020b). In the meantime, SAMR started a well-publicised antitrust investigation into Alibaba Group Holdings Ltd. An editorial in the People’s Daily commented that, ‘This investigation does not mean that the country’s attitude towards the encouragement and support of the platform economy has changed’, but stated that it was necessary for ‘improving the socialist market economy system and promoting high-quality development’ (Davidson, 2020).

In terms of platform governance, two common practices on the platforms will be forbidden according to the upcoming Anti-monopoly Guidelines on the Platform Economy Field (SAMR,

2020a). The first is Alibaba’s so-called ‘Choosing One from Two’ policy, whereby online merchants have to sign exclusive agreements with Alibaba and are prevented from selling products on other platforms (Davidson 2020). According to Article 15 of the Guidelines, the dominant platform company may abuse its power and control the commercial behaviour of platform users and restrict market competition. Moreover, as SAMR (2020a) observes, these restrictions on users are not only in the form of written contracts, but may also be enforced by the platform companies through ‘oral agreement, architecture design, rules and algorithms of the platform’. All these restrictions on users will be deemed unfair competition, and will be investigated by SAMR. This detailed list of prohibitions indicates that the Chinese state will take ‘governance by platform’ and

216

the context of the Chinese labour market into consideration when enacting new regulations. The second practice, Pricing Variation, as Article 17 outlines, will also be deemed unfair competition.

For marketing and maximising profit, the dominating platforms usually use data extraction and big-data analytics to roll out differential pricing based on the personal data of consumers. While the legitimacy of data extraction is still controversial, so-called pricing variation strategies will be classified as anti-competitive behaviour and any suspect platforms will be investigated according to the new guidelines (SAMR, 2020a).

One month after the release of the guidelines, on 11 December 2020, China’s Politburo restated the importance of reinforcing anti-monopoly policies to prevent ‘disorderly capital expansion’ (Y. Liu, 2020). Following the announcement, several tech giants in China have been investigated and penalised. On 14 December 2020, Alibaba was fined for ‘not properly reporting increased share stake in their retail chain in-time’, and the Tencent-backed China Literature Group was fined for not reporting an acquisition (Frater, 2020). While the Anti-monopoly Guidelines on the Platform Economy Field had not taken effect at the time of writing, this punishment is backed by a 2008 anti-monopoly law, and it shows the state’s clear determination to curb the platformisation led by tech giants. On the same day, two Chinese live streaming platform were also put on the investigation list when SAMR began its investigation into the proposed merger of game live-streaming companies DouYu and Huya, suggesting that this merger would also breach anti-monopoly laws (Frater, 2020). As examined in Chapter 4, Tencent has become the largest shareholder of both Douyu and Huya (Douyu, 2020), and it will take 80 per cent of the Chinese game streaming market after the merger, thus causing it to become a monopoly (Frater, 2020).

As previous chapters have noted, these dominant Chinese live streaming platforms are squeezing out smaller businesses and individual users, causing serious regulatory issues and

217

content degeneration. The recent actions by the state to curb anti-competitive behaviours of the platforms not only show signs of dealing with monopolies, but they also indicate the state’s increasing realisation of a need to curb the governance power of the platform companies. The guidelines thus highlight the algorithmic and architectural control of the platform, the informal contract between users and platform companies, and data extraction and pricing variations. All these techniques correspond to the concentrating ownership of the platforms, compulsive use of exclusive contracts and informal agreements, manipulative structures, rogue agencies, the poor working conditions of streamers, and a general trend towards degenerate and homogenised content in China’s live streaming industries. Although the detail of measures to enforce the guidelines are still in discussion, this proposed agenda and recent special operation/campaign have shown that the Chinese state is prepared to initiate a program of longer-term regulatory enforcement for platform regulation.

Digital sovereignty and live streaming in a global context

The enforcement platform economy regulation demonstrates the will of the Chinese state to act on anti-trust laws, but clearly the state retains its power over digital sovereignty. In a global context, transnational corporations, global production networks and international governing entities increasingly incorporate different actors into digital capitalism, pressing developing countries to conform and to participate in the US-dominated techno-economic framework. In this framework,

China has become a downstream manufacturing powerhouse in the past few decades (Zhao, 2008).

However, the situation on the internet morphed when China asserted its digital sovereignty. For

China, although state sovereign authority is contested by the global capitalist system led by the

United States, the Chinese state always asserts its sovereign rights and borders (Hong & Goodnight,

2020). While using the ‘Wall’ to block international platforms was one way of claiming such

218

digital sovereignty, its introduction was arguably somewhat rushed. Subsequently the measure has been widely criticised for violating human rights. However, the rationale behind its rollout was justified by the state in these terms: China believes the unilateral control exerted by the United

States over the digital media resulted in global structural imbalances, and therefore it became necessary to defend its own interests and cybersecurity (Hong & Goodnight, 2020, p. 17). From this point of view, it was reasonable that China consistently blocked Google, Twitter and YouTube, and this consequently provided an opportunity for local digital platforms to grow without exposure to fierce global competition (Zhao, 2019, p. 12). Moreover, the huge domestic user base and the state’s continuous efforts around building digital infrastructure then laid the foundations for the rise of Chinese internet companies, thus constituting a ‘parallel universe’ in the form of a digital media ecosystem that rivals US companies (Birkinbine et al., 2016, pp. 342–48).

In terms of live streaming, China quickly blocked Twitch when live streaming became popular from 2018 (BBC News, 2018b). After that time, it was no longer impossible for Chinese viewers or stream or to use Twitch in a normal way. Furthermore, the Administrative Measures for the Operation of Network Performance require that from March 2017, foreign streamers or streamers external to mainland China should apply for permission to Ministry of Culture and

Tourism before streaming on Chinese platforms (MCT, 2016b). From the perspective of Western viewers, watching live streaming on Chinese platforms is never a good option given the language barrier and taste variations. In this sense, there is very little cross-cultural negotiation from a civil society level, with one important exception: international students or other overseas ethnic Chinese can stream on Chinese platforms when they are abroad. Douyu has recruited many international students to show Western lives to the Chinese domestic audience in the form of live shows (Douyu,

2017). There are some interesting examples where, at the national level, the state has tried to utilise

219

live streaming for political purposes. To take one example, on 23 July 2020, China ordered the

United States to close its Consulate General in Chengdu City as a counterplay to the American closure of the Chinese Consulate General in Houston. The whole process was then live streamed in CGTN’s official channel on YouTube, with tens of thousands of live viewers watching and discussing in the Live Chat (CGTN, 2020).

From a platform level, it seems that US and Chinese platforms have been learning monetisation mechanisms from each other. For example, the preference of promoting live shows over game streaming has expanded to Twitch: ‘Just Chatting’ is a new category on Twitch from

2018, but it has quickly emerged as a category of content that outweighs game streaming. On ‘Just

Chatting’ channels, streamers mainly chat with viewers or showcase other creative activities without gameplay. Although this new category bothers some hardcore game players, reports show that the growth of Twitch over these two years is largely due to the growth of non-game content, including the ‘Just Chatting’ feature (Khalid, 2019). In addition, both Douyu and Twitch have introduced ‘gamblified’ monetisation mechanisms in recent years (Abarbanel & Johnson, 2020), despite gambling being strictly forbidden in China. There were reported cases of viewers losing millions to Douyu (Wang & Lan, 2021). Clearly, for the Western live streaming platforms, new genres of live streaming content have brought about a higher level of interaction that generates more profit for the platform companies; however, this could also give rise to regulatory issues, given that the live show format entails more toxic content from viewers compared with game streaming (Khalid, 2019) and the new monetisation of the live interaction will also intensify the process of commodification and datafication, as it has in China.

It is clear that the platform logic itself cannot be limited by borders or national sovereignty, and the ‘parallel universe’ still has intersections in terms of platformisation and commercial

220

activities. Platformisation has been evolving simultaneously around the world without strong restrictions, and such expansion has been confronted by economic and political conflicts between

China and the United States – the ban on WeChat and TikTok was predictable. Nonetheless, the expansion of live streaming platforms went smoothly and these platform companies have tried to avoid invading each other’s territory. Twitch already has a dominant position in North America and the European Union, as it seeks further expansion in Australia (Gillezeau, 2020). And Chinese live streaming companies such as YY and Douyu are in the early stages of extending their international market reach, starting from Asian countries such as Singapore, India and Japan

(Cheng, 2019; Singh, 2019; C. Zhang, 2019). In this situation of international expansion, who should be responsible for the content regulation, and what kind of regulatory body should be established to govern the platform and prevent the platform from exploiting users? Given that inconsistency has always been an issue for platform self-regulation (Common, 2020), and the effectiveness of external regulation is weakened by the different standards and conflicting legal systems of different countries, can we imagine a situation where users are empowered and granted autonomy to regulate themselves?

Conclusion and directions for future research

Over the last decade, we have witnessed dramatic changes in the digital platforms in a global context, and there is an urgent need to review outdated media and communication policies to deal with ever-changing national contexts. The problem of harmful content, data misuse, labour exploitation and the increasing political and economic power of platform companies is becoming more urgent, calling for the analysis of ‘regulation by the platform’. How do platforms control and mould communicative activities through their proprietary algorithms, interface, contracts and

221

policies? Consequently, we also need to reconsider the ‘regulation of the platform’ and consider how public sectors can work together to regulate the platforms in order to defend and maintain the public interest in the long-term.

From this point, the regulation of Western countries is moving from an era where platform companies regulate themselves to a context of increasing nation-state driven intervention (Flew &

Gillett, 2020). By contrast, or at least in terms of regulating live streaming sectors, the Chinese state has always tended to pursue a policy of direct intervention, and now it has moved forward to build up a new regulatory framework whereby it passes the regulatory responsibility and workload over to the platforms through forms of self-regulation and co-regulation by non-state actors. This direct intervention may be launched at any time when there are concerns about political and social stability, but in most situations the state acts as a stakeholder or facilitator in the multi-stakeholder system. Often the state relies on a paternalistic model of regulation to maintain the public interest in the form of special operations and campaigns, as a reaction to the imbalanced economic structure and grassroots efforts.

Therefore, the platformisation of live streaming in China is not totally subjected to the platform logics, but it is partly constrained or promoted by the state, while hinting at a democratic and participatory potential as the grassroots effort is driving the state to make further changes.

Unlike in the West, we need to seek alternatives to using frameworks of labour exploitation, or models that counterpose freedom of speech versus content moderation; social obligations seems a more amenable approach for understanding platform regulation in a Chinese context. In particular, the novel content-based paradigm of live streaming is still largely outside platform logic. More importantly, with less vague and more consistent state policies, and potential community regulation models, the grassroots approach offers the capability to truly govern and control the

222

platform structure that affects viewers. Therefore, analysis from the perspectives of participatory culture, economic counter-power, configurations of firm–state–user community relations and paternalistic model of regulation can all assist in furthering the understanding of the mechanisms of platformisation and platform regulation of Chinese live streaming. In addition, evolving Chinese regulatory policies and industrial practices provide abundant experience of how to handle the implicitly illegal content, or ‘legal for now but unethical’ content, that is so prevalent on social media platforms.

Despite the divergence, the theoretical framework outlined in the first chapter is capable of situating and explaining the regulation of Chinese live streaming platforms. The CPEC/PR approach provides us with a framework to understand how imbalanced power relationships have taken hold, and thus become an obstacle to regulation: this is revealed through the analysis of the historical transformation of related culture production, distribution and circulation of live streaming content productions within the logic of platform economy (Poell et al., 2019; Van Dijck et al., 2018). The political-economic system underpinning these unfair relationships is exposed, as are the key regulatory stakeholders. In addition, no matter how complex and intertwined the multi- stakeholder ecosystem seems to be in the Chinese context, a platform governance framework can help to clarify the regulatory relationships and lead us to practical solutions. First, we need to ask how ‘regulation by platform’ takes place – that is, how platforms and affiliated commercial entities might control and mould communicative practices through algorithms, interfaces, contracts and platform policies. Second, we need to understand how ‘regulation by platform’ can itself be regulated by public sectors in a justifiable and clear way, thereby making the platform economy more fully beneficial and sustainable for the majority rather than the few (Gorwa, 2019b, p. 865).

223

Moreover, the analysis of the platform regulation in China offers opportunities for researchers to better understand the concept of ‘affordance’ and avoid technological determinism.

Although live streaming platforms are evolving twenty-first century technologies, the live streaming ecosystem is unique because of the human actors who utilise it. The possibility of live streaming being more widely used by the Chinese underclass, or grassroots users within distinctive social structures, is shaping very different dynamics of information production, distribution and circulation. More open operations in terms of moderation, and even the development of platform architectures involving community members and public authorities, platform affordance and user practices, can mutually benefit from each other’s participation (Poell, 2020, pp. 653–54).

Finally, there are several limitations to this research that should be mentioned. Because regulatory issues remain sensitive in Chinese contexts, my research into live streaming companies and streamers, particularly interviews with staff from agent companies, often encountered difficulties, with participants reluctant to answer questions. Many participants preferred avoidance, and some rejected my requests for an interview. Thus, although the information collected from the interviews is highly informative, it could be more comprehensive if access to unequivocally cooperative participants were available. In addition, this study takes a more holistic approach to analysing the complexity of the platformisation of live streaming, which means that some specific issues have not been fully addressed. For example, privacy and copyright issues are becoming increasingly significant on live streaming platforms (Matsui, 2016;

Tao, 2018), especially when the these companies are becoming more global rather than being limited to a domestic scope. In dealing with oligopolistic or monopolistic markets, not only can we restrain big platforms, but supporting smaller platforms may also need to be a strategy to confront tech giant-led platformisation. In addition to tax reductions and governmental loans that have been widely adopted by the state to support small tech business (Hong, 2017a), platform regulation responsibility could also be

224

varied, depending on the size of the company. Small live streaming platforms, as noted in Chapter 3, can hardly afford the high costs of real-time monitoring as well as obtaining licences, and many of these companies tend to be more speculative and thus indulge the growth of violating content. The closure of smaller platforms in recent years has actually contributed to the oligopoly of tech giants. In this situation, regulators should ask whether it is possible to relieve the self-regulation responsibility of the smaller platforms and let a regulatory body to do certain tasks. For example, Australia’s e-Safety Commissioner operates a Social Media Tier Scheme,29 which set ups different levels of regulatory responsibility depending on the size and influence of the platform, where a self-regulation system is voluntary for smaller platforms but mandatory for the largest. The point is that to maintain a healthy market and confront monopolistic conduct, it is necessary to ensure that self-regulation responsibility itself does not become a burden for smaller platforms and, further, that it does not become an obstacle to a free market.

In China, is it possible to relieve the regulatory cost and responsibility of the small platforms while keeping the content legitimate? As there are many existing third-party moderators undertaking moderation for the platforms, could the government directly approve the authorised third-party moderators to do this work? All these questions could be addressed by future research.

29 See Social media Tier Scheme from Australian ESafety Commissioner, https://www.esafety.gov.au/about- us/who-we-are/social-media-tier-scheme

225

References

Abarbanel, B., & Johnson, M. R. (2020). Gambling engagement mechanisms in twitch live

streaming. International Gambling Studies, 20(3), 393–413.

Abbott, K. W., & Snidal, D. (2009). The governance triangle: Regulatory standards institutions

and the shadow of the state. The Politics of Global Regulation, 44, 44–88.

Anand, P. (2020, Jan 8). How big is Amazon’s Twitch? Revenues reveal ad struggles. The

Information. https://www.theinformation.com/articles/how-big-is-amazons-twitch-

revenues-reveal-ad-struggles

Andrejevic, M. (2004). The webcam subculture and the digital enclosure. In M. Andrejevic,

MediaSpace (pp. 207–222). Routledge.

Andrejevic, M. (2007). Surveillance in the digital enclosure. The Communication Review, 10(4),

295–317.

Article 19. (2020, Jun 13). Facebook: Improvements in transparency reporting more urgent

amid coronavirus pandemic. Article 19. https://www.article19.org/resources/facebook-

improvements-in-transparency-reporting-more-urgent-amid-coronavirus-pandemic

Auverset, L. A. (2017). Social TV is the new digital watercooler: Personality traits, behaviors,

and trends in second-screen media activity. [Doctoral dissertation, University of

Alabama].

Baruah, T. D. (2012). Effectiveness of Social Media as a tool of communication and its potential

for technology enabled connections: A micro-level study. International Journal of

Scientific and Research Publications, 2(5), 1-10.

226

BBC. (2018, Aug 7). Tales from modern China. South East Tales from Modern China. Episode 5

of 6. BBC. https://www.bbc.co.uk/programmes/n3ct3g02

BBC News. (2018, Sep 21). China blocks Twitch game-streaming service. BBC News.

https://www.bbc.com/news/technology-45600272

Benkler, Y. (2006). The wealth of networks: How social production transforms markets and

freedom. Yale University Press.

Bilibili. (2017, Apr 8). Basics of Discipline Committee. Bilibili.

https://www.bilibili.com/read/cv363368

Bilibili. (2020, Dec 28). 2020 年十二月第四周风纪委员周报 (Discipline Committee report

Dec Week 4). Bilibili. https://www.bilibili.com/read/cv9022892

Bilibili Housekeeper. (2018, Apr 26). Discipline Committee update V4.0. Bilibili.

https://www.bilibili.com/read/cv417902

Bingham, C. M. (2020). Talking about Twitch: Dropped frames and a normative theory of new

media production. Convergence, 26(2), 269–286.

Birkinbine, B., Gómez, R., & Wasko, J. (2016). Global media giants. Routledge.

Bolsover, G. (2017). Computational propaganda in China: An alternative model of a widespread

practice, S. C. Woolley & P. N. Howard, Eds. https://ora.ox.ac.uk/objects/uuid:b081ab3f-

a54d-40ad-896d-462d1f386042

Brady, A.-M. (2009). Marketing dictatorship: Propaganda and thought work in contemporary

China. Rowman & Littlefield.

Brady, A.-M. (2017). Plus ça change? Media control under xi Jinping. Problems of Post-

Communism, 64(3–4), 128–140.

227

Bruns, A. (2008). Blogs, Wikipedia, Second Life, and beyond: From production to produsage.

Peter Lang.

Bucher, T. (2012). Want to be on the top? Algorithmic power and the threat of invisibility on

Facebook. New Media & Society, 14(7), 1164–1180.

Bucher, T. (2018). If ... then: Algorithmic power and politics. Oxford University Press.

CAA. (2020, Jul 2). Code of conduct for e-commerce live streaming. CAA. http://www.china-

caa.org/uploads/downloads/zbyxgf0624.pdf

CAC. (2016, Nov 4). CAC issued ‘Internet live service management regulations’. CAC.

http://www.cac.gov.cn/2016-11/04/c_1119846202.htm

CAC. (2017, May 2). Internet news information service management regulations. CAC.

http://www.cac.gov.cn/2017-05/02/c_1120902760.htm

CAC. (2018). 具有舆论属性或社会动员能力的互联网信息服务安全评估规定 (Regulations

on security evaluations on internet service providers that have public opinion nature or

the capacity for social mobilization). CAC. http://www.cac.gov.cn/2018-

11/15/c_1123716072.htm

CAC. (2018b, Feb 13). CAC investigated and punished a number of severely violating live

stream platforms and streamers. CAC. http://www.cac.gov.cn/2018-

02/13/c_1122415948.htm

CAC. (2020, Mar 18). 2020 年网络扶贫工作要点印发实施 (Key points of online poverty

alleviation in 2020 are publised and implemented). CAC. http://www.cac.gov.cn/2020-

03/18/c_1586073208099659.htm

Cai, J., & Wohn, D. Y. (2019). Live streaming commerce: Uses and gratifications approach to

understanding consumers’ motivations. Proceedings of the 52nd Hawaii International

228

Conference on System Sciences,

https://www.researchgate.net/publication/331024408_Live_Streaming_Commerce_Uses_

and_Gratifications_Approach_to_Understanding_Consumers’_Motivations

Cameron, J. A. (2016). The on-screen water cooler: Effects of televised user-generated

comments on cognitive processing, social presence, and viewing experience [PhD thesis,

University of Tennessee].

Cao, X. (2019). Bullet screens (danmu): Texting, online streaming, and the spectacle of social

inequality on Chinese social networks. Theory, Culture & Society, 33(6), 3–23.

CCTV. (2020, Jun 8). 莫让网课变“网游”(Don’t make online lessons become ‘online games’).

CCTV.

https://tv.cctv.com/2020/06/08/VIDEQp3NRk8L9YO8vJfGDtoi200608.shtml?spm=C53

156045404.PKXC0xLPAnP9.0.0

CGTN. (2020, Jul 20). Live: Exterior of the U.S. Consulate General in Chengdu [Video File].

CGTN. https://www.youtube.com/watch?v=I9E09BekUH4

Chan, K. W. (2010). The global financial crisis and migrant workers in China:‘There is no future

as a labourer; returning to the village has no meaning’. International Journal of Urban

and Regional Research, 34(3), 659–677.

Chasmar, J. (2018, Jul 17). Facebook apologizes to Diamond and Silk: 'We badly mishandled'

their censorship concerns. Washington Times.

https://www.washingtontimes.com/news/2018/jul/17/facebook-apologizes-diamond-silk-

we-badly-mishandl

229

Chen, H. (2019, Feb 6). Chinese government identifies esports as a profession, PUBG details

esports plans for China. The Esports Observer. https://esportsobserver.com/china-recap-

feb6-2019

Chen, L. (2020, Jan 22). How is technology changing Chinese New Year? CGTN.

https://news.cgtn.com/news/2020-01-22/How-is-technology-changing-Chinese-New-

Year--Nswc5sHWyk/index.html

Cheng, X. (2019, Mar 5). YY 花 14.5 亿美元全资收购的 Bigo 是一家什么样的公司？(What

kind of company is Bigo, which YY spent US$1.45 billion wholly-owned to acquire?).

http://news.iresearch.cn/content/201903/284983.shtml

Chen, X., & Chen, Z. (2020). Between the marked and the unmarked: Twin semiotic paradoxes

of the barrage in China’s livestreaming fandom. Media, Culture & Society, 42(5), 727–

744.

Chinadaily. (2010, Aug 6). Minister hits at ‘vulgarity’ in culture sector. China Daily.

http://www.chinadaily.com.cn/china/2010-08/06/content_11113082.htm

Chuanmeitoutiao. (2017, March 16). 最新的 588 家持证机构名单在这，视听许可证炒至

3500 万元 (The list of the latest 588 licence holders is here, and the audio-visual licence

is up to 35 million yuan). Sohu. http://www.sohu.com/a/129114810_570250

Chun, W. H. K. (2011). Programmed visions: Software and memory. MIT Press.

CNNIC. (2017). 39th statistical report of Chinese internet development. CNNIC.

http://www.cnnic.net.cn/hlwfzyj/hlwxzbg/hlwtjbg/201701/P020170123364672657408.pdf

CNNIC. (2019). CNNIC: The 44th survey report. CNNIC.

https://cnnic.com.cn/IDR/ReportDownloads/201911/P020191112539794960687.pdf

Cohen, J. E. (2017). Law for the platform economy. UCDL Review, 51, 133.

230

Cohen, N. S. (2008). The valorization of surveillance: Towards a political economy of Facebook.

Democratic Communiqué, 22(1), 5–22.

Common, M. F. (2020). Fear the reaper: How content moderation rules are enforced on social

media. International Review of Law, Computers & Technology, 34(2), 126–152.

Consalvo, M. (2017). Player one, playing with others virtually: What’s next in game and player

studies. Critical Studies in Media Communication, 34(1), 84–87.

Cross, K. (2016). Press F to revolt. In Y. B. Kafai, G. T. Richard & B. Tynes (Eds.), Diversifying

Barbie and Mortal Kombat: Intersectional perspectives and inclusive designs in gaming

(pp. 23–34). ETC Press.

Cui, D., & Wu, F. (2016). Moral goodness and social orderliness: An analysis of the official

media discourse about Internet governance in China. Telecommunications Policy, 40(2–

3), 265–276.

Culture-Association, B. I. (2017, Dec 20). 北京网络直播发起自律公约 (Beijing Live

streaming Released Self-discipline Convention).

http://www.bjwlwh.cn/xhhd/20160413/2726.html

Cunningham, S., Craig, D., & Silver, J. (2016). YouTube, multichannel networks and the

accelerated evolution of the new screen ecology. Convergence, 22(4), 376-391.

Cunningham, S., Craig, D., & Lv, J. (2019). China’s livestreaming industry: Platforms, politics,

and precarity. International Journal of Cultural Studies, 22(6), 719–736.

Custer, C. (2016, Apr 16, May 15). China’s live-streaming platforms are all under investigation

for sex, violence, and crime. Sina. https://www.techinasia.com/chinas-livestreaming-

platforms-investigation-sex-violence-crime

231

David, M., Kirton, A., & Millward, P. (2015). Castells,‘Murdochization’, economic

counterpower and livestreaming. Convergence, 23(5), 497–511.

Davidson, H. (2020, Dec 24). China targets Alibaba with anti-monopoly investigation. The

Guardian. https://www.theguardian.com/business/2020/dec/24/china-targets-alibaba-

with-anti-monopoly-investigation

Davis, M., & Xiao, J. (2021). De-Westernizing platform studies: History and logics of Chinese

and U.S. platforms. International Journal of Communication, 15.

https://ijoc.org/index.php/ijoc/article/view/13961 de Kloet, J., Poell, T., Guohua, Z., & Yiu Fai, C. (2019). The platformization of Chinese society:

Infrastructure, governance, and practice. Chinese Journal of Communication, 12(3), 249–

256.

Deng, I. (2020, Oct 13). Live-streaming e-commerce is the fastest-growing area of China’s

internet, but buyer’s remorse is common. SCMP. https://www.scmp.com/tech/e-

commerce/article/3105315/live-streaming-e-commerce-fastest-growing-area-chinas-

internet

Dexter. (2011, Dec 11). PIS’ debut video is finally out. GosuGamers.

https://www.gosugamers.net/news/16827-pis-debut-video-is-finally-out

Ding, X. (2016, May 13). Secret trader of live broadcast revealed: Eat and drink and clothing of

90% of the streamers are controlled. ICPCW.

http://www.icpcw.com/Information/Tech/News/3273/327382.htm

Donald, S. H., & Keane, M. (2002). Media in China: New convergences, new approaches (pp. 3–

17). In S. H. Donald, M. Keane, & Y. Hong (Eds), Media in China: Consumption,

content and crisis. Routledge.

232

Douyu. (2016, Feb 14). 斗鱼直播扣分系统上线 直播间违规管理方法 (Douyu credit

deduction system update, Live Streaming Room Regulation Code). Douyu.

https://www.douyu.com/cms/gong/201602/14/3078.shtml

Douyu. (2017, Jun 7). 留学生主播招募 逛遍全球怕不怕 (Recruitment of international

student streamers, are you afraid to travel around the world). Douyu.

https://www.douyu.com/cms/huo/201706/07/5463.shtml

Douyu. (2019). Prospectus of DouYu International Holdings Limited.

https://www.sec.gov/Archives/edgar/data/1762417/000104746919004260/a2239266z424

b4.htm

Douyu. (2020a). DouYu 2019 annual report. Douyu. https://ir.douyu.com/Annual-Reports

Douyu. (2020b). Douyu directory. Douyu. https://www.douyu.com/directory

Douyu. (2020c). 斗鱼公会主播入驻服务协议 (Douyu Agents Streamers Service Agreement).

Douyu. https://www.douyu.com/protocal/roomclub

Douyu. (2020d). 正能量直播 (Positive energy live streaming). Douyu. Retrieved Feb 2 from

https://www.douyu.com/g_znl

Douyu. (2020e, Oct 12). Huya and Douyu enter into merger agreement. Douyu.

https://ir.douyu.com/2020-10-12-Huya-and-DouYu-Enter-into-Merger-Agreement

Douyuservice. (2017, Nov 17). 斗鱼新排序算法上线啦——热度高才是王道 (Douyu’s new

ranking algorithm is online – high ‘redu’ is king). Douyu.

https://www.douyu.com/cms/gong/201711/17/6722.shtml

233

Ducheneaut, N., Moore, R. J., Oehlberg, L., Thornton, J. D., & Nickell, E. (2008). Social TV:

Designing for distributed, sociable television viewing. International Journal of Human–

Computer Interaction, 24(2), 136–154.

Duffy, B. E. (2017). (Not) getting paid to do what you love: Gender, social media, and

aspirational work. Yale University Press.

Enli, G., Raats, T., Syvertsen, T., & Donders, K. (2019). Media policy for private media in the

age of digital platforms. European Journal of Communication, 34(4), 395–409.

European Commission. (2019). Progress on combating hate speech online through the EU Code

of Conduct 2016–2019.

https://ec.europa.eu/info/sites/info/files/aid_development_cooperation_fundamental_right

s/assessment_of_the_code_of_conduct_on_hate_speech_on_line_-_state_of_play__0.pdf

Facebook. (2020a). Community Guidelines.

https://www.youtube.com/howyoutubeworks/policies/community-

guidelines/#community-guidelines

Facebook. (2020b). Community Standards. https://www.facebook.com/communitystandards

Facebook. (2020c). Facebook Oversight Board. https://www.oversightboard.com

Facebook. (2020d). Terms of Service. https://www.facebook.com/terms.php

Facebook Help. (2020). Appeal Community Guidelines actions.

https://support.google.com/youtube/answer/185111?hl=en

Fan, J., & Zhang, Q. (2018). The impact of interactivity on virtual gifts giving intent – based on

live-streaming platforms. Paper presented to 3rd International Conference on Humanities

Science, Management and Education Technology (HSMET 2018).

234

Fang, T. Y., & Hao. (2017, Dec 15). Live streaming bubble burst after 15 months. A Play

Ground. https://aplayground.com/app/?p=144226

Filtering Practices of Social Media Platforms (115th Cong), Judiciary Committee (2018).

https://judiciary.house.gov/legislation/hearings/filtering-practices-social-media-platforms

Finance Sina. (2019, Jul 7). 27 省份平均工资出炉 (Average salary in 27 provinces has been

released). Sina. https://finance.sina.com.cn/china/gncj/2019-07-07/doc-

ihytcitm0216579.shtml

Finance Sina. (2020, Aug 10). 斗鱼虎牙合并消息尘埃落定 大股东腾讯发初步建议书

(Douyu Huya merger news settles, major shareholder Tencent issues preliminary

proposal). Sina. https://finance.sina.com.cn/chanjing/gsnews/2020-08-10/doc-

iivhuipn7911106.shtml

Flew, T. (2015). Social media governance. Social Media + Society, 1(1).

https://journals.sagepub.com/doi/full/10.1177/2056305115578136

Flew, T. (2016). Global media and national policies: The return of the state. Palgrave

Macmillan.

Flew, T. (2019). The platformized internet: Issues for internet law and policy. Journal of Internet

Law, 22(11), 3–16.

Flew, T. (2021). Platform regulation and governance. In Regulating Platforms. Polity Press.

Flew, T., & Gillett, R. (2020). Platform policy: Evaluating different responses to the challenges

of platform power. Paper accepted for International Association for Media and

Communication Research (IAMCR) annual conference, Tampere, Finland.

235

Flew, T., Iosifidis, P., & Steemers, J. (2016). Global media and national policies: The return of

the state. In Global media and national policies: The return of the state (pp. 1–15).

Springer.

Flew, T., & Suzor, N. P. (2018). Global media industries and media policy. In Making Media

(pp. 159–170). Amsterdam University Press.

Fong, M. W. (2009). Digital divide between urban and rural regions in China. The Electronic

Journal of Information Systems in Developing Countries, 36(1), 1–12.

Fortney, L. (2019, Oct 20). How Amazon’s Twitch platform makes money. Investopedia.

https://www.investopedia.com/investing/how-does-twitch-amazons-video-game-

streaming-platform-make-money

Frater, P. (2020, Dec 14). Monopoly regulators to probe Tencent’s DouYu-Huya merger and

punish China Literature and Alibaba. Variety. https://variety.com/2020/biz/asia/china-

monopoly-regulators-probe-douyu-huya-merger-punish-alibaba-1234852880/

Friedman, G. (2014). Workers without employers: Shadow corporations and the rise of the gig

economy. Review of Keynesian Economics, 2(2), 171–188.

Fu, Y. (2020, Jun 30). 8 年融资近 20 亿的直播平台陷欠薪旋涡：情况比较复杂 (The live

streaming platform, which has financed nearly 2 billion in 8 years, is in a whirlpool of

salary arrears: the situation is more complicated). Sohu.

https://www.sohu.com/a/404951726_649045

Fuchs, C. (2013). Social media: A critical introduction. Sage.

Fuchs, C. (2014a). Digital labour and Karl Marx. Routledge.

Fuchs, C. (2014b). Digital prosumption labour on social media in the context of the capitalist

regime of time. Time & Society, 23(1), 97–123.

236

Galloway, A. R. (2004). Protocol: How control exists after decentralization. MIT Press.

Gao, M. (2018). 浅谈 “网红” 传播热的社会效应 (Discussing the social effect of popular of

online celebrities). The Collection (收藏), 1, 58.

Gehl, R. W. (2014). Reverse engineering social media: Software, culture, and political economy

in new media capitalism. Temple University Press.

Gibbs, S. (2017, May 3). Facebook Live: Zuckerberg adds 3,000 moderators in wake of murders.

The Guardian. https://www.theguardian.com/technology/2017/may/03/facebook-live-

zuckerberg-adds-3000-moderators-murders

Gibbs, S. (2017, April 25). Facebook under pressure after man livestreams killing of his

daughter. The Guardian.

https://www.theguardian.com/technology/2017/apr/25/facebook-thailand-man-

livestreams-killing-daughter

Gillespie, T. (2014). The relevance of algorithms. In T. Gillespie, P. J. Boczkowski, & K. A.

Foot (Eds.), Media technologies: Essays on communication, materiality, and society

(p. 167). MIT Press.

Gillespie, T. (2017). Governance of and by platforms. SAGE handbook of social media (pp. 254–

278). Sage.

Gillespie, T. (2018). Custodians of the internet: Platforms, content moderation, and the hidden

decisions that shape social media. Yale University Press.

Gillezeau, N. (2020, Aug 24). Live-stream platform Twitch eyes Australian expansion.

Australian Financial Review. https://www.afr.com/companies/media-and-marketing/live-

stream-platform-twitch-eyes-australian-expansion-20200819-p55nd6

237

Ginosar, A. (2013). Media governance: A conceptual framework or merely a buzz word?

Communication Theory, 23(4), 356–374.

Global Times. (2020, Dec 15). News anchor joins hands with live streaming celebrity to boost

Hubei's economic resumption; $5.7 million of Hubei products sold in 2 hours.

https://www.globaltimes.cn/content/1184870.shtml

Gorwa, R. (2019a). The platform governance triangle: Conceptualising the informal regulation of

online content. Internet Policy Review, 8(2), 1–22.

Gorwa, R. (2019b). What is platform governance? Information, Communication & Society,

22(6), 854–871.

Gow, M. (2017). The core socialist values of the Chinese dream: Towards a Chinese integral

state. Critical Asian Studies, 49(1), 92–116.

Gräve, J.-F. (2017). Exploring the perception of influencers vs. traditional celebrities: Are social

media stars a new type of endorser? Proceedings of the 8th International Conference on

Social Media & Society.

Greenfield, T., & Greener, S. (2016). Research methods for postgraduates. John Wiley & Sons.

Guan, Z., & Zhang, L. (2020). 阿里巴巴的进化史与小资本主义的平台化: 对本土语境平台

化的考察 （The evolutionary history of Alibaba and the platformization of small

capitalism: An investigation of the platformization in local context). 国际新闻界

（Chinese Journal of Journalism & Communication), 42(2), 28–49.

Guarriello, N.-B. (2019). Never give up, never surrender: Game live streaming, neoliberal work,

and personalized media economies. New Media & Society, 21(8), 1750–1769.

Guo, Q. (2014). Review on the latest development of Chinese Media Industry. China Media

Report Overseas, 10(3), 82–89.

238

Guynn, J. (2007, March 30). IT’S JUSTIN, LIVE! ALL DAY, ALL NIGHT! / S.F. startup puts

camera on founder’s head for real-time feed, and a star is born. San Francisco Chronicle.

https://www.sfgate.com/news/article/IT-S-JUSTIN-LIVE-ALL-DAY-ALL-NIGHT-S-F-

2606536.php

Haggart, B. (2020). Global platform governance and the internet-governance impossibility

theorem. Journal of Digital Media & Policy, 11(3), 321–339.

Hambrick, M. E., Simmons, J. M., Greenhalgh, G. P., & Greenwell, T. C. (2010). Understanding

professional athletes’ use of Twitter: A content analysis of athlete tweets. International

Journal of Sport Communication, 3(4), 454–471.

Han, D. (2020, 2020 Jul 21). 直播平台被约谈后仍现低俗内容? (After being interviewed, the

live streaming platform still shows vulgar content?). Chinanews.

http://www.chinanews.com/sh/2020/07-21/9243620.shtml

Han, Y., & Shengchen, Z. (2017). 中国网络游戏视频直播现状研究——以斗鱼 TV 为例

(Game Live streaming of China – Case of Douyu TV). 今传媒 (Today’s Mass Media),

25(4), 20–22.

Hao, Y. (1999). From rule of man to rule of law: An unintended consequence of corruption in

China in the 1990s. Journal of Contemporary China, 8(22), 405–423.

Harcha, J. Z., & Pereira, P. C. (2014). Revolutions of Resolution. tripleC: Communication,

Capitalism & Critique. Open Access Journal for a Global Sustainable Information

Society, 12(1), 315–327.

Hardt, M. (1999). Affective Labor. boundary 2, 26(2), 89–100.

Hardt, M., & Negri, A. (2000). Empire. Harvard University Press.

239

He, K. (2016, Jan 19). Discover live-streamer. Tencent News.

http://news.qq.com/a/20160119/037064_1.htm

Hearn, A., & Schoenhoff, S. (2016). From celebrity to influencer. In P. D. Marshall, & S.

Redmond (Eds.), A companion to celebrity (pp. 194–212). John Wiley & Sons.

Helberger, N., Pierson, J., & Poell, T. (2018). Governing online platforms: From contested to

cooperative responsibility. The Information Society, 34(1), 1–14.

Help.Twitch. (2020a, April 11). How to manage harassment in chat. Twitch.

https://help.twitch.tv/s/article/how-to-manage-harassment-in-chat?language=en_US

Help.Twitch. (2020b, Feb 11). Joining the affiliate program. Twitch.

https://help.twitch.tv/s/article/joining-the-affiliate-program?language=en_US

Hilvert-Bruce, Z., Neill, J. T., Sjöblom, M., & Hamari, J. (2018). Social motivations of live-

streaming viewer engagement on Twitch. Computers in Human Behavior, 84, 58–67.

Ho, P. (2001). Who owns China’s land? Policies, property rights and deliberate institutional

ambiguity. The China Quarterly, 166, 394–421.

Ho, V., & Fung, A. (2016). Cultural policy, Chinese national identity and globalisation. In

T. Flew, & P. Iosifidis (Eds), Global media and national policies: The Return of the State

(pp. 106–121). Springer.

Hong, Y. (2014). Between corporate development and public service: The cultural system reform

in the Chinese media sector. Media, Culture & Society, 36(5), 610–627.

Hong, Y. (2017a). Networking China: The Digital Transformation of the Chinese Economy.

University of Illinois Press.

240

Hong, Y. (2017b). Pivot to internet plus: Molding China’s digital economy for economic

restructuring. International Journal of Communication, 11.

https://ijoc.org/index.php/ijoc/article/view/6105

Hong, Y., & Goodnight, G. T. (2020). How to think about cyber sovereignty: The case of China.

Chinese Journal of Communication, 13(1), 8–26.

Hong, Y., & Xu, J. (2019). Toward fragmented platform governance in China: Through the lens

of Alibaba and the legal-judicial system. International Journal of Communication,

13(21). https://ijoc.org/index.php/ijoc/article/view/12025

House of Lords Select Committee on Communications. (2019). Regulating in a digital world.

UK Parliament.

https://publications.parliament.uk/pa/ld201719/ldselect/ldcomuni/299/299.pdf

Howard, P. N. (2011). Castells and the media: Theory and media (Vol. 2). Polity Press.

Hu, P., & Yu, F. (2019). 网络主播与情感劳动: 一项探索性研究 (Live streamers and

affective labor: An exploratory study). Journalism and Communication Study, 2(3).

Huai, G. (2020, Oct 11). 淮安清江浦警方捣毁一淫秽直播平台 (Huaian Qingjiangpu police

smashed an obscene live broadcast platform). Xinhua Net.

http://www.js.xinhuanet.com/2020-10/11/c_1126592289.htm

Huaixiansen. (2017, Dec 15). 全民帝师反向石锤五五开？(Dishi launched smokinggun on

White55?). Sohu. https://www.sohu.com/a/208575625_461082

Huang, D. (2017). Social media and activism of grassroots NGOs in China. In M. Kent, K. Ellis,

& J. Xu (Eds.), Chinese social media: Social, cultural, and political implications (pp. 22–

40). Routledge.

241

Huang, P. C. (2017). Dispatch work in China: A study from case records, part I. Modern China,

43(3), 247–287.

Human Rights Watch. (2018, Feb 20). Germany: Flawed social media law.

https://www.hrw.org/news/2018/02/14/germany-flawed-social-media-law

Huya. (2018, Jun 5). Huya first quarter 2018 Unaudited financial results.

http://ir.huya.com/2018-06-05-Huya-Reports-First-Quarter-2018-Unaudited-Financial-

Results

Huya. (2019, March 4). Huya Inc fiscal year 2018 unaudited fnancial results.

http://ir.huya.com/2019-03-04-HUYA-Inc-Reports-Fourth-Quarter-and-Fiscal-Year-

2018-Unaudited-Financial-Results

Huya. (2020). Huya 2019 annual report. http://ir.huya.com/index.php?s=120 iiMedia. (2016). 2016 Chinese live sreaming report. http://www.iimedia.cn/42373.html iiMedia. (2020, Aug 14). 2020 上半年中国在线直播行业研究报告 (Research report on

China’s online live streaming industry in the first half of 2020). iiMedia.

https://www.iimedia.cn/c400/73538.html

Influencer. (2020, Dec 12). 25 useful Twitch statistics for influencer marketing managers.

https://influencermarketinghub.com/twitch-statistics/#toc-1

Institute, P. I. R. (2019). 中国网络直播行业发展现状分析 (Analysis of the status of China's

live streaming industry). https://bg.qianzhan.com/report/detail/300/191029-

db68392a.html

Jenkins, H., & Deuze, M. (2008). Convergence culture. Sage.

242

Jia, L., & Zhou, T. (2017). The ‘making’of an online celebrity. In M. Kent, K. Ellis, & J. Xu

(Eds.), Chinese social media: Social, cultural, and political implications (pp. 42–58).

Routledge.

Jiang, M. (2014). The co-evolution of the internet, (un)civil society & authoritarianism in China.

In J. de Lisle, A. Goldstein, & G. Yang (Eds.), The internet, social media, and a

changing China (pp. 28–48). University of Pennsylvania Press.

Jimroglou, K. M. (1999). A camera with a view: JenniCAM, visual representation, and cyborg

subjectivity. Information, Communication & Society, 2(4), 439–453.

Jing, M. (2019, Apr 8). No smoking, no tattoos, no bikinis: Inside China’s war to ‘clean up’ the

internet. SCMP. https://www.scmp.com/tech/policy/article/3005252/no-smoking-no-

tattoos-or-bikinis-inside-chinas-war-clean-internet

Jingyan, Z. (2016, Nov 12). The secret of supervising live-streaming. The Paper.

http://www.thepaper.cn/newsDetail_forward_1428933

Johnson, M. R., & Woodcock, J. (2019a). ‘And today’s top donator is’: How live streamers on

Twitch.tv monetize and gamify their broadcasts. Social Media + Society, 5(4).

https://doi.org/10.1177/2056305119881694

Johnson, M. R., & Woodcock, J. (2019b). ‘It’s like the gold rush’: The lives and careers of

professional video game streamers on Twitch.tv. Information, Communication & Society,

22(3), 336–351.

Kaytoue, M., Silva, A., Cerf, L., Meira Jr, W., & Raïssi, C. (2012). Watch me playing, I am a

professional: A first study on video game live streaming. Proceedings of the 21st

International Conference on World Wide Web, Lyon.

243

Keane, M. (2016). Internet+ China: Unleashing the innovative nation strategy. International

Journal of Cultural and Ceative Industries, 3(2), 68–74.

Keane, M., & Chen, Y. (2019). Entrepreneurial solutionism, characteristic cultural industries and

the Chinese dream. International Journal of Cultural Policy, 25(6), 743–755.

Khalid, A. (2019, Nov 13). The future of live-streaming, for better or worse, depends on Twitch.

Quartz. https://qz.com/1747158/twitch-grows-as-non-gamer-live-streaming-expands-on-

the-platform

Khamis, S., Ang, L., & Welling, R. (2017). Self-branding,‘micro-celebrity’and the rise of social

media influencers. Celebrity Studies, 8(2), 191–208.

Khan, I. (2019, Dec. 15). Why Twitch is still the king of live game streaming. New York Times.

https://www.nytimes.com/2019/12/15/business/tech-video-game-streaming-twitch.html

Khudyakov, A. A. (2017). The iron hand in a velvet glove: Domestic soft power and the struggle

for cultural influence in China [Master’s thesis, University of Oslo].

Kim, J. (2012). The institutionalization of YouTube: From user-generated content to

professionally generated content. Media, Culture & Society, 34(1), 53–67.

King, G., Pan, J., & Roberts, M. E. (2013). How censorship in China allows government

criticism but silences collective expression. American Political Science Review, 107(2),

326–343.

King, G., Pan, J., & Roberts, M. E. (2014). Reverse-engineering censorship in China:

Randomized experimentation and participant observation. Science, 345(6199), 1251722.

Knight, J., Shi, L., & Song, L. (2006). The rural–urban divide and the evolution of political

economy in China. In L. Garbo (Ed.), Human development in the era of globalization:

Essays in honor of Keith B. Griffin (pp. 44–63). Edward Elgar.

244

Kostka, G. (2019). China’s social credit systems and public opinion: Explaining high levels of

approval. New Media & Society, 21(7), 1565–1593.

Krämer, N. C., Winter, S., Benninghoff, B., & Gallus, C. (2015). How “social” is Social TV?

The influence of social motives and expected outcomes on the usage of Social TV

applications. Computers in Human Behavior, 51, 255–262.

Kücklich, J. (2005). Precarious playbour: Modders and the digital games industry. fibreculture,

5(1), 1–5.

Lai, C. (2017, May 26). Chinese live-streaming platform punished for broadcasting fake

‘Forbidden City’ video. HKFP. https://www.hongkongfp.com/2017/05/26/chinese-live-

streaming-platform-punished-broadcasting-fake-forbidden-city-video

Lan, R. (2019). 众包直播系统中的用户行为分析及应用 (User behavior analysis and

application in crowdsourced live streaming system). University of Science and

Technology of China.

Langlois, G. (2013). Participatory culture and the new governance of communication: The

paradox of participatory media. Television & New Media, 14(2), 91–105.

Langlois, G., & Elmer, G. (2013). The research politics of social media platforms. Culture

machine, 14. http://svr91.edns1.com/~culturem/index.php/cm/article/viewArticle/505

Laotou, Y. (2019, Sep 21). 真棒！斗鱼推出青少年保护模式，家长终于可以放心让孩子看

直播了 (Awesome! Douyu launched a youth protection model, and parents can finally

let their children watch the live stream safely). Sina.

http://k.sina.com.cn/article_6591108808_188dc56c800100ir7c.html?cre=tianyi&mod=pc

pager_fintoutiao&loc=4&r=9&rfunc=100&tj=none&tr=9

245

Laoyu. (2019). 细数网络直播行业的这三年 (Pondering the live streaming in these three

years). Huxiu. https://www.huxiu.com/article/283804.html

Latham, K. (2000). Nothing but the truth: News media, power and hegemony in South China.

The China Quarterly, 163, 633–654.

Lazzarato, M., & Virno, P. (1996). Radical thought in Italy: A potential politics. University of

Minnesota Press.

Li, D. (2020, Oct 30). 直播行业大变局 百度巨额收购 YY 国内业务 BAT 欲三分天下？

(Big changes in the live streaming industry, Baidu acquires YY’s domestic business with

a huge amount. BAT wants to share the world?). Sohu. Retrieved Nov 22 from

https://www.sohu.com/a/428390261_153054

Li, G. C. (2020). Distinction in China: The rise of taste in cultural consumption. London School

of Economics and Political Science.

Li, L. (2019). Zoning China: Online video, popular culture, and the state. MIT Press.

Li, Q. (2020, May 29). 600m with $140 monthly income worries top. Global Times.

https://www.globaltimes.cn/content/1189968.shtml

Li, Z. (2020, June 20). 领导干部走进 618 电商直播间 带货还需带人带产业 (Leaders enter

the 618 e-commerce live broadcast room, e-commerce also need to bring people and

bring industry). Xinhua Net. http://www.xinhuanet.com/fortune/2020-

06/20/c_1126137474.htm

Liang, B., & Lu, H. (2010). Internet development, censorship, and cyber crimes in China.

Journal of Contemporary Criminal Justice, 26(1), 103–120.

Lifang. (2018, Jan 20). Live streaming revenue exceeds 30 bln yuan in China in 2017. Xinhua

Net. http://www.xinhuanet.com/english/2018-01/20/c_136911336.htm

246

Lin, J., & de Kloet, J. (2019). Platformization of the unlikely creative class: Kuaishou and

Chinese digital cultural production. Social Media+ Society, 5(4).

https://journals.sagepub.com/doi/full/10.1177/2056305119883430

Liu, H. (2020). 放手“冯提莫们”的斗鱼，才是真正的成熟了 (Let go of the "Feng Timo"

makes Douyu mature). Chinanews.

https://m.chinanews.com/wap/detail/zw/business/2020/01-03/9050500.shtml

Liu, R., & Xu, J. (2018). Co-governance of live online video: Based on the perspective of policy

network analysis. Chinese Journal of Journalism & Communication, 40(12), 32–49.

Liu, s. (2012). 《谁认识马云 2：危机中的救赎》(Who knows Jack Ma 2: Redemption in a

crisis). China Friendship Publishing Company.

Liu, X. (2016). Live streaming in China: Boom market, business model and risk regulation.

Journal of Residuals Science & Technology, 13(8), 1–7.

Liu, Y. (2020, Dec 14). China’s antitrust regulator fines affiliates of Alibaba, Tencent and SF

Holding for ‘monopolistic behaviour’. SCMP. https://www.scmp.com/business/china-

business/article/3113860/chinas-antitrust-regulator-fines-affiliates-alibaba-tencent

Live, F. (2020). How to moderate comments. Facebook.

https://www.facebook.com/facebookmedia/solutions/facebook-live

Lobato, R. (2016). The cultural logic of digital intermediaries: YouTube multichannel networks.

Convergence, 22(4), 348-360.

Lochrie, M., & Coulton, P. (2012). Sharing the viewing experience through second screens.

Proceedings of the 10th European Conference on Interactive TV and Video, Berlin.

247

Lomas, N. (2017). Facebook’s content moderation rules dubbed ‘alarming’ by Child Safety

Charity. TechCrunch. https://techcrunch.com/2017/05/22/facebooks-content-moderation-

rules-dubbed-alarming-by-child-safety-charity

Lqbal, M. (2019, Feb 27). Twitch revenue and usage statistics. Business of Apps.

https://www.businessofapps.com/data/twitch-statistics

Lu, J., & Weber, I. (2007). State, power and mobile communication: A case study of China. New

Media & Society, 9(6), 925–944.

Lu, R. (2014, Aug 22). China’s ‘hormone economy’: Monetizing male loneliness. Foreign

Policy. http://foreignpolicy.com/2014/08/22/chinas-hormone-economy-monetizing-male-

loneliness

Lu, Y. (2016, May 9). Live streaming platform ‘cracking down on pornography’ war. China

Youth Daily. http://zqb.cyol.com/html/2016-05/09/nw.D110000zgqnb_20160509_1-

06.htm

Lu, Z. (2015). 流行歌曲与音乐文化的符号域 (Popular songs and the symbolic domain of

music culture). Guizhou Social Sciences, 308(8).

Lu, Z. (2019). Live streaming in China for sharing knowledge and promoting intangible cultural

heritage. interactions, 27(1), 58–63.

Lu, Z., Xia, H., Heo, S., & Wigdor, D. (2018). You watch, you give, and you engage: A study of

live streaming practices in China. Proceedings of the 2018 CHI Conference on Human

Factors in Computing Systems, Montreal, Canada.

Lynley, M. (2011, Mar 10). Live-streaming site Justin.tv buffing up for e-sports channels.

VentureBeat. https://venturebeat.com/2011/03/10/justin-tv-esports-channel

248

Ma, W. (2020, Oct 12). How China's digital farmers can grow a post-COVID-19 future.

WEforum. https://www.weforum.org/agenda/2020/10/how-digital-farmers-in-china-can-

grow-a-post-covid-19-future

Mai, Y., Meng, L., & Ye, Z. (2017). Regional variation in the capital structure adjustment speed

of listed firms: Evidence from China. Economic Modelling, 64, 288–294.

Martini, M. (2018a). Online distant witnessing and live-streaming activism: Emerging

differences in the activation of networked publics. New Media & Society, 20(11), 4035–

4055.

Martini, M. (2018b). Online distant witnessing and live-streaming activism: Emerging

differences in the activation of networked publics. New Media & Society, 20(11), 4035–

4055.

Marwick, A. E. (2013). Status update: Celebrity, publicity, and branding in the social media

age. Yale University Press.

Matsui, S. (2016). Does it have to be a copyright infringement? Live game streaming and

copyright. Texas Intellectual Property Law Journal 24, 215.

McGlynn, A. (2020, Dec 12). Cyberpunk 2077 is a hit in China, despite not having an official

release there. PC Games China. https://www.pcgamesn.com/cyberpunk-2077/sales-china

MCT. (2009, Jun 26). 文化部、商务部关于加强网络游戏虚拟货币管理工作的通知 (Notice

of the Ministry of Culture and the Ministry of Commerce on strengthening the

management of online game virtual currency). Ministry of Culture.

https://www.mct.gov.cn/whzx/bnsj/whscs/201111/t20111128_751250.htm

249

MCT. (2016a). Ministry of Culture to rectify illegal network broadcast platform (文化部整治违

规网络直播平台). Ministry of Culture. http://www.gov.cn/xinwen/2016-

04/14/content_5064016.htm

MCT. (2016b). Notice of the Ministry of Culture on printing and fistributing the Measures for

the Administration of the Operation of Network Performances. Ministry of Culture.

http://www.mcprc.gov.cn/whzx/bnsjdt/whscs/201612/t20161212_464766.html

Ministry of Education. (2018, Dec 28). 教育部办公厅关于严禁有害 APP 进入中小学校园的

通知 (Notice of the General Office of the Ministry of Education on Prohibiting Harmful

Apps from Entering Primary and Secondary School Campuses).

http://www.moe.gov.cn/srcsite/A06/s3321/201901/t20190102_365728.html

Moiz, A. (2017, April 19). Livestreaming Q1'17: Twitch vs YouTube, avg. spend, LTV and tip

volumes. Streamlabs. https://blog.streamlabs.com/livestreaming-q117-twitch-vs-youtube-

avg-spend-ltv-and-tip-volumes-8263c6a9f99e

Mosco, V. (2009). The political economy of communication (2nd ed.). Sage.

Napoli, P. M. (2019). Social media and the public interest: Media regulation in the

disinformation age. Columbia University Press.

Nascimento, G., Ribeiro, M., Cerf, L., Cesário, N., Kaytoue, M., Raïssi, C., Vasconcelos, T., &

Meira, W. (2014). Modeling and analyzing the video game live-streaming community.

Web Congress (LA-WEB), 2014.

NDTV. (2014, Aug 6). Twitch’s birthplace: Justin.tv shuts down. NDTV.

https://gadgets.ndtv.com/internet/news/twitchs-birthplace-justintv-shuts-down-571534

Needed, M. (2018, Dec 8). Twitch by the numbers: Stats, users, demographics & fun facts.

Muchneeded. https://muchneeded.com/twitch-statistics/

250

Nieborg, D. B., & Poell, T. (2018). The platformization of cultural production: Theorizing the

contingent cultural commodity. New Media & Society, 20(11), 4275–4292.

Nielsen. (2013, Jun 17). Action figures: How second screens are transforming TV viewing.

Nielsen. https://www.nielsen.com/ae/en/insights/article/2013/action-figures-how-second-

screens-are-transforming-tv-viewing

Norman, D. A. (1999). Affordance, conventions, and design. interactions, 6(3), 38–43.

Pan, J. (2018). 网络直播用工关系的劳动法思考 (Thinking on the labor law of live streaming

employment relations). Journal of China Institute of Labor Relations, 32(4), 70–77.

Panneton, C. (2019). Appropriating play: Examining Twitch.tv as a commercial platform

[Doctoral dissertation, University of Western Ontario].

The Paper. (2020). 科技向善，打赏向“恶”：斗鱼财报没有说出的秘密 (Technology for

good, tip for ‘evil’: The secret of Douyu’s financial report).

https://www.thepaper.cn/newsDetail_forward_7672406

Partin, W. C. (2020). Bit by (Twitch) bit: ‘Platform capture’ and the evolution of digital

platforms. Social Media + Society, 6(3),

https://journals.sagepub.com/doi/full/10.1177/2056305120933981

Pellicone, A. (2016). Performing play: Cultural production on twitch.tv. Proceedings of the 2016

CHI Conference Extended Abstracts on Human Factors in Computing Systems, New

York.

People.cn. (2020, Jun 5). Eight departments work togather to carry out special operations for the

live streaming industry. http://yuqing.people.com.cn/n1/2020/0605/c209043-

31737011.html

251

Phillips, A. (2017). Welcome to my fantasy zone: Bayonetta and queer femme disturbance. In

A. Shaw, & B. Ruberg (Eds.), Queer Game Studies (pp. 109–124). University of

Minnesota Press.

Pires, K., & Simon, G. (2014). Dash in twitch: Adaptive bitrate streaming in live game streaming

platforms. Proceedings of the 2014 Workshop on Design, Quality and Deployment of

Adaptive Video Streaming, New York.

Plantin, J.-C., Lagoze, C., Edwards, P. N., & Sandvig, C. (2018). Infrastructure studies meet

platform studies in the age of Google and Facebook. New Media & Society, 20(1), 293–

310.

Poell, T. (2020). Three challenges for media studies in the age of platforms. Television & New

Media, 21(6), 650–657.

Poell, T., Nieborg, D., & van Dijck, J. (2019). Concepts of the digital society: Platformisation.

Internet Policy Review, 8(4). https://doi.org/10.14763/2019.4.1425

Popper, B. (2014, Aug 8). Justin.tv, the live video pioneer that birthed Twitch, officially shuts

down. The Verge. https://www.theverge.com/2014/8/5/5971939/justin-tv-the-live-video-

pioneer-that-birthed-twitch-officially-shuts

Post, Y. E. (2018, Jan 24). 闹翻后，网络主播告公司索要 41 万 (After the fight, the streamers

asked the company for 410,000). Sina.

http://k.sina.com.cn/article_1653603955_628ffe73034002i4n.html?sudaref=www.google

.com&display=0&retcode=0

Powazek, D. M. (2002). Design for community: The art of connecting real people in virtual

places. New Riders.

252

Price, M. E. (2002). Media and sovereignty: The global information revolution and its challenge

to state power. MIT Press.

PUBGAssistance. (2017, Nov 27). 电竞圈超级主播签约费最高达三亿元 (Live Streaming

Signning Fee is up to 300 Million RMB). Sina.

http://k.sina.com.cn/article_3793766510_e220486e001002qzh.html

Qiu, J. L. (1999). Virtual censorship in China: Keeping the gate between the cyberspaces.

International Journal of Communications Law and Policy, 4(1), 25.

Qiu, J. L. (2009). Working-class network society: Communication technology and the

information have-less in urban China. MIT Press.

Qiu, J. L. (2018). China’s digital working class and circuits of labor. Communication and the

Public, 3(1), 5–18.

Recktenwald, D. (2016). ‘Donation alerts’ on Twitch.TV: Commodification of community and

attention. The IMPEC 2016 Conference, Lyon.

https://www.researchgate.net/publication/333260592_Donation_Alerts_on_Twitch

Recktenwald, D. (2017). Toward a transcription and analysis of live streaming on Twitch.

Journal of Pragmatics, 115, 68–81.

Recktenwald, D., & Du, Y. (2016). Lagging behind Twitch or on its own path: Pressures and

perks on domestic online live streaming in China. Proceedings of 3rd Annual Chinese

DiGRA Conference, Taiwan.

Reeve, A., & Reeve, M. (2011). Punch and Judy at the beach and in the mall. Visual Culture in

Britain, 12(1), 17–31.

Rosen, G. (2019, March 14). Protecting Facebook Live from abuse and investing in manipulated

media research. Facebook. https://about.fb.com/news/2019/05/protecting-live-from-abuse

253

Ruberg, B. (2020). ‘Obscene, pornographic, or otherwise objectionable’: Biased definitions of

sexual content in video game live streaming. New Media & Society,

https://journals.sagepub.com/doi/abs/10.1177/1461444820920759.

Ruberg, B., Cullen, A. L., & Brewster, K. (2019). Nothing but a “titty streamer”: Legitimacy,

labor, and the debate over women’s breasts in video game live streaming. Critical Studies

in Media Communication, 36(5), 466–481.

SAMR. (2020a, Nov 10). Anti-monopoly guidelines on the platform economy field.

http://www.samr.gov.cn/hd/zjdc/202011/P020201109757564420041.docx

SAMR. (2020b, Nov 10). 市场监管总局关于《关于平台经济领域的反垄断指南（征求意见

稿）》公开征求意见的公告 (SAMR’s announcement for public suggestions on anti-

monopoly guidelines on the platform economy field).

http://www.samr.gov.cn/hd/zjdc/202011/t20201109_323234.html

SAPPRFT. (2007, Dec 29). 互联网视听节目服务管理规定 (Provisions on the administration

of internet audio-visual program service).

http://www.lawinfochina.com/Display.aspx?lib=law&Cgid=269280

SAPPRFT. (2016, Dec 12). General administration issued ‘on strengthening the network audio-

visual program live service management issues related to notice’.

http://www.chinasarft.gov.cn/art/2016/9/19/art_113_31668.html

SAPPRFT. (2017a, Jun 22). 国家新闻出版广电总局要求“新浪微博”、“ACFUN”等网站关停

视听节目服务 (SARFT Require Weibo and ACFUN to Stop Audio-visual Services).

http://www.sapprft.gov.cn/sapprft/govpublic/6702/348195.shtml

254

SAPPRFT. (2017b, Jun 1). 总局进一步加强网络视听节目创作播出管理 (SARFT further

strengthens the management of network audiovisual program creation and broadcasting).

http://www.sapprft.gov.cn/sapprft/contents/6582/336049.shtml

SAPPRFT. (2018a, Dec 12). 关于进一步规范网络视听节目传播秩序的通知 (Notice on

further regulating the transmission order of internet audio-visual programs).

http://www.sapprft.gov.cn/sapprft/govpublic/6684/1798.shtml

SAPPRFT. (2018b, Apr 4). 国家广播电视总局严肃处理“今日头条”“快手”传播有违社会道德

节目等问题 (SARFT punish Toutiao and Kuaishou for disseminating programme that

offends social morals). http://www.sapprft.gov.cn/sapprft/contents/6582/363639.shtml

Schiller, H. (1989). Disney, Dallas and electronic data flows: The transnationalization of culture.

1989). Cultural transfer or electronic imperialism: The impact of American television

programs on European television (pp. 33–43). C. Winter.

SCMP. (2019, Jan 23). Why live streaming is becoming China’s most-profitable form of

electronic media. https://www.youtube.com/watch?v=bXbatlWx-

tY&ab_channel=SouthChinaMorningPost

Scott, C. (2017). The regulatory state and beyond. In P. Drahos (Ed.), Regulatory theory:

Foundations and application (p. 265). ANU Press.

Senft, T. M. (2013). Microcelebrity and the branded self: A companion to new media dynamics.

Blackwell.

Shamma, D. A., Kennedy, L., & Churchill, E. F. (2012). Watching and talking: Media content as

social nexus. Proceedings of the 2nd ACM International Conference on Multimedia

Retrieval, Hong Kong.

255

Shamma, D. A., Shaw, R., Shafton, P. L., & Liu, Y. (2007). Watch what I watch: Using

community activity to understand content. Proceedings of ACMMIR, Augsburg Bavaria.

Shanghai Bureau of Statistics. (2020, Jun 30). 2019 年上海市城镇单位就业人员平均工资

(Average salary of employee in urban units in Shanghai in 2019). Tjj.

http://tjj.sh.gov.cn/tjxw/20200629/fc54a4124e70400098921b6fc5680e1e.html

Shaw, A., Drogin Rodgers, E. B., Rodgers, R. P., & Wiggins, B. (2014). E-Sport spectator

motivation. In R. Rogers (Ed.), Understanding Esports: An Introduction to the Global

Phenomenon (pp. 73–84). Lexington Books.

She, X. (2020, Aug 10). 斗鱼虎牙合并消息尘埃落定 大股东腾讯发初步建议书 (The

merger of Douyu & Huya has been confirmed, major shareholder Tencent issues

preliminary proposal). Sina. https://finance.sina.com.cn/chanjing/gsnews/2020-08-

10/doc-iivhuipn7911106.shtml

Shen, H. (2019, Aug 11). A brief analysis of major legal issues in the overseas listing of pan-

entertainment platforms. Lexology.

https://www.lexology.com/library/detail.aspx?g=f335ff38-8a3e-4881-adee-f13017e51dfe

Shen, J., & Xiao, Y. (2020). Emerging divided cities in China: Socioeconomic segregation in

Shanghai, 2000–2010. Urban Studies, 57(6), 1338–1356.

Shen, M. (2020, Sep 17). 农产品直播带货利好农户 走出产地流向大市场 (Live streaming of

agricultural products is beneficial for farmers, letting them going out of the place of

production to a bigger market). Xinhua Net. http://www.xinhuanet.com/food/2020-

09/17/c_1126505587.htm

256

Shen, X. (2020a, May 13). Live-streaming sellers and blockchain engineers are officially jobs in

China now. SCMP. https://www.scmp.com/abacus/tech/article/3084264/live-streaming-

sellers-and-blockchain-engineers-are-officially-jobs

Shen, X. (2020b Apr 3). Shopping in live streaming apps is booming in China thanks to the

COVID-19 pandemic. SCMP.

https://www.scmp.com/abacus/tech/article/3078208/shopping-live-streaming-apps-

booming-china-thanks-covid-19-pandemic

Shi, C. (2018, Feb 14). 焦点访谈，揭开天佑被封杀真因 (Focus report, revealing the true

cause of god blessing). Sohu. http://www.sohu.com/a/222688224_313745

Shu, T. (2004). 当网络游戏被荧屏 “删除” (When online games are ‘deleted’ by the screen).

Computer and Internet, 10, 1.

Sina. (2017, Jan 26). 文化部严管直播平台 严整“三俗”内容 (The Ministry of Culture strictly

controls the live broadcast platform and strictly rectifies the content of the ‘three

secular’). https://ent.sina.cn/tv/tv/2017-01-26/detail-ifxzyxmu8033467.d.html?from=wap

Sina. (2018, Jul 27). 哔哩哔哩动画客户端下架 B 站发布公告称将加强审核严格整改

(Bilibili Client is removed, it issued an announcement saying that it will strengthen the

moderation). Sina. https://tech.sina.com.cn/roll/2018-07-27/doc-ihfvkitx4498247.shtml

Singh, P. (2019, Feb 19). Bigo to expand in India, Middle East and the US. Entrepreneur.

https://www.entrepreneur.com/article/328511

Sjöblom, M., Törhönen, M., Hamari, J., & Macey, J. (2019). The ingredients of Twitch

streaming: Affordances of game streams. Computers in Human Behavior, 92, 20–28.

257

Smith, T., Obrist, M., & Wright, P. (2013). Live-streaming changes the (video) game.

Proceedings of the 11th European Conference on Interactive TV and Video, Como, Italy.

Smythe, D. W. (1977). Communications: Blindspot of western Marxism. CTheory, 1(3), 1–27.

Soo, Z., & Ho, C. K. (2019, Apr 19). Inside China tech: How live-streaming in China is

monitored and censored. SCMP. https://www.scmp.com/podcasts/article/3005878/inside-

china-tech-how-live-streaming-china-monitored-and-censored

Spilker, H. S., Ask, K., & Hansen, M. (2020). The new practices and infrastructures of

participation: How the popularity of Twitch.tv challenges old and new ideas about

television viewing. Information, Communication & Society, 23(4), 605–620.

Spilker, H. S., & Colbjørnsen, T. (2020). The dimensions of streaming: Toward a typology of an

evolving concept. Media, Culture & Society, 42(7–8), 1210–1225.

Srnicek, N. (2017). Platform capitalism. John Wiley & Sons.

State Council Xinhua (2018). China to merge ministry of culture, tourism administration.

http://english.www.gov.cn/state_council/ministries/2018/03/13/content_28147607650988

8.htm

Su, C., & Flew, T. (2020). The rise of Baidu, Alibaba and Tencent (BAT) and their role in

China’s Belt and Road Initiative (BRI). Global Media and Communication,

https://journals.sagepub.com/doi/abs/10.1177/1742766520982324

Summa, G. (2011). Social TV: The future of television in the internet Age [Master’s thesis,

Massachusetts Institute of Technology].

Sun, P., Zhao, G., Liu, Z., Li, X., & Zhao, Y. (2020). Toward discourse involution within

China’s internet: Class, voice, and social media. New Media & Society.

https://doi.org/10.1177/1461444820966753

258

Tang, M. (2020). From “bringing-in” to “going-out”: Transnationalizing China’s internet capital

through state policies. Chinese Journal of Communication, 13(1), 27–46.

Tao, Z. (2018, Feb 17). 法院叫停西瓜视频，首个游戏直播禁令 (The court suspends

Watermelon Video, the first game live streaming ban). Netease.

https://www.163.com/dy/article/E880S8D205373Y5B.html

Tapscott, D., & Williams, A. D. (2008). Wikinomics: How mass collaboration changes

everything. Penguin.

Taylor, N. T. (2016). Now you’re playing with audience power: The work of watching games.

Critical Studies in Media Communication, 33(4), 293–307.

TechSina. (2017, Mar 18). 还记得那张直播千播大战图吗？已经有六分之一平台无法使用

(Remember the picture of The War Between the Thousands? One sixth of the platforms

are already unavailable). Sina. https://tech.sina.cn/i/gn/2017-03-08/detail-

ifycaafm5491458.d.html?from=wap

Terranova, T. (2000). Free labor: Producing culture for the digital economy. Social text, 18(2),

33–58.

Toffler, A., & Alvin, T. (1980). The third wave. Bantam Books.

Tomasic, R., & Fu, J. (2005). Regulation and corporate governance of China’s top 100 listed

companies: Whither the rule of law? Research Committee on the Sociology of Law (ISA

RC12), Annual Conference, Paris.

The Truman Show. (2020). IMDb. https://www.imdb.com/title/tt0120382

Twitch. (2019, Nov 19). Twitch terms of service. https://www.twitch.tv/p/legal/terms-of-service

259

Twitch. (2020a, April 14). About account enforcements and chat bans.

https://help.twitch.tv/s/article/about-account-suspensions-dmca-suspensions-and-chat-

bans?language=en_US

Twitch. (2020b, Dec 12). How to subscribe. https://help.twitch.tv/s/article/how-to-

subscribe?language=en_US

Twitch. (2020c, Dec 20). Twitch advertising. https://twitchadvertising.tv/

Twitch. (2020d, Apr 7). Twitch community guidelines.

https://www.twitch.tv/p/legal/community-guidelines/

UUU9. (2018, Mar 02). 《Dota》主播 PIS 被要求转型《绝地求生》否则解约 ("Dota"

Streamer PIS is required to transit to "PUBG" or terminate the contract).

http://news.uuu9.com/yjzu/201803/424101.shtml van Brandenburg, R., Niamut, O., Veenhuizen, A., & Hoekman, G.-J. (2016). Towards new

forms of news gathering through crowdsourced live mobile streaming systems IBC 2016

Conference, Amsterdam. http://digital-

library.theiet.org/content/conferences/10.1049/ibc.2016.0032

Van Dijck, J. (2020). Seeing the forest for the trees: Visualizing platformization and its

governance. New Media & Society,

https://journals.sagepub.com/doi/full/10.1177/1461444820940293 van Dijck, J., Nieborg, D., & Poell, T. (2019). Reframing platform power. Internet Policy

Review, 8(2), 1–18.

Van Dijck, J., & Poell, T. (2013). Understanding social media logic. Media and Communication,

1(1), 2–14.

260

Van Dijck, J., Poell, T., & De Waal, M. (2018). The platform society: Public values in a

connective world. Oxford University Press. van Ditmarsch, J. (2013). Video games as a spectator sport. [Master’s thesis, Utrecht

University]. https://dspace.library.uu.nl/handle/1874/288162

Walker, A. (2014). Watching us play: Postures and platforms of live streaming. Surveillance &

Society, 12(3), 437–442.

Wang, H., & Linyuan, P. (2017). 网络直播监管机制及路径研究 (Study on live streaming

regulation mechanism and path). 出版广角 (View on Publishing), 6, 78–82.

Wang, J. (2017). “Stir-frying” internet finance: Financialization and the institutional role of

financial news in China. International Journal of Communication, 11, 22.

Wang, Q. (2018). Media differentiation between Chinese state media and commercial online

media: A multi-stakeholder media system in the digital China. International

Communication Research Journal, 53(1), 2–25.

Wang, S. (2020). Chinese affective platform economies: Dating, live streaming, and

performative labor on Blued. Media, Culture & Society, 42(4), 502–520.

Wang, S., & Chen, S. (2019, Apr 21). 试论“三俗”内容的法律规制 (Laws and regulations for

content that commits the ‘three vulgarity’). CAC. http://www.cac.gov.cn/2020-

04/21/c_1589016056807150.htm

Wang, W. Y., & Lobato, R. (2019). Chinese video streaming services in the context of global

platform studies. Chinese Journal of Communication, 12(3), 356–371.

Wang, Z., & Lan, J. (2021, Jan 21). 沉迷斗鱼直播“鱼丸预言”互动 男子输掉 140 万购房款

(Addicted to Douyu ‘Fish Ball Prediction’ interactive game, a man loses 1.4 million

261

property funding). Sina. https://finance.sina.com.cn/tech/2021-01-21/doc-

ikftssan9019479.shtml

Wangnitianxinweijixin. (2018, Jun 11). A discussion on Discipline Committee. Bilibili.

https://www.bilibili.com/read/cv575799

Weber, I. (2005). Digitizing the dragon: Challenges facing China's broadcasting industry. New

Media & Society, 7(6), 791–809.

Weber, I., & Jia, L. (2007). Internet and self-regulation in China: The cultural logic of controlled

commodification. Media, Culture & Society, 29(5), 772–789.

Wei, Z. (2017, July 10). Ministry of Culture remediation network ‘three customs’ shut down

‘Wukong TV’ and other 12 mobile phone performance platform. Xinhua Net.

http://news.xinhuanet.com/politics/2017-06/29/c_1121235571.htm

Weisz, J. D., Kiesler, S., Zhang, H., Ren, Y., Kraut, R. E., & Konstan, J. A. (2007). Watching

together: Integrating text chat with video. Proceedings of the SIGCHI conference on

Human factors in computing systems, Montréal, Canada.

Wellman, B. (2001). Physical place and cyberplace: The rise of personalized networking.

International Journal of Urban and Regional Research, 25(2), 227–252.

Winseck, D. (2017). Reconstructing the political economy of communication for the digital

media age. The Political Economy of Communication, 4(2).

https://polecom.org/index.php/polecom/article/view/72

Winseck, D. (2020). Vampire squids,‘the broken internet’and platform regulation. Journal of

Digital Media & Policy, 11(3), 241–282.

262

Witkowski, E. (2011). Following Ms_Fabulous: Women, live-streaming, and do-it-yourself

visibility in e-sports. Annual meeting for the Digital Games Research Association,

Hilversum.

Witkowski, E., Recktenwald, D., Manning, J., & Ge, Z. (2016). Livestreaming in theory and

practice: Four provocations on labour, liveness and participatory culture in games

livestreaming. Proceedings of the 22nd international Symposium on Electronic Art, Hong

Kong.

Woodcock, J., & Johnson, M. R. (2019a). The affective labor and performance of live streaming

on Twitch.tv. Television & New Media, 20(8), 813–823.

Woodcock, J., & Johnson, M. R. (2019b). Live streamers on Twitch.tv as social media

influencers: Chances and challenges for strategic communication. International Journal

of Strategic Communication, 13(4), 321–335.

Worldbank. (2019, Dec 6). GDP per capita (current US$) - China, United States, World.

https://data.worldbank.org/indicator/NY.GDP.PCAP.CD?locations=CN-US-1W

Wright, J. (2014). Regional variation in Chinese internet filtering. Information, Communication

& Society, 17(1), 121–141.

Wu, C. (2017). 直播行业将建“黑名单”制度 (Live streaming industry establishes ‘blacklist’

system). Qiangjiang Evening News. http://qjwb.zjol.com.cn/html/2017-

07/09/content_3543644.htm?div=0

Wu, H. X. (2004). Daoist wisdom and popular wisdom: A sociolinguistic analysis of the

philosophical maxims in the Daode Jing and their proverbial equivalents. In V. Shen, &

W. Oxtoby (Eds.), Wisdom in China and the West. The Council for Research in Values

and Philosophy.

263

Xia, H. (2020, May 3). China’s live-streaming service users reach 560 mln: Report. Xinhua.

http://www.xinhuanet.com/english/2020-05/03/c_139027174.htm

Xie, Q., Zhou, Yijin. (2016, Aug 12). History, present and future live video streaming. Nfmedia.

http://www.nfmedia.com/cmzj/cmyj/jdzt/201608/t20160812_369897.htm

Xinhua. (2018a, Mar 13). China to merge Ministry of Culture, Tourism administration. Retrieved

http://english.www.gov.cn/state_council/ministries/2018/03/13/content_28147607650988

8.htm

Xinhua. (2018b, Aug 29). 网络直播平台高价挖人乱象：互挖墙脚成竞争常态 (Online live

streaming platform digs people at high prices: Mutual digging becomes the norm of

competition). Xinhua Net. http://www.xinhuanet.com/zgjx/2018-08/29/c_137426786.htm

Xinhua. (2019, Jan 8). 《2018 主播职业报告》发布 21%职业主播月收入过万 (‘2018

Streamer Profession Report’ released 21% of professional streamers have monthly

income of over 10,000). http://www.xinhuanet.com/info/2019-01/08/c_137725701.htm

Xinhua. (2020, Jun 11). 网课变网游：是时候为网游营销划界了 (Online lessons become

online games: it's time to draw a boundary to online game marketing). Xinhua Net.

http://www.xinhuanet.com/2020-06/11/c_1126099557.htm

Xu, J., & Zhang, G. (2020). The rise and fall of the ‘King of Hanmai’ – MC Tianyou. Celebrity

Studies, 1–6. https://doi.org/10.1080/19392397.2020.1783747

Xu, L., & Zhang, H. (2019). “人气游戏”: 网络直播行业的薪资制度与劳动控制 (‘Popular

game’: Salary system and labor control in the live streaming industry). Chinese Journal

of Sociology, 39(4), 61–83.

Xue, X. (2014). The research on communication science of YY voice communication platform

[Master’s Thesis, Dalian University of Technology].

264

Yan, Y. (2014, Dec 15). WeChat: Has not yet limited the individual accounts in the political

affairs and would not close the accounts on a large scale. (微信：暂未对时政类个人账

号进行限制 不会大规模封号). ifeng.

http://nb.ifeng.com/kejishuma/detail_2014_08/09/2735541_0.shtml

Yang, G. (2009). The power of the internet in China: Citizen activism online. Columbia

University Press.

Yang, G. (2014). The return of ideology and the future of Chinese internet policy. Critical

Studies in Media Communication, 31(2), 109–113.

Yang, G., & Jiang, M. (2015). The networked practice of online political satire in China:

Between ritual and resistance. The International Communication Gazette, 77(3), 215–

231.

Yangyu, B. S., Guangqian. (2017, Nov 24). Douyu exhibits on ECGC to empower Wuhan to

create the ‘fourth pole’ of the Internet (斗鱼参展 ECGC 赋能武汉打造互联网“第四

极”). People.cn. http://game.people.com.cn/n1/2017/1124/c218877-29666534.html

Yeung, E. (2020). China internet report 2019. SCMP. https://www.scmp.com/china-internet-

report

Yin, Y., & Fung, A. (2017). Youth online cultural participation and Bilibili: An alternative form

of democracy in China? In R. Luppicini, & R. Baarda (Eds.), Digital Media Integration

for Participatory Democracy (pp. 130–154). IGI Global.

Yosilewitz, A. (2019, Dec 20). State of the stream 2019: Platform wars, the new king of

streaming, most watched game and more! Streamelements.

https://blog.streamelements.com/state-of-the-stream-2019-platform-wars-the-new-king-

of-streaming-most-watched-game-and-more-ab0596d5c13d

265

YouTube. (2020, Aug 11). YouTube Trusted Flagger program.

https://support.google.com/youtube/answer/7554338?hl=en

Yu, L. (2018, Mar 9). For the second consecutive year, the proportion of GDP has exceeded 4%

for the cultural industry of Wuhan（连续两年占 GDP 比重超过 4% 武汉文化产业开

启换道超越）. Xinhua Net. http://www.hb.xinhuanet.com/2018-

03/09/c_1122509636_2.htm

Yuan, J. (2020, Feb 19). 疫情下的直播者：有人睡觉引 1857 万人围观 (Live broadcasters

under the epidemic: Someone slept and attracted 18.57 million viewers). Sina.

https://tech.sina.com.cn/i/2020-02-19/doc-iimxxstf2597783.shtml

YY. (2020). ALL categories. https://www.yy.com/catalog

YY Operation Team. (2018, Dec 12). Violating channel/personnel management regulations.

https://www.yy.com/n/33_537.html

Zang, Y. (2007). 从 “阳春白雪” 到 “下里巴人”——精英文化的大众化探析 (From

Yangchun Baixue to Xiali Baren – analysis on the popularization of elite culture).

Modern Audio-Video Arts, 8, 24–26.

Zeng, Y., & He, F. (2017). 我国网络直播行业的特点分析与规范治理 (Analysis of

Characteristics and Standardized Governance of my country’s webcast industry).

Research on Library Science, 6, 57–60.

Zhang, C. (2019, Aug 1). Douyu plans to launch live streaming service in Japan. Esportsinsider.

https://esportsinsider.com/2019/08/douyu-expands-into-japan

266

Zhang, C., & Liu, J. (2015). On crowdsourced interactive live streaming: A twitch.tv-based

measurement study. Proceedings of the 25th ACM Workshop on Network and Operating

Systems Support for Digital Audio and Video, Istanbul, Turkey.

Zhang, G. (2019). Zhibo: An ethnography of ordinary, boring, and vulgar livestreams [PhD

thesis, RMIT, Melbourne].

Zhang, G., & Hjorth, L. (2017). Live-streaming, games and politics of gender performance: The

case of Nüzhubo in China. Convergence, 25(5–6), 807–825.

Zhang, L. (2006). Behind the ‘Great Firewall’: Decoding China’s internet media policies from

the inside. Convergence, 12(3), 271–291.

Zhang, L. (2020). When platform capitalism meets petty capitalism in China: Alibaba and an

integrated approach to platformization. International Journal of Communication, 14, 21.

Zhang, W. (2020, Aug 18). Chinese web celebs appointed to party-backed youth organization.

Sixtone. http://www.sixthtone.com/news/1006068/chinese-web-celebs-appointed-to-

party-backed-youth-organization

Zhang, X., Xiang, Y., & Hao, L. (2019). Virtual gifting on China’s live streaming platforms:

Hijacking the online gift economy. Chinese Journal of Communication, 12(3), 340–355.

Zhang, X., & Ming, J. (2018, Mar 19). 兼职直播成热门 当心轻松背后的陷阱!合约签订要小

心 (Part-time live streaming has become hot, beware of the traps behind the ease! Be

careful when signing contracts). Guangxi Television.

http://gx.sina.com.cn/news/sh/2018-03-19/detail-ifyskeua9467475.shtml

Zhang, Y. (2020, Oct 15). 英雄联盟 S10 总决赛现“黑屏解说” 直播平台版权受限如何突围

(League of Legends S10 Finals shows ‘black screen commentary’: How to break through

267

copyright restrictions on live streaming platforms). Sina.

https://finance.sina.com.cn/chanjing/gsnews/2020-10-15/doc-iiznezxr6124299.shtml

Zhang, Z. (2020). Infrastructuralization of Tik Tok: Transformation, power relationships, and

platformization of video entertainment in China. Media, Culture & Society,

https://journals.sagepub.com/doi/abs/10.1177/0163443720939452?journalCode=mcsa

Zhao, E. (2016). Online Entertainment| Professionalization of Amateur Production in Online

Screen Entertainment in China: Hopes, Frustrations and Uncertainties. International

Journal of Communication, 10, 19.

Zhao, E. (2017). The bumpy road towards network convergence in China: The case of over-the-

top streaming services. Global Media and China, 2(1), 28–42.

Zhao, E. (2019). Digital China’s informal circuits: Platforms, labour and governance.

Routledge.

Zhao, L. (2017, April 12). Reveal the men behind the streamers: Agencies are more

‘complicated’. http://www.chinanews.com/cj/2017/04-12/8196923.shtml

Zhao, L., & Liu, X. (2017, April 12). 谁来监管网络直播行业经纪机构 ( who is supervising

live streaming agencies?). Legal Daily.

http://www.ce.cn/culture/gd/201704/12/t20170412_21897141.shtml

Zhao, Y. (1998). Media, market, and democracy in China: Between the party line and the bottom

line. University of Illinois Press.

Zhao, Y. (2008). Communication in China: Political economy, power, and conflict. Rowman &

Littlefield.

Zhao, Y. (2013). The content management of media in China. International Journal of Digital

Television, 4(3), 321–330.

268

Zhou, B. (2017). 谁在使用视频直播?——网络视频直播用户的构成, 行为与评价分析(Who

is using live streaming? Analysis of the composition, behavior and evaluation of online

video live stream users). 新闻记者 (news journalist), 3, 52–62.

Zhou, Y. (2006). Historicizing online politics: Telegraphy, the internet, and political

participation in China. Stanford University Press.

Zhu, Z., Yang, Z., & Dai, Y. (2017). Understanding the gift-sending interaction on live-

streaming video websites. International Conference on Social Computing and Social

Media, Las Vegas.

Zittrain, J., & Palfrey, J. (2008). Internet filtering: The politics and mechanisms of control. In R.

Deibert, J. Palfrey, R. Rohozinski et al. (Eds.), Access denied: The practice and policy of

global Internet filtering (pp. 5–28). MIT Press.

Zou, S. (2018). Producing value out of the invaluable: A critical/cultural perspective on the live

streaming industry in China. tripleC: Communication, Capitalism & Critique. Open

Access Journal for a Global Sustainable Information Society, 16(2), 805–819.

269