How equalitarian regulation of online hate speech turns authoritarian: a Chinese perspective

ABSTRACT This article reveals how the heterogeneous legal approaches of balancing online hate speech against equality rights in liberal democracies have informed China in its manipulative speech regulation. In an authoritarian constitutional order, the regulation of hate speech is politically relevant only because the hateful topics are related to regime-oriented concerns. The article elaborates on the infrastructure of an emerging authoritarian regulatory patchwork of online hate speech in the global context and identifies China’s unique approach of restricting political contents under the aegis of protecting equality rights. Ultimately, both the regulation and dis-regulation of online hate speech form a statist approach that deviates from the paradigm protective of equality rights in liberal democracies and serves to fend off open criticism of government policies and public discussion of topics that potentially contravene the mainstream political ideologies.


Introduction
The universal adoption of digital technologies has made it more convenient to spread various types of speech, including hate(ful) speech that impinges on racial, religious and gender equality, all of which constitute themes of socio-political significance. 1 Due to broad cultural discrepancies and political disparities, however, countries around the world endorse tremendously different regulatory systems to deal with unpopular speech that may be branded as hateful. 2 To be fair, most of these legal rules are rooted in conventional speech law supposed to be largely applicable in digital context. 3 But the idiosyncratic feature of the digital environment, together with the development of international politics and economic globalisation, attaches interdisciplinary considerations to the regulation of Internet speech. 4 Indeed, online speech regulation is an integral part of global Internet governance captured by the tension between the state, multi-stakeholders and the Internet itself. 5 Whereas the initial stage of Internet governance was characterised by the decentralisation of technological control and the marginalisation of the role of the state, 6 the beginning of the second decade of this new century seems to be bidding farewell to that entirely liberal yet at times anachronistic regulatory mode. This is a trend duly visible in a multitude of recently enacted laws and regulations to combat online hate speech around the world. 7 Thus, the regulation of online speech falls, inevitably, within the purview of the intransigent debates on the division of roles between state and society and is increasingly subject to the narrative of the confrontation between authoritarianism and constitutionalism. 8 In today's world, the governance models of Internet speech can be discerned roughly in a taxonomy of 'three empires of Internet': China, the EU and the US. 9 As such, more profound features are identifiable in the distinctions between speech regulation in authoritarian countries and that in democratic countries. 10 Amidst this turmoil, China often plays a leading role in exemplifying how authoritarian governments regulate Internet speech, in sharp contrast to liberal democracies. 11 Nevertheless, the borderline between authoritarian states and democratic states in their specific regulation of online speech is, on the surface, somewhat blurring, although the fundamental distinctions in the regulatory regimes remain intact. On the one hand, even democratic states need to stretch out their hands occasionally into a domain born to be free from regulation in order to find a compromise between free speech and other constitutional values. 12 On the other hand, authoritarian countries are also seized by capitalism, so they must take market factors into consideration when regulating the Internet, albeit to different degrees. 13 Therefore, certain measures that impose restrictions on online speech in both types of countries, when perceived independently of their underlying institutional, social and political structures, may be overlapping. This is the contemporary global context in which authoritarian states may well usurp international legal rules to reach overt or covert goals of changing the international legal order gradually across the borders. 14 It is against such a backdrop that this article seeks to explore certain features of the regulation of online hate speech that China is developing in conjunction with its overall target of online speech regulation. To be clear, hate speech, as it is so termed also in China, is situated only peripherally in the spectrum of political speech that the Chinese government aims to regulate wholeheartedly by means of various types of laws and regulations. This is because, as an authoritarian state, China lays its regulatory focus on traditional, core political speech that could be deemed subversive, inciteful and seditious to the ruling Chinese Communist Party (CCP), whereas racial, religious and gender-oriented speech are often regarded as less relevant in China's domestic political context. This is so even though ethnic tension and suppression of religious freedom often constitute a major topic in China's foreign relations, and, thus, a potential catalyst for political upheaval. 15 Therefore, not only does China's regulation of online hate speech exist in an emerging patchwork of legal rules that are politically oriented, but this regime also borrows extensively from what is known as the regulation of online hate speech in international and comparative perspectives. However, the implications of the application of similar legal techniques in a completely different setting are not to be ignored, since they may sometimes generate similar, if not unexpected and undesirable, effects on the daily use of such legal infrastructure elsewhere. For these reasons and those stated above, it makes sense to look into the Chinese case as a primary model in contradistinction to what is understood as the regulation of online hate speech universally. This article identifies a trend in Chinese law in which regulators take the protection of equality rights as a legitimate claim to regulate online hate speech but, in fact, subject such speech regulation to the broader regulation of core political speech. To engage with this unique approach, it is necessary to grasp its roots and paths in international and comparative perspectives. First, this article builds on the transition from traditional regulation of hate speech and addresses the new technological, legal, political and constitutional challenges posed by online hate speech to the protection of equality rights. Second, the article analyses the heterogeneous approaches by which European and US courts deal with those challenges and discusses differing roles of the state, private actors, and media organisations. Third, the article explains how those approaches have informed Chinese law while being used in different ways within an authoritarian constitutional order. Finally, the article develops a case study of the dis-regulation of 'White-Left' that commentators deem to represent Western liberalism in China, questioning both theoretical and practical aspects of that statist approach in light of the tension between free speech and equality rights.

The conventional imbroglio of hate speech and equality rights
Traditional hate speech law in Europe and the US focuses on the challenges to racial, religious and gender equality, which necessitates nuanced constitutional protection of equality rights. 16 A comparison of the judicial approaches in these two major systems belies a dichotomy between content-based and content-neutral speech. On the one hand, European courts apply a systemic interpretation of Articles 10, 14 and 17 of the European Convention on Human Rights (ECHR), highlighting an equalitarian and dignitarian reading of the protection of speech. 17 The Strasbourg jurisprudence ranging from Glimmerveen to Jersild well illustrates this. 18 More recently, the Lilliendahl case shows that the Strasbourg court stresses an 'assessment of the content of the expression' in determining the degree of protection granted to hate speech. 19 On the other hand, US courts have almost always held on to a 'content-neutral' stance regarding the regulation of hateful speech. Apart from the earlier deviation from this standard in Beauharnais, 20 both RAV and Virginia, 21 the two landmark cases, are testament to the rule that US courts do not endorse the regulatory approach of distinguishing between different types of speech based on their content or topic. The more recent Matal case reiterates the fact that the First Amendment protection would extend to hateful speech 'that demeans on the basis of race, ethnicity, gender, religion, age, disability, or any other similar ground'. 22 Therefore, there is a clear borderline between the two classical approaches: US courts would only consider whether they may prohibit certain hateful speech to stop imminent violence regardless of the discriminatory elements in speech, whereas European courts are obliged to draw a red line where they must protect equality rights from hateful and extremist speech. Despite such a dichotomy of judicial approaches in terms of the content of speech, both European courts and US courts would engage with a more profound debate between those who stand for an absolutist reading of free speech rights and those who strive for the equality rights of minorities. 23 This imbroglio of hate speech and equality rights has become the litmus test of international standards now enshrined in the prevalent conventions such as the Universal Declaration of Human Rights, the International Covenant on Civil and Political Rights (ICCPR) and the Race Convention, all of which are corollaries of rounds of earlier international negotiations that reflected conflicting values in dealing with hateful speech. 24 At the end of the day, the US' standpoints emanating from the jurisprudence of the First Amendment helped dilute the proposals led by the USSR that represented overall suppression of hate speech prone to incite national, racial and religious resentment and culminated in the criminalisation of war propaganda or genocide. 25 Apart from Article 19 of the ICCPR, a reconcilable approach adopted by most countries is hallowed in Article 20 of the ICCPR, which accommodates not only content-based elements such as national, racial and religious hatred but also relatively content-neutral standards such as discrimination, hostility and violence. 26 As a result, prevalent international jurisprudence is characterised by considerations that espouse both the criminalisation of hate crimes and legitimate caveats against misuse of speech-repressive laws. 27  profoundly influenced by, and associated with, the post-war European political context that precludes holocaust denial. In its leading case Faurisson, the United Nations Human Rights Committee (UNHRC) upheld the French court's conviction of anti-Semitist speech regardless of the speaker's intent under the three-step test of Article 19. 28 In another enlightening case, Rabbae, the UNHRC established Article 20 as a bridge between free speech and equality rightswhere, however, courts will need to draw a fine line of distinction between substantial harms and criticism of religious leaders or tenets such as that allowed under Dutch law. 29 Indeed, the leading regional jurisdiction that has developed a variety of judicial approaches toward finetuning of such international standards is the European Court of Human Rights (ECtHR), which grants member states of the ECHR a 'margin of appreciation' in making such decisions. 30 However, the ECtHR and domestic European courts are less tolerant of hateful speech than are US courts, where discriminatory considerations may be triggered.

New challenges of online hate speech to the protection of equality rights
With the development of the Internet, artificial intelligence (AI) and big data, the free flow of information and speech grows at such an exponential speed that hate speech in social media becomes an inevitable theme in Internet governance. 31 Arguably, various types of hateful speech in the virtual environment contribute to the distribution of political, national, racial, religious and gender hatred that, in turn, could be reflected in the physical world in ubiquitous forms of hostility, discrimination and violence. 32 Online hateful speech is common both in liberal democracies that promote free speech and in authoritarian regimes that impose ruthless censorship. longer confined to national territories, but can be extended to recipients across different jurisdictions. 34 Thus, states would often issue long-arm regulations of cross-border data and online hateful speech that reach remote senders and publications. 35 In that sense, all jurisdictions, democratic or non-democratic, need to deal with the conflict-ridden consequences of such online hateful speech, irrespective of whether the regulatory approaches they adopt are content-based or not. The technicalities of the regulation of online hate speech, although country-specific and region-oriented, are mostly backed up by considerations of discriminatory risks and could be seen as part of global Internet governance at least on an ontological level. 36 While digital media proffers the general public larger space for free expression, such seemingly equal rights to impart and receive information may become vulnerable to substantive inequality as a result of the marginalisation of certain minority groups at whom hate speech is targeted. 37 This is because minority groups are often positioned in a disadvantageous and hazardous environment such that their speech may not be heard or communicated effectively. 38 Besides, hateful and dangerous speech could be the precursor of physical violence. 39 This situation could be aggravated by the fact that the harms of online hate speech, if not subjected to regulatory and technological measures, are spread widely and exist permanently in the Internet, which deepens certain factual inequalities of different minority groups. 40

Technological risks
Digital technologies often make it difficult to detect and identify certain types of ideological discrimination and inequality. While formalistic equality between different groups is widely recognised, the harms that online hate speech causes to such equality rights may still obtain in an anonymous and dormant manner that potentially shapes users' way of thinking and behaviour. 41 Although online hate speech could be filtered via technological 33 Notably, in the latter case, ethnic tension constitutes a fundamental threat to political stability and regime existence; see Elliott (n 15) 40. 34  measures, such measures do not provide a fundamental solution, as the decentralisation techniques serve only to remove speech from one platform while users may transpose it to another. 42 In addition, certain drawbacks of AI-generated arithmetics prevent hate speech from being identified effectively, because the understanding of human language and speech goes far beyond the literal input of linguistic meanings. 43 Words meant to be entirely neutral can be imbued with hateful connotations in different contexts, and complex human culture makes it easy to develop countermeasures to eschew AI-generated filtering software. 44 Worse still, filtering keywords hinders the general public from exercising their right to free expression. 45

Political and constitutional factors
Traditional constitutional protection of equality rights presupposes the bipolar confrontation between state power and individual rights and hinges on the constitutional mechanism of preventing state power from encroaching on the civil liberties of individuals. With the revolutionary advent of digital technologies, however, part of the protégé of such constitutional power turns into digital media, a fourth constitutional power, 46 which alters the established model of Montesquieurian trias politica. 47 The decentralisation of media power enables a variety of influential distributors to acquire control of information in a way comparable to the governance of speech by public authorities. 48 Where the law fails to respond to such a fundamental change of political and constitutional infrastructure in a timely and effective manner, the multiplication of such public roles in media could further weaken the voice of minorities and endanger the protection of equality rights. 49

Legal challenges
Online hate speech also poses unprecedented challenges to the protection of equality rights when compared with the classical legal approaches to discriminatory practice in the past. First, corresponding legislation may lag behind 42 Nash (n 31) 449. 43  technological progress. Whereas the anonymity, immediacy and interoperability of the Internet makes it extremely convenient to distribute messages across the world, the automatic sharing and volatility of hate speech renders it difficult to provide a timely remedy for the harms to equality rights. 50 National legislation with respect to online regulations could plummet into a standoff because, even in traditional domains, there is no uniformly agreed set of rules of hate speech and equality rights. 51 Moreover, the nature of digital media that carries hate speech brings significant obstacles to direct supervision by government authorities, since it is almost impossible for them to ask even explicit but unidentified sites to delete hateful information thoroughly. 52 Still other websites have a variety of ways, such as using neutral domain names and a deceptive user interface, to cover up their discriminatory purposes. 53 These potentially harmful websites are hardly visible to supervisory bodies where one sets off to search for discriminatory information directly. 54 Besides, government authorities, above all, in the US, may not deal with complaints about hateful speech in a serious manner where there is no imminent danger of violence. Finally, despite the existing international standards, courts around the world would interpret cross-border hate speech in entirely different legal, political and cultural contexts. 55 The equalitarian dilemma: heterogeneous contours of regulating online hate speech These new challenges call for concerted efforts of regulation and cautious response to the potential harms of online hate(ful) speech. 56 In the wake of growing terrorism, racial tension and globalisation problems, online hate speech has elicited broad social and political impacts on minority groups such as among refugees and immigrants and impinged on democratic countries that embrace both free speech and equality rights. many US tech companies including Facebook initiated their own governance codes and adjusted their platform rules so as to curb hateful speech. In conjunction with such efforts, the EU also issued its Code of Conduct on Countering Illegal Hate Speech Online, which represents a non-binding call for a nondiscriminatory, tolerant and respectful Internet while stressing the importance of free speech. 58 However, such an emerging global regulatory patchwork still derives from the doctrines of protecting free speech and equality rights rooted in the heterogeneous contours of the constitutional and legal traditions of these countries. In essence, a liberal democracy requires restriction of public power including that of the media with regard to the potential encroachment on individual rights. 59 Yet the power of the media, if it can be dubbed as such, is different from state power in that the media power stems from the civil liberty of free expression, and restricting media in the way restrictions are imposed on public authorities may impair media freedom itself. 60 As a result, digital media often enjoys greater freedom of speech and transfers such freedom to its users. 61 Conversely, the protection of equality rights could be weakened disproportionately if the potential conflicts between these two fundamental rights could not be reconciled in this new context. Based on its stronghold of the protection of equality rights, Germany issued the 2017 Network Enforcement Act (Netzwerkdurchsetzungsgesetz -'NetzDG'), which requires social media to take down manifestly unlawful content within 24 h of receipt and imposes fines of up to €5 million on sites that violate such rules. 62 Nonetheless, it is this law that has become the focus of debates on the regulation of online hate speech with regard to its potential merits and drawbacks. 63 Admittedly, post-war European hate speech law relies on the protection of equality rights to mitigate racial and ethnic tension as a socio-political task. Thus, the ECtHR found prosecution of holocaust denial compatible with Article 14 of the ECHR when it upheld German defamation law in prohibiting such speech and sanctioning such an attack on the Jewish community. 64 However, the legal techniques of addressing relevant hate speech in member states could be rather sophisticated. For instance, German courts would not deal with hate speech against Jews directly as a violation of the constitutional right to equality, but tend to address such speech along the lines of human dignity under Article 1 of the German Basic Law (Grundgesetz, 'GG'). 65 In that connection, the German criminal law allows for state prosecution of insult of a member of any group persecuted by the National Socialist regime, 66 while making holocaust denial and its justification an offence. 67 The German Federal Constitutional Court (FCC) embraced such laws as the 'general law' under Article 5(2) of the GG that could apply legitimately to restrict the dissemination of this type of hate speech. 68 Under this interpretive framework, holocaust denial constitutes a false allegation of fact, although sanction of hate speech with group target can be compromised where the speech is meant to criticise the government. 69 By contrast, the Fourteenth Amendment to the US Constitution, which addresses equal protection, isn't and can never turn into an amendment to or containment of the First Amendment, so that US courts would not adopt a balancing act in its decisions. Instead, US courts abide by the bedrock principles of the First Amendment and refrain from restricting hateful speech merely for the reason that it contains discriminatory and derogatory messages. Following Chaplinsky, 70 a narrow category of speech was not treated as 'speech' for the purpose of the First Amendment. In Rav, 71 however, Scalia J, on behalf of the court, almost overthrew this rule by expressing the court's hostility toward any content-based regulations, i.e. there should be no content review of even fighting words. However, White J and colleagues suggested that states were free to ban any specific type of fighting words and that the court should consider whether the hate speech caused 'pressing public concerns'. 72 This line of reasoning was affirmed in Virginia when the court held it legitimate for states to ban a threatening message regardless of the content. 73 Such a logic indicative of an absolutist approach of the protection of free speech has continued to foment in digital context. In Matal, the Supreme Court struck down a federal law prohibiting trademark names that disparage others, finding that the law itself constitutes 'viewpoint discrimination', 74 which makes it difficult even to pass the Hudson test. 75 While hateful speech is not an exception to the protection of free speech under the First Amendment, its regulation does often pose an equalitarian dilemma. At best, the heterogeneous contours of regulation of online hate speech by European and US courts render the global regulatory framework a patchwork that calls for coordination and cooperation in legal, media and technological policies. This framework often requires redefining the roles of the state, private actors and media organisations. 76 Above all, multinational digital companies now play a decisive part across different jurisdictions, as the power of content review is relegated from public authorities to private content reviewers such as under the regulatory model of NetzDG. Clearly, the fear of dilated 'private censorship' is legitimate. 77 But it may also be true that such a 'privatised process' is channelled into the reasonable calculations of European and US jurisprudence. 78 Whichever way it is, such a new regulatory model is based on the premise of a largely free Internet in a democratic society, whereas this scenario leaves much technical leeway for nonliberal regimes to look to when designing their own regulatory system in an attempt to manipulate speech regulation.

An authoritarian approach: China's regulation of online hate speech under the aegis of equality rights
China's statist regulatory patchwork While China's authoritarian constitutional framework may not provide an effective approach protective of free speech or equality rights, the flourishing use of an even incomplete Internet by a significant proportion of Chinese citizens does call for expedient regulation of online hate speech that exists in various forms of discrimination. Indeed, China's constitution proscribes discrimination based on race, nationality, religion or gender in addition to the prescription of equal legal status of all Chinese citizens. 79 Moreover, comparable to the speech-regulatory framework in Europe, China's constitution allows for restriction of speech in terms of national security, public morals and others' rights. 80 Such constitutional mandates, however, are not to be construed as a balance between equality rights and speech rights 74 Matal (n 22) 4. 75  or as prioritising either when it comes to hate speech. Instead, they are conceptualised under a regulatory patchwork of hate speech in the digital environment, which China partly transplants from the Western legal system but steers in a different direction that hallows the 'national security' defined by the party-state. 81 Above all, China's online hate speech regulation is geared to its cybersecurity laws, which specify the constitutional doctrine of national security enshrined in the National Security Act. 82 Under the Cybersecurity Act, an underlying legal source of this regime, such a statist approach takes the form of allowing the government to assume 'cyber sovereignty' over Internet governance whose overwhelming substance is the maintenance of the monopolistic rule of the authoritarian regime itself. 83 This statist regulatory patchwork, under which state regulation is superior to the protection of speech, consists of three layers of rules. First, broadly defined racist and religious hate speech deemed a threat to China' national security is regulated effectively through prior restraint and government censorship. This approach often results in suppression of both free speech and equality. Second, a harshly applied approach is the subsequent punishment of online hate speech that may trigger social unrest and jeopardise public morals. This approach may take the form of both administrative penalty and criminal offence. However, it presupposes a limited space for speech and involves less consideration of the protection of equality rights than of social stability. Finally, while China has no constitutional review system, a rights-based regulatory approach is taking shape, above all, through the development of intermediary liabilities of Internet service providers that host, cache and transmit hate speech. Whilst there is still rather limited latitude for free speech, such an approach represents that of an indirect tort which draws the attention of both legislators and courts, and might involve, ultimately, qualified considerations of balancing speech against equality. These three layers of rules are discussed in turn below. proportionality. Above all, the Chinese government enacts a large number of administrative regulations in relation to online publication and information services, which share certain entrenched and overlapping norms. For instance, these regulations ban the publication of materials that 'incite the national hatred or discrimination, undermine the solidarity of the nations, or infringe upon national customs and habits', 'propagate evil cults or superstition' or 'endanger public ethics or the fine cultural traditions of any nationalities'. 85 In practice, these terms are often hardly distinguishable from other terms prescribed in conjunction with them, such as 'endanger the unity of the nation, sovereignty or territorial integrity' or 'endanger national security or damage the honour or benefits of the State'. 86 A recent government regulation targeting online speech singles out contents 'advocating terrorism or extremism, or instigating any terrorist or extremist activity', 'inciting ethnic hatred or discrimination to undermine ethnic solidarity' and 'detrimental to state religious policies, propagating heretical or superstitious ideas' that must be filtered by information service providers (ISPs). 87 Where a user violates such government policies, that online platform is obliged to take measures such as 'warning for rectification, restricting functions, suspending updating, and closing accounts'. 88 Meanwhile, the ISP must 'eliminate illegal information and contents in a timely manner, keep relevant records, and report to the relevant competent authorities'. 89 Any ISP that fails to comply with these content review requests will face severe administrative penalties amounting to temporary bans from network information services or permanent industrial bans. 90 Arguably, overburdening ISPs with strict censorship duties and administrative liabilities could result in overall bans of any politically sensitive contents by technological measures including marginal elements stirring up hatred or discrimination. Typically, ISPs undertake an opaque censorship process in which users are not allowed to participate. 91 Moreover, there are no accessible appeal mechanisms or legal remedies regarding the outcomes of censorship. This is observable in the regulation of speech ranging from racial and religious hatred to gender topics. Recently, the mobile social platform WeChat has shut down nearly 20 accounts run by campus LGBTQ clubs and non-profit organisations for suspicion of materials contravening mainstream political values, although they used to enjoy some privileges of equal protection sufficient to eschew stringent censorship. 92 That gap is now closed because, under the aforementioned legislation, public opinion authorities in China, which comprise both government propaganda departments and state media, are increasingly tying sexual minorities with 'colour revolutions' imported from overseas, as long as these groups trespass the red line of political speech. 93

Criminal law as an instrument of subsequent punishment
Second, the statist approach of regulating online hate speech aims at restricting freedom of speech and religious freedom by virtue of criminal law. Whilst China's constitution provides for the equality of 'all nationalities' and citizens' equal legal status, 94 Articles 249 and 250 of the Criminal Code of the PRC build up the major framework that regulates the incitement of racial hatred and racial discrimination as well as the publication of racially discriminatory and insulting materials: Article 249: Whoever provokes ethnic hatred or discrimination, if the case is serious, is to be sentenced to three years or fewer in prison, put under limited incarceration or surveillance, or deprived of their political rights. If the case is especially serious, they are to be sentenced to three to ten years in prison.
Article 250: Whoever is directly responsible for publishing materials that discriminate or insult minority nationalities, if the case is serious and results in grave consequences, is to be sentenced to three years or fewer in prison, or put under limited incarceration or surveillance. 95 Notably, Article 249 encompasses any attempt of incitement of racial hatred or discrimination against all nationalities, including the majority Han nationality, through language or words or by any other means. By contrast, Article 250 is targeted at any attack on the customs, traditions and life habits of minorities in China (excluding the Han nationality itself but including any preferences and taboos in their material and spiritual lives). However, the components of both offences are vaguely defined, so the standards for criminalising what amounts to a 'serious' degree and an 'especially serious' degree of racist hate speech assume large uncertainties. In the case of Pai, for instance, the defendant was charged with uploading 'terrorist pictures' on his personal blog whereas the defendant was unaware of the relationship between terrorism and the uploaded pictures of the flag of East Turkestan and the flag of Saudi Arabia. 96 Without explaining why this would elicit racial hatred and the specific harms that might be caused, the court adopted the prosecutors' view that online dissemination of these pictures served to instigate racial hatred to a serious degree that merited sanctions. 97 Furthermore, these offences could be used to cover up government campaigns of cracking down on core political dissident speech. In the case of Ma, for example, the defendant was sanctioned for disseminating extreme religious thoughts through mobile phones and for uploading, caching and transmitting pictures relating to extreme religious thoughts in order to stir up racial conflictswhile the real purpose was to criticise the government's religious policies. 98

Civil law as an instrument of marginal protection of speech and equality rights
Finally, China has developed some basic instruments in its civil law to protect equality rights against hate speech, although this rights-based approach is far from constituting a balancing act and provides merely limited protection of either right. Above all, China's new Civil Code defines the protection of 'personality rights' including the 'inviolability and integrity of person' and one's 'reputation' and 'honour'. 99 Additionally, it creates civil liabilities for tort of such rights: Article 1183: Where any harm caused by a tort to a personal right or interest of a natural person inflicts a serious mental distress on the victim of the tort, the victim of the tort shall have the right to require compensation for the infliction of mental distress. 100 In practice, such liabilities may be applicable to cases where hate speech concerning races across different regions, choice of religions, or gender equality causes serious harms to individuals, although harms of 'mental distress' are often measured by one's materialistic loss, such as unemployment. 101 More importantly, the Civil Code has adopted the 'notice and takedown '  an overall manner by defining the ISPs' obligation to delete, block and disconnect upon prima facie evidence of tort as well as their joint liabilities for infringing contents in case of actual knowledge. 102 While this regime covers online hate speech in a way comparable to the German NetzDG, cases adjudicated so far dwell overwhelmingly in the area of copyright and personality rights such as reputation and privacy rights. This is because most online hate speech has been filtered or penalised as a result of applying the aforementioned two approaches by government authorities and courts. With regard to hate speech, there is no decisive role for ISPs to play beyond rigid compliance with government policies.
Out of fear of additional monetary liabilities, ISPs practise stringent censorship of spontaneous hate speech but leave the guided speech with hateful elements intact. In fact, a wide range of online hate speech is part of a systematic, decades-long statist propaganda project driven by the party-state apparatus and its zealous followers inculcated with blind nationalism for the purpose of criticising and denouncing a political target by a statist attack strategy reminiscent of the Cultural Revolution. Recent victims include Hong Kong pro-democracy protestors, 103 Wuhan novelist Fang Fang who reflected critically on the Covid-19 outbreak, 104 and New Yorker journalist Jiayang Fan who was attacked because of her overseas Chinese identity. 105 More recently, Xu, a China-born journalist and researcher working for an Australian think tank, became the target of a vehement online harassment campaign when official videos deprecated her work, numerous social media accounts ascribed the global Xinjiang boycott to her, and infuriated 'patriots' threatened to harass her parents in China. 106 Stigmatisation of 'White Left': laissez-faire in regulating hate speech?
Whereas hate(ful) speech is often considered political speech, those hateful topics concerning race, religion and gender are not directly associated with the CCP's concern for regime survival. But this does not mean that hate speech is apolitical and irrelevant in China. On the contrary, the regulation of hate speech is often complementary to that of subversive speech. This is palpable in the fact that the most prevalent hate speech in China is derived from, and supported by, government discourses that play up, galvanise and racialise the 'national uniqueness' of 'Chineseness'. 107 In essence, the regulation or dis-regulation of such online hate speech is highly manipulable; this allows the one-party state to stir up a collective (un)consciousness of 'racial patriotism' which, chained to the mainstream ideology of nationalism, militates against any liberal-minded penchant in the mainstream parlance. 108 Despite those aforementioned regulations of racial, religious, gender and regional discrimination, China's government authorities, Internet platforms and fervent nationalistic netizens embrace collectively oriented hate(ful) speech against any target labelled as 'anti-patriotic' or 'traitors of the motherland'. The stigmatisation of 'White Left' (Baizuo) in China's Internet is a most eminent case that prima facie is part of the global racist discourse and at the same time betrays the politicisation of dis-regulated hate speech. A widely used definition of this term runs as follows: Baizuo is used generally to describe those who 'only care about topics such as immigration, minorities, LGBT and the environment' and 'have no sense of real problems in the real world'; they are hypocritical humanitarians who advocate for peace and equality only to 'satisfy their own feeling of moral superiority'; they are 'obsessed with political correctness' to the extent that they 'tolerate backwards Islamic values for the sake of multiculturalism'; they believe in the welfare state that 'benefits only the idle and the free riders'; they are the 'ignorant and arrogant westerners' who 'pity the rest of the world and think they are saviours'. 109 In the past several years, hostility toward Baizuo on the Chinese Internet have escalated, and more obscene, slanderous and insulting words are used against any target that could be branded with this title. Indeed, Chinese online 'patriots' use the term 'Baizuo' to label both Merkel and Obama to attack the refugee and gender policies of Western countries, 110 regardless of the skin colour of the actual targets. Although a racial label itself, the term connotes more a politically 'left' identity than a racial one. To be sure, these debated elements of cultural and identity politics are rarely seen and only peripherally represented, if they exist at all, in Chinese society. The fact that these remote topics are scorned, belittled and disparaged widely on the Chinese Internet without being censored or punished presents a stark contrast to the regulation of hate(ful) speech that could be channelled eventually to regime-related vulnerabilities such as unfair minority, religious and gender policies attributable to the authoritarian system.
As a highlight of its political nature, this hateful term can be applied mutatis mutandis to any Chinese who carry or exhibit similar hallmarks of such pro-liberalism ideology so as to limit their speech. For instance, many Chinese netizens conceived of Professor Rao, a famous neurobiologist in China who openly criticised the smearing use of Baizuo and Chinese netizens' support of Donald Trump, as an identifiable 'Baizuo' characteristic of their 'elitist sense of superiority, bigotry and ignorance'. 111 Seen from an ideological perspective, the ease of disseminating such an anti-Baizuo mentality among Chinese netizens is rooted in a widespread pre-Hobbesian philosophy of power politics and social Darwinist theory of 'survival of the fittest' under China's authoritarian regime. 112 Consequently, this hateful term acquires its domestic immediacy and influential implications in China in the wake of conflicts between established urban residents and mass immigrants from rural areas, ethnic and religious tensions between Han and non-Han minorities, and the confrontation between sexually conservative groups and LGBT or feminists. 113 By adopting the 'White' as a racial factor, such a debate on this prevailing hate(ful) term against alleged race traitors has gained a strong tinge of political discussion 'without an awareness of its racial nature'. 114 Obviously, a guideline for a nationalist campaign encapsulating vague definitions of national security, cyber sovereignty and incitement to racial, religious and gender hatred and discrimination in China's legal framework is conducive to the ineptitude of the hate speech laws in regulating such a unique type of hate(ful) speech with rich and comprehensive externalities. The widespread use of the so-called 'Baizuo' on the Internet is evidence of a politicised speech campaign in China, which dwells in a lassez-faire state within China's existing regulatory patchwork. Notably, this campaign increasingly takes place under the tutelage of 'combatting Western hegemony', 115 and fits perfectly into China's statist approach of constructing the image of a rising power against that of a steadily falling Western civilisation. Oddly and ambivalently, the use of this term among many Chinese nationalists is coupled with their profound concern, anxiety and, perhaps, sympathy regarding the 'degeneration' of the White race itself, which is exemplified most saliently by the narrative 'blacks and Muslims will take 111 Ibid. 112 Zhang (n 84) 101-04. 113 Cheng (n 107) 285-86. 114 Ibid 288. 115 Zhang (n 84) 90-94. over Europe' due to the extremely low birth rate of the White race and the 'extremely humanitarian' mindset in the Western society toward those immigrants. 116 Thus, such a campaign of hateful speech can generate extra-territorial influence over Western politics. 117

Conclusion
The post-war hate speech law around the world is characterised by a transition from the classical liberty-focused discourse to a balance-oriented approach in Europe and, alternatively, deeper engagement of the First Amendment with equality in race, religion and gender in the US. This imbroglio of free speech and equality rights continues to dominate the digital scenario with regard to most hate(ful) speech despite the technological, political and legal challenges pertaining to diversified structures of various ISPs and their social media platforms. While their regulatory approaches are taking shape with different legitimate stances, the disparity between the EU and the US in dealing with online hate speech by equality rights does provide a multitude of instruments and flexibilities for authoritarian regimes including China to emulate, albeit with different purposes.
In post-socialist China, the regulation of online hate speech under the aegis of protecting equality rights is largely characterised by a guideline of nationalist campaign and demoralised pragmatism that informs a patchwork of administrative law, criminal law and civil law. Administrative law serves as an online valve of prior restraint guided by a vague concept of national security and cyber sovereignty. Criminal law acts as subsequent punishment to deter any anti-state and anti-social online speech. Civil law provides merely marginal protection of speech and equality rights by imposing stringent liabilities on ISPs and content providers. All these legal boundaries bear the hallmark of the equal protection of different nationalities, races, religions, regions and genders.
A deeper look into a unique online hate(ful) term in China (Baizuo) and its immunity from such regulation reveals that China's online hate speech law represents a statist approach that deviates from the paradigm protective of equality rights in liberal democracies. Whereas the law imposes restrictions on online speech by virtue of mandates of equal protection, the statist campaign of hate(ful) speech often trumps the latter and serves to fend off open criticism of government policies and public discussion of topics that potentially contravene the mainstream political ideologies.
Moreover, such a legal regime encourages the promotion of its extraterritoriality. Ultimately, the regulation and dis-regulation of such statist hate(ful) speech runs counter to both European law and US law, which aim to reconcile potential conflicts between liberal interests and equalitarian interests.