Total Mobilization from Below: Hong Kong ’ s Freedom Summer

This article examines the origins and dynamics of an extraordinary wave of protests in Hong Kong in 2019 – 2020. Despite lacking visible political opportunities and organizational resources, the protest movement drew resilient, mass participation unparalleled in the city ’ s history and much of the world. Drawing from original on-site surveys and online datasets, we conceptualize the Anti-Extradition Law Amendment Bill Movement as a form of “ total mobilization from below. ” The totality of the mobilization depended on a set of interactive mechanisms: abeyant civil society networks concealed after the 2014 Umbrella Movement were activated by threats over extradition and institutional decay, whereas affective ties developed through conflicts and mutual assistance were amplified by digital communication. The movement ’ s characteristics in terms of protest scale, mobilizing structure, use of alternative spaces, and group solidarity are examined. The spasmodic moments of mobilization are explained by a nexus of network building that took place in an unreceptive environment and at a critical juncture. The roles of threats and emotions in mass mobilizations are also analysed.

"taste freedom" when returning from the mainland to Hong Kong. 11 The prominent land developer Joseph Lau filed a judicial review to contest the bill. 12 Yet, the pro-establishment bloc unanimously expressed unconditional support for the bill after being summoned by Beijing officials in May 2019.
Alongside the changing nature of Beijing's governance in Hong Kong after the Umbrella Movement, the democratic opposition's organizational resources and institutional platform had been weakened significantly. Although the prodemocracy camp secured more seats in the 2016 Legislative Council (LegCo) elections than it ever previously had, the government disqualified six elected legislators for taking their oaths of office improperly. A few dozen leading activists were sentenced to prison for participating in illegal assemblies. Meanwhile, the rise of localism created an ideological rift within the opposition camp, preventing veteran pro-democracy parties from recruiting young members and new localist parties from accessing pro-democracy networks. 13 While the state had seemingly cowed the populace into submission, many informal civil society networks had remained in place, albeit latently. Without mounting overt challenges, activists preserved their values, identity and vision during this abeyance. 14 This interactive process highlights how the "eventfulness" of a momentous event has the power to profoundly transform social structures, produce political subjectivity and shape public acceptance of unconventional actions. 15 Nevertheless, the mere existence of abeyance networks alone does not explain the timing and scale of mass mobilization. Mass mobilization requires protesters to have the frames to articulate their grievances, the means to communicate with one another and the solidarity to coordinate actions.
In another vein, Charles Tilly and Jack Goldstone contend that threats can be equally crucial as opportunities and resources in mobilizing contentious action. 16 Paul Almeida insists that threats are potent in transforming individual cognition into collective action, indicating the "probability that existing benefits will be taken away or new harms inflicted if challenging groups failed to act collectively." 17 These conceptualizations move beyond considering threat in terms of a negative opportunity and address the costs associated with inaction. Specifically, threats often foster mobilization in hybrid and authoritarian contexts where the responsive institutions and elite division commonly present in democracies are absent. Yongshun Cai's and Ming-sho Ho's studies of contentious events in China and Taiwan affirm that grievances are often better articulated and mobilization better organized in response to threats. 18 Research on 11 HK01 2019. 12 South China Morning Post 2019. 13 Veg 2017. 14 Taylor 1989. 15 Ku 2019Lee and Sing 2019;Cheng, Chung and Cheng 2021. 16 Goldstone and Tilly 2001. 17 Almeida 2003, 347. 18 Cai 2008Ho 2015. Hong Kong's democracy movement has generated analogous concepts, such as "rearguard politics," "civil society in self-defence," "citizen self-mobilization" and the "nascent movement society" to explicate mass mobilization. 19 However, these studies tend to regard authoritarian encroachment or institutional loopholes as a backdrop and focus on the organizational and communicative aspects of mobilization. How threats are perceived among the crowd and are translated into affective ties remains unclear. 20 Building on earlier works, we propose to distinguish threats by considering their temporality and severity. 21 We conceptualize two forms of threat: 1) "systemic threat" that involves a fundamental erosion of rights, values and institutions; and 2) "repressive threat" that involves state actions ranging from legalistic repression and police tactics to counter-mobilization. Systemic threat requires the evaluation of pre-existing institutional conditions in a society whereas repressive threat arises in relation to state responses to dissent. Both forms of threats combine to influence individuals' cognition and risk acceptance, which in turn shape their decisions of whether to join collective actions.
Furthermore, the relational and cultural turns of contentious politics offer alternative lenses through which to mediate the mainstream structural-rationalist approach. Instead of seeing social movements as having a dispositional tendency, Tilly, Tarrow and McAdam emphasized the transactions among individuals and groups and favour a mechanisms-oriented approach to social change. 22 James Jasper adds that emotions are present in almost every phase and aspect of social movements. 23 Emotions are not only the formative conditions of social movements but also constituents of affective ties among protesters. 24 At critical junctures, movement actors often experience a range of emotional upheavals. 25 A mixture of positive and negative feelings such as hope, awe, fear, anger, sadness and guilt interact, and sometimes become shared feelings among strangers working for a common goal. Protest tactics can be dramatically radicalized, and can meet with more public acceptance, when encountering repression or countermobilization that protesters consider unfair or unjust. 26 These relational and cultural lenses contextualize the tactical innovations and social ties that were built during Hong Kong's Freedom Summer.

The Conditions and Characteristics of Total Mobilization from Below
Charles Tilly defines mobilization as "the process by which a group goes from being a passive collection of individuals to an active participant in public life." 27 Sidney Tarrow introduces the cycles of contention as "phases of heightened contention across a social system with a diffusion of collective action from more mobilized to normally less mobilized sectors of the population." 28 In contrast, the notion of total mobilization has long been associated with topdown initiatives organized by the state during wartime and peacetime "megaprojects." The controversial theorist Ernst Jünger defines total mobilization as "the channelling and management of the entire force of the nation, the people, or the proletarians of the world, into a process of production and self-production driven by the highest possible level of technological and managerial order." 29 Our definition borrows from this in terms of collective resources and mutual reliance characteristics but contests the assertion that an organizational hierarchy is the precondition for total mobilization across classes, sectors, spaces and generations. Instead, we propose the term "total mobilization from below" to conceptualize the unprecedented and resilient mobilization in Hong Kong's Freedom Summer. It serves as the foundation for the following explanation of how abeyance networks, threats and emotions were constituted during this critical juncture. It also serves as a conceptual tool to unpack the dynamics unfolding from many contemporary networked movements against democratic backsliding worldwide. 30 We will use the evolution of protest frames, claims and tactics to illustrate the concept.
Total mobilization from below depends on a combination of interactive mechanisms. The concealed civil society networks after the Umbrella Movement facilitated public deliberation, initial mobilization and frame articulation, despite the absence of a centralized movement leadership. The most popular slogan at the movement's beginning was "no extradition to China" ( fannsungzung 反送中), a Cantonese homonym that carries the double meaning of antiextradition and opposition to the "death" perceived to be brought about by the extradition bill. While the origin of this slogan remains disputed, its popular acceptance was realized through collective efforts of civil society networks: promotion by opposition parties and community groups, articulation by key opinion leaders on digital media, and the Civil Human Rights Front's (CHRF) adoption of the slogan in the two mass rallies in June and beyond.
The mobilization, however, would not have become unparalleled in scale and duration without attending to the participants' cognitive liberation and the building of affective ties. From the outset, the lack of central leadership meant that no leaders or organizations enjoyed the legitimacy and authority to command and coordinate the movement. Protest claims and action repertoires had to be constantly adjusted to appeal to different audiences while maintaining internal solidarity. The demands of the Anti-ELAB Movement, summarized as the "Five 27 Tilly 1978, 69. 28 Tarrow 2022 Cited from Costea andAmiridis 2017. 30 Lührmann andLindberg 2019;Caraway 2021. Demands," were not proposed by any movement organization. Instead, they were a collective endeavour that took place both offline and online and between multiple actors. The earliest demands of the movement emerged from a post on 12 June 2019 on the LIHKG forum. 31 After the political suicide of Marco Leung Ling-kit on 15 June, which aroused public outcry, the demands were revised to observe Leung's final words. The Five Demands soon became the uncontested protest claims; they were, "withdraw the bill; retract the riot definition; establish an independent commission of inquiry into police conduct; no arrest and prosecution of protesters; institutionalize genuine universal suffrage," corresponding to the political accountability, institutional oversight and political development prescribed under the Basic Law. 32 From this perspective, total mobilization from below is different from other common terms applied to social movements. Unlike a revolution (or an uprising/revolt) that aims to systematically overthrow the current political order, total mobilization from below observes the constitutional order and its protagonists frame their struggle in line with the systemic threats to the current political system. While both revolutions and total mobilizations from below are critical junctures, their protagonists differ in the ways in which they see government accountability and movement goals. 33 Unlike revolution, which often resorts to violence to take political power or achieve individual goals not necessarily approved of by the majority, total mobilization from below develops its protest repertoire through deliberated protocols and claims. Despite the repeated occupation of luxury malls and the spread of vandalism during the Anti-ELAB Movement, the police and the media found no instances of looting. This unusual conduct suggests that the movement was politically driven, and affective ties might have restrained self-interested behaviour.
We conceptualize total mobilization from below as mass acts of defiance performed collectively and persistently at a critical juncture. In the face of a crumbling socio-political order, total mobilization from below is often not a calculated move but constitutes the sum of actions and reactions to conjoined threats and group emotions. Threats over extradition to an authoritarian jurisdiction, emotions associated with protest policing, and the construction of collective identities sustained a decentralized but coordinated network in the Anti-ELAB Movement. Under this networked structure, frames were aligned, opportunities were explored and resources were crowdsourced dynamically. The totality of this mobilization had four characteristics in terms of scale, structure, space and solidarity among different segments of society: 1) enacting a massive-scale of mobilization including individuals who would not otherwise participate; 2) endorsing a diverse protest repertoire whose elements were linked to one another horizontally; 3) the use of alternative spaces for contention in order to explore 31 https://lihkg.com/thread/1224111/page/1. 32 , 22. 33 See Bayat 2017 new opportunities and resources; and 4) fostering affective ties to maintain internal solidarity.

Data and Methods
To examine the total mobilization of the Anti-ELAB Movement, this paper relies on a variety of data sources. First, we analysed the signatories of online petitions in May 2019, which initiated the territory-wide mobilization. Second, we conducted 26 on-site surveys between 9 June 2019 and 1 January 2020, covering every major protest of the movement and obtaining the motivations, tactics, emotions and demographics of more than 13,000 protest participants. We followed established methods to use the spatial distribution of the protesters as the sampling frame and a systematic sampling procedure to select individual respondents. 34 Third, we drew on computational network analysis of 25 million comments on the LIHKG forum, the main mobilization platform of the movement, between 1 April 2019 and 31 January 2020. The topic analysis will reveal how digital communication allowed protesters to construct frames and cultivate identities in the absence of centralized leadership. Finally, we report several testimonies from organizers and volunteers of professional and community groups to elucidate the role of abeyance networks.

Networks and Practices during the Abeyance
Despite the absence of territory-wide mobilization, the post-2014 period saw the rise of grassroots groups, sectoral initiatives and digitally enabled connective actions. 35 These abeyance networks emphasized the connecting of politics to everyday life, networking informal groups and broadening citizens' democratic imaginations. 36 These post-Umbrella Movement groups and platforms used "hidden transcripts" to subtly disseminate their understandings of civil liberties and citizenship and developed a loose "network of networks" through grassroots penetration and digital communication. 37 These groups were often informal and covert, lacking a hierarchical structure and official membership. Trust was built upon personal, social, or professional ties, whereas actions were coordinated through WhatsApp groups or Facebook pages. Their covertness allowed them to observe the boundary between advocacy and mobilization while at times being critical and vocal. Admittedly, these community networks alone cannot account for the unprecedented mobilization in 2019. Yet, they served as the online or offline nodes by which to perform special functions at specific stages.
At least 60 grassroots community groups were formed after the Umbrella Movement. An example was a community network named Sai Wan Changing. 34 Yuen et al. 2022. 35 Lee and Chan 2018. 36 Mathews 2018Pan 2020. 37 Scott 1990Castells 2015. Between 2016 and 2018, the group organized hundreds of festive events, public film screenings, book exchange boxes and information boards, as well as a community school at the western end of Hong Kong Island. These events often featured books, films and songs that were heavily censored on major commercial outlets. These micro-level and spatial practices sought to redefine the connective function of the urban space and to deepen the meaning of democratic life. By May 2020, the group had 83,000 members on Facebook. When protests became part of the city's everyday experience during the Freedom Summer, group members actively shared images and news about the movement and called for collective action.
Co-founded by veteran activists, Fixing Hong Kong was a representative grassroots network composed of educated youth and blue-collar workers, many of whom are not traditional supporters of the pro-democracy camp. They met almost every evening to provide free in-house repair services in redevelopment neighbourhoods in To Kwa Wan. The hidden agenda of this face-to-face interaction was to help residents realize how their poverty and displacement were situated in the political context. Between 2015 and 2018, the activists engaged in more than 2,000 visits. Their innovative model expanded their volunteer team to include ethnic minorities and attracted established nongovernmental organisations to seek collaboration. One of its founders explained how they built up strong ties and participatory experiences at the grassroots level: We want to make politics matter in everyday encounters. We do not restrict ourselves to repairing goods and offering welfare. What we aim to heal is each other's souls, through hard work. And we have deepened the ties with those we encountered. Our clients invite us to their festive events, and we bring them back to attend forums. 38 In parallel, more than 60 professional groups of doctors, lawyers, social workers, artists and accountants were also formed. 39 They shared the conviction that many traditional associations or trade unions in their sectors were either too conservative or hemmed in by administrative red tape and that a smaller and informal group of people sharing similar views tends to be more flexible and better disposed for action.
A group of outreach social workers formed the Reclaiming Social Work Movement. Aiming to inject progressive politics into social issues, these social workers focused on fieldwork and public education. The group had approximately 30 core members in its WhatsApp group that were willing to work on issues related to electoral politics and social empowerment. The group, despite being small in scale, created dissenting voices among the professional organizations, pushing them to acknowledge the need for reform. During the early months of the Freedom Summer, the group remained set on pushing forward the agenda of supporting young protesters. Some of them even stepped up to 38 Interview, activist, Hong Kong, 11 November 2018. 39 Ma 2020 the frontline to mediate between protesters and police and formed alliances with other informal civil society groups.
Another notable group, the Progressive Lawyers Group, adopted a more proactive and grassroots approach than the Hong Kong Bar Association and the Law Society of Hong Kong would sanction. When controversial legal issues arose, they produced easy-to-understand versions of legal concepts and prosecution procedures. They also worked with volunteer lawyers to bail out arrested activists and protesters after 2015. By early 2019, their group had reached a size of more than 120 legal professionals. In summer 2019, they gave more than three hundred interviews to share their legal analysis with the local and foreign press. A convenor of the group articulated how managing differences built up solidarity in an informal group: We believe that our legal expertise can serve society better. While some colleagues stress advocacy, others contest boundaries. We value these differences as they are deliberative. Regardless or our rank and politics, we have been more committed. We make ourselves available for countless bails, trials, commentaries and interviews. 40 Alongside this, many online and social media platforms continued to allow activists and groups to connect with their potential supporters while lowering the cost of content distribution. 41 When the political environment was non-receptive to mass mobilization, digital communication helped people to remain connected with political actors, aided in aggerating resources and preserved dissenting voices.
First, a few online media platforms emerged in response to what they perceived as a highly censored media landscape after 2014. Often formed by professional journalists, these platforms ensured their editorial autonomy through crowdfunding or subscriptions. Notable examples included Stand News, Citizen News and the Fact Wire News Agency, which offered a variety of investigative reports, expert columns and feature stories. They provided live broadcasts and fact-checking during 2019 and became the primary source of movement information for many. 42 Second, political commentators revamped their channels on YouTube. These key opinion leaders provided instant analysis of current affairs and rallied huge and loyal audiences. MemeHK, SingJai and D100 were popular channels whose digital subscriptions increased two-to-threefold during the movement. Third, LIHKG, a Reddit-like online forum, was established in 2016. It was a spin-off of Golden Forum that was active in the Umbrella Movement. Due to concerns that state-backed internet commenters had infiltrated the original forum, the new forum adopted a set of precautions. LIHKG registered an overseas server as a measure against censorship and cyber-attacks. It adopted a registration system that required email addresses provided by internet service providers for identity verification. It provided a user-friendly interface that 40 Interview, barrister, Hong Kong, October 2019. 41 Chan 2020. 42 See Lee et al. 2021b. highlighted the most discussed topics and allowed users to upvote or downvote topics. These unique features of LIHKG, in contrast to Twitter or Facebook, created a pluralistic yet centralized communication platform that enabled the brainstorming of tactics and frames during 2019. 43 Altogether, these informal and digital networks not only enabled the total mobilization, but also preserved progressive values during the abeyance period, developed frames in the initial mobilization and facilitated division of labour to sustain the protests. During the 16 June rally, for instance, we counted 153 professional, community and political groups. These civil society networks were not movement leaders, but they provided support for ordinary citizens to engage in collective action.

Movement Experiences and Initial Mobilization
When the Hong Kong government introduced the extradition bill on 12 February 2019, formal organizations remained the early risers in organizing collective protest actions. 44 On 15 March, Demosisto, a youth-led political party, organized a sit-in at the government headquarters in Admiralty. Nine activists were arrested, and the bill passed its first reading. On 31 March and 28 April, the CHRF organized two rallies and claimed turnouts of 12,000 and 130,000, respectively. These protests created the momentum for pro-democracy politicians to filibuster in the LegCo bill committee and stall the amendment. Meanwhile, online discussion about the bill became more intense. On 5 June, Apple Daily, the largest prodemocracy newspaper, produced a trilogy of short films entitled Forests and Fields for Animal Fugitives, On the Chopping Board and Imprisoned Night. These films attracted more than 1.5 million views on YouTube before the 9 June rally. These organized actions initiated by veteran movement organizations and the media successfully raised awareness about the extradition bill and highlighted its impact on the existing order.
However, what truly galvanized society-wide opposition was the emergence of the online petition campaign in opposition to the extradition bill in late May. The petition campaign first started with the city's secondary schools, when alumni and students from each school drafted petition statements and circulated them online. Many of these petitions invoked school mottos to frame the bill as an erosion of the city's core values and institutions, while some appealed to senior government officials who were alumni. 45 The campaign swiftly spread to different industries: lawyers, bankers, accountants, journalists, doctors, nurses and tech industry workers all crafted petitions in their own words. Parents, homemakers, immigrants, churchgoers and residential groups followed suit. Our research found 487 petitions being circulated online by early June, which gathered more than 270,000 signatures in total. Secondary schools made up 55 per cent of these petitions, tertiary educational institutions 16 per cent, primary schools 10 per cent, professionals 9 per cent, community organizations 5 per cent, religious groups 3 per cent, and overseas networks 2 per cent. Most of the petition initiators did not hold any positions in movement organizations or political parties. They were ordinary citizens who counted on digital and social networks to mobilize others.
Who were the signatories? Table 1 shows the correlations between the signing of an anti-extradition bill petition in May 2019 and participation in major social protests since 2012. Those who participated in previous protests were more likely to sign a petition. This pattern reveals that protest participation has an enduring impact over one's lifetime, but the impact is mediated by the nature of the protest events. Participants in the 2012 anti-national education campaign (in response to a school curriculum on "moral and national education" proposed by the Education Bureau of Hong Kong in 2012) tended to sign petitions through their secondary school or university networks. Participants in the annual July 1 rallymarking the anniversary of the British handover of Hong Kong to China in 1997more frequently signed petitions through their university or professional network. The ripple effect of the Umbrella Movement was particularly strong. More than 75 per cent of its participants signed a petition through their university/alma mater or industry sector. Memories of the June 4 Tiananmen Incident have been kept alive with an uninterrupted vigil at Victoria Park that has lasted for more than three decades, which significantly mediated the local populace's perceptions of threat and resistance. 46 The above pattern of participation history reveals the tendency for self-mobilization at a critical juncture of the Anti-ELAB Movement. Instead of relying on the calls or resources of movement organizations, networked individuals were capable of expending their social capital to defend their liberties and ethos. 47

Conjoined Threats and the Unprecedented Scale of Mobilization
Protest scale helps to indicate the extent of social grievances and movement strength. It provides protesters with the signal that their claims are shared by many. While the above discussion on abeyance networks and online petitions has addressed the question of how the Anti-ELAB Movement came about, the question of why it did remains unanswered. Our empirical findings suggested that the systematic threat concerning the erosion of the city's values and institutions imposed by the extradition bill was a primary cause of the unprecedented mobilization. The regime's interceptive strategies, including legalistic repression, police tactics and counter-mobilization, thereafter constituted concrete examples of a repressive threat.  Authors' on-site survey on 16 June 2019 with a sample size of 876. Entries in the "Yes" and "No" columns are percentages of respondents who signed an anti-extradition bill petition according to their participation in a specific past protest. Respondents could choose multiple petitions, so the total percentage in each column can exceed 100. The X 2 values are derived from cross-tabulating the participation in the four protests in agreement with the statement. * p < 0.05; **p < 0.01; ***p < 0.001 According to the organizers, 1 million and 2 million people participated in the rallies on 6 and 16 June, respectively. Some big data-based analytics provided estimates of up to 755,000 and 1.44 million, respectively. 48 Our two population surveys in August 2019 and May 2020 indicated that 36.4 to 42.2 per cent of the local population of around 7 million participated in the movement, suggesting a resilient mobilization after the rallies. The 2020 survey found that the movement drew on significant participation from groups not typically associated with social movement involvement: 49.3 per cent were aged 40 or older, 21.4 per cent had a secondary education or less, 31.9 per cent were from upper-middle-class and high-income households, 14.4 per cent had centrist or pro-establishment ideologies and 10.4 per cent had no prior protest experience.
The scaling up of mass participation within one week in June 2019 was shaped by the interactions between the state responses to the protests and the public threat perception. After the first rally, the Hong Kong government issued a statement acknowledging different views in society but insisted that the second reading of the bill would resume on 12 June. 49 Tens of thousands of protesters, mostly young ones, then occupied Harcourt Road and surrounded the LegCo building on the date of the second reading. Police and protesters clashed, and the meeting was adjourned. Later that afternoon, Carrie Lam labelled the protesters as rioters. The decision to carry on with the bill reading and the riot designation revealed a resolute government and amplified perceived threats regarding the lack of political accountability. Although Lam announced the suspension of the bill on 15 June, this coincided with the suicide of Marco Leung Ling-kit, after he had unfurled a banner denouncing the bill. While the influence of this contingency cannot be fully measured, our field observations recorded that nearly one in twenty protesters were holding white flowers to pay tribute to the deceased in the days following Leung's death. Leung's five demands on the banner would become the major protest claims. Table 2 summarizes the motivations for participating in the three massive rallies on 9 June, 16 June and 8 December. The two on-site surveys in June found that worries over the "extradition of pro-democracy activists and politicians to mainland China," the "extradition of the general public critical of political affairs of mainland authorities," "the destruction of the rule of law in Hong Kong" and "the end of one country, two systems" were the participants' most important concerns. More than 90 per cent of respondents considered these threats to dissention and civil liberties to be their primary concerns. Other threats to Hong Kong's status as a financial centre and global connectivity formed the second tier of concern, whereas participants were least motivated by worries over personal safety and household wealth. After the extradition bill was officially withdrawn on 23 48   Respondents were asked to rank their preference on a 5-point Likert scale. The percentages for extremely worried and worried responses were combined.
October, unregulated protest policing was seen as a repressive threat that reinforced a systemic threat over the erosion over civic rights and accountable institutions. In the December 8 survey, while 96.8 per cent of the respondents expressed their dissatisfaction with the police handling of the protests, 94.7 per cent continued to believe that the extradition bill threatened the semiautonomous city's constitutional status.
Moreover, systemic threats continued to amplify from the first to the second rally, whereas repressive threats became more salient as the movement unfolded. Normally, the larger the rally is, the higher the chances of having diverse participants and hence nuanced responses. Yet the results on 16 June suggested that fear and anxiety were widespread and consensual, despite the increase in protest scale. All indexes of concern rosefor example, personal safety concerns over the possibility of "extradition of you, your family, or friends" increased from 56.2 to 79.9 per cent. The data suggest that the respondents leaned towards the "extremely worried" end of the spectrum and that their worry had increased significantly in one week. Figure 1 further captures the scale of mobilization from March 2019 to February 2020. The trends show an initial increase in the number of protesters per protest over time; thereafter, the number of protests began to increase and the number of protesters per protest began to decline. The point of interception between the two trends happened in mid-August, around the time when the police began to consistently deny protest applications on July 27, precipitating several events. On 7 August, protesters started to vandalize property following several attacks on civilians. On 11 August, a young volunteer medic was shot by a beanbag round and was believed to have been blinded. This incident galvanized the occupation of road junctions and the airport in the weeks that followed. Meanwhile, the police used more resolute means to disperse the crowds, from firing teargas and beanbags, beginning in June, to making mass arrests and deploying water cannons, beginning in August. As of 30 January 2020, 21,000 rounds of teargas had been fired, 10,000 arrests had been made and more than 3,000 injuries were registered in connection to the Anti-ELAB Movement. 50 What explains the changing pattern of mobilization? One possibility is that state repression significantly increased the cost of participation. Another possibility is that protesters made tactical adjustments by shifting toward protesting in communities and workplaces. While both factors shaped the movement trajectory, the threat of repression often prevailed. Polling data indicated that public approval of the police force dropped from 61 to 39.4 per cent from early June to early August 2019. 51 In another poll, the number of respondents who gave a score of zero for trust in the police force rose from 5.3 per cent in early June to 51.5 per cent in early October 2019. 52 Throughout the period, these polls 50 Stott et al. 2020. 51 HKPORI 2020. N=1006 and N=1015, 3-6 June and 1-6 August 2019. 52 CUHKCOPOP 2020. N=1048 and N=751, 23 May-3 June and 8-14 October 2019.
showed that 70 to 80 per cent of respondents supported the establishment of an independent commission to investigate policing conduct and strategy. But the regime never conceded to this widely supported demand. In other words, increased repression reinforced the systemic threat that the political institutions could no longer shield freedoms of assembly and regulate the coercive force of the state. While the systematic threat was crucial in the early stage of mobilization, it did not fade away but was instead amplified by the repressive threat as the movement unfolded.

Networked Structure and Diversified Repertoires
Repertoires of contention are strategies shared by movement actors who resonate with such actions and find them useful. Yet repertoires can also limit options, since new consensus takes time to build. 53 William Sewell contends that the boundary of contention can be intensely contested and revised at critical junctures. 54 Hong Kong's Freedom Summer largely echoes Sewell's thesis. Scenes of violent clashes became routine. The newly adopted repertoire, including wearing protective gear, paralyzing traffic, staging sit-ins at shopping malls and making appeals to an international audience through social platforms like Twitter, was widely practised.

Source:
Kong's (2020) estimates with the authors' verification and recalculation, based on a total of 528 protests. 53 Tilly 2008. 54 Sewell 2005 With no central leadership, the total mobilization from below was sustained by highly mobile, agile and open-source networks among formal associations, informal groups and ordinary individuals. 55 The motto "be water," inspired by Bruce Lee and adopted by the Anti-ELAB Movement, was the best illustration of this networked structure. 56 It authorized an innovative repertoire to overcome the problem of tactical inertia that connective actions often face. It inspired other action protocols on digital platforms to restrain or suppress ideological differences. The conflicts between the nonviolent camp and the militant camp were left behind, with both camps taking turns in due time. The dynamics were in sharp contrast to the static occupation of physical space and the leadership struggles that characterized the Umbrella Movement and many protests worldwide in the early 2010s. 57 On 11 June 2019, Christian groups held public prayer gatherings near the government headquarters and sang "Sing Hallelujah to the Lord." Religious gatherings could be held without a permit, and they could generate a peaceful framing for the protest. In the following weeks, mothers organized unity sit-ins, students besieged police stations, opposition politicians negotiated with the police on the frontlines, medical practitioners supplied first aid to the injured, social workers

Notes:
Entries are the percentages of respondents who participated in the action. The two protests were large rallies involving participants from across the territory and thus respondents are taken as a representative sample of the Anti-ELAB Movement as a whole.
55 Costanza-Chock 2012. 56 Ting 2020. 57 Ho 2020Ting 2020. delivered care to the mentally distressed and lawyers set up hotlines to bail out the arrested. These diverse practices and services were delivered without prior planning and abundant resources; they were exercised through the pooling of individual expertise in informal networks.
According to the authors' on-site surveys at the protests on 20 October and 8 December (Figure 2), one-tenth to one-third of protesters were willing to adopt "high-risk activism," such as stopping police from advancing and engaging in flash-mob actions in shopping malls or outside police stations. While the number of arrests were a reminder of the danger of frontline actions, many young protesters persisted. Even at the frontline, the division of labour continued and may have intensified. Some wore full tactical gear, donning black clothes, goggles, gas masks and helmets to hide their identities and formed makeshift barricades within minutes. Others benefited from informants on the LIHKG Forum and thousands of public and private Telegram groups. Messages were shared about police deployment and supply chains so that protesters could adjust routes and arrange safe shelters. Despite their radical actions, the militants were not alienated from the peaceful protesters. On the afternoon of 1 September, the police besieged thousands of frontline protesters rallying at the airport. The police suspended all public transport connecting to the airport, but thousands of private vehicles rushed to the airport to rescue those left behind by the evening. Another illustrative example was the sieges of the Hong Kong Polytechnic University and Chinese University of Hong Kong campuses and the various attempts to rescue those trapped inside from 11 to 29 November.
Nearly two-thirds of the protesters resorted to "low-risk activism," such as forming human chains outside schools and posting messages on Lennon Walls in every district. Lennon Walls spread across eighteen districts and were often set up at public transport interchanges. They served as sources of alternative information and as arenas of connectivity. One of the most illustrative examples of the networked structure of the movement happened on 23 August. With little preparation, an estimated crowd of 300,000 citizens formed a 50-mile human chain along three main mass transit lines in a recreation of the Baltic Way, a peaceful political demonstration in the Soviet Union which took place on 23 August 1989. The idea was floated on LIHKG only four days before the event, but once it received majority approval, participants in Telegram groups and other social media platforms soon created maps, brainstormed routes and deployed volunteers to ensure that the chain would connect.
The protesters engaged in resource mobilization to sustain their action. Nearly half of them donated money or other materials, such as goggles, helmets, masks and food coupons. One of the most organized attempts was the creation by veteran democrats, lawyers and academics of the 612 Humanitarian Relief Fund. 58 The fund aimed to provide financial assistance for medical treatment, 58 612 Humanitarian Relief Fund 2020. psychological counselling and legal advice, and representation for injured, arrested or otherwise affected protesters. By 31 May 2020, the fund had received HK$113.15 million in donations, of which 73 per cent consisted of small donations. The fund responded to more than 14,000 requests for support and spent more than 80 per cent of its reserve funds. However, this level of support could not have been possible without the voluntary action and pro bono work of hundreds of doctors, lawyers and social workers. Resources were mobilized through voluntarism in civil society.
The movement also extended into the economic realm through the notion of the "yellow economic circle." This new repertoire aimed to counter the dominance of pro-Beijing forces in the business sector and support everyday defiance. The percentage of protesters who engaged in political consumptionboycotting pro-establishment businesses and "buycotting" (that is, increasing consumption at) pro-movement onesincreased from approximately 80 per cent in October to 99 per cent in December. Protesters shifted to this new arena when street protests became more dangerous. By creating an alternative arena in which to sustain dissent on an everyday level, political consumption formed a moral economy based on democratic and egalitarian ideals. It pressured the authorities and leveraged consumer power to prevent business owners from supporting the government vocally.

Sectoral Mobilization and Alternative Space for Contention
Space "constitutes an outcome and medium of contentious politics" and can serve as an arena for the powerless to transform existing social relations and as a means to make local grievances relevant to global constituencies. 59 Unlike previous activism in Hong Kong featuring preordained routes or occupied zones, the Freedom Summer constituted an alternative space for contention. The extension across different political geographies and into neighbourhoods, workplaces and the international arena was inspired by the "be water" motto and motivated by the concern of how to rally sectors of society and communities that usually do not participate in rallies. After the storming of the LegCo building on 1 July, public support remained high. 60 But the threats of legal prosecution and regime retribution also increased. Hence, protesters urgently needed alternative arenas and means of contention to put sustained pressure on the authorities. 61 While the diffusion into different physical and virtual spaces was not a coordinated action, it did translate into cross-sectoral mobilization. Figure 3 shows a breakdown of 528 instances of protests between March 2019 and February 2020 by protest type. From August 2019 onward, community protests, sectoral strikes and flash-mobs clearly replaced mass rallies as the 59 Sewell 2001, 55. Martin and Miller 2003. 60 Lee et al. 2021a, 2. 61 Li and Whitworth 2021 predominant protest forms. When the initial phase of the Anti-ELAB Movement came to a standstill, community mobilization offered alternative spaces and new opportunities with which to rally new constituencies. The aggregation of these grassroots, workplace and global initiatives resembled the logic of the mass rallies. On 7 July 2019, protests shifted to the Salisbury Garden area in Tsim Sha Tsui. This was the first time since 1989 that a mass rally had been held outside of Hong Kong Island. The massive turnout inspired other community rallies that connected local issues to wider political disenchantment, thereby increasing the protests' relevance to their audience. The interactions at grassroots communities helped activists and ordinary protesters to share knowledge and coordinate tactics. Based on the authors' own calculation, out of the 120 approved public processions, 65 per cent were applied for by veteran activists. These individuals negotiated with the police over protest routes, identified the required number of pickets, and shouldered the liability for wrongdoing during the protests. This know-how determined why certain applications were approved but others not before the police banned all protests.
However, these high-risk forerunners alone could not organize the community rallies. They were dependent on a decentralized yet connected network. According to the organizers of the community rallies: Once the community rally was approved, we recruited helpers on LIHKG and other public Telegram groups. It only took us one evening to hire 200 pickets. A few of them were from our inner circle. But the majority were strangers.
We pooled manpower from the well-known Telegram platforms. Then, we contributed

Figure 3: Diffusion of Community and Sectoral Mobilizations
Source: Authors' calculations.

Note:
The y-axis records the total number of protests of each type in a given month. "Mass rally" = mass rally protests; "Flash-mob" = flash-mob protests; "Community" = community protests; Sectoral = sectoral strikes.
according to our expertise in small groups. Some of us took care of publicity; others arranged logistics. Once we decided the next moves, we circulated the details back to the main groups. Our small Telegram groups continued to function after the protests, with members helping other groups to apply for a protest permit, defending Lennon Walls or organizing video screenings. 62 Sectoral mobilization further transformed the movement. In the past, only those with strong pro-democracy associational affiliations such as teachers, lawyers, social workers and university students had joined protests, but mostly in individual capacities or for sectoral interests. Yet, as early as July 2019, cross-sectoral mobilizations had proliferated to include civil servants, medical professionals, priests, accountants, artists, flight attendants, transport workers, tech workers, mothers, secondary school students and the elderly. 63 Many of them organized sit-ins or rallies on the basis of their professional ethoses or social identities. On 5 August, protesters initiated a city-wide general strike of workers, students and businesses. According to the Hong Kong Confederation of Trade Unions, more than 350,000 people from 50 industrial sectors participated in the general strike, making it the largest labour action since the Canton-Hong Kong Strike in 1925. Lacking any legal protection, strikes used to be difficult to organize in Hong Kong. But on 5 August, participants either refused to turn up for work or called in sick. Seven major assemblies were held throughout Hong Kong to rally those who participated.
The strike did not last long and two other strikes in November and January were less well participated. However, the democratic labour action on 5 August spilled over to other arenas and allowed groups and individuals to rally new supporters. First, many protesters blocked roads and disrupted public transit on that day. Their improvised actions aimed to prevent people from working and offer them an excuse to give to their employers for their absence. Two unintended consequences of the seven assemblies were instilling the notion that protests should not be confined to specific physical spaces and testing the public tolerance for disruptive actions. Second, the strike's short duration promoted discussion over organizational platforms and resources. This discussion in turn motivated many community organizers and professional workers to consider unionisation for democracy and to contest the state-corporatist structure. 64 Networked groups kept the movement lively by recognizing the contributions of others. The alternative space for contention combined civil society infrastructures and digital communication networks to facilitate a division of labour appropriating participants' varied degrees of risk tolerance, skills and personal, associational or community ties. This distributed role connected individual protesters to their sphere of contention with comparative advantages. Each participant thus held a sense of ownership and injected momentum into the movement.

Affective Ties and Group Solidarity
Emotions in protests "give ideas, ideologies, identities and even interests their power to motivate." 65 Anger, in particular, tends to "put fire in the belly and iron in the soul." 66 However, fear or anger alone is not enough to sustain mobilization. During trauma, the alignment between individual cognition and collective action often undergoes a transformative process that builds affective ties and group identification. 67 On the one hand, the assertive police actions and associated arrests, injuries and suicides of protesters were novel reminders to the public of the irregularity and disorder in society. The live broadcasting and sharing on social media of these almost daily confrontations generated a strong sense of moral outrage among citizens. On the other hand, the networked structure was effective in generating protocols and identities for the participants through which they could interpret the situation and make sense of their participation, intensifying emotions to mediate threats.
It would be mistaken to suggest that the Hong Kong government made no concessions. But the announcement of the "suspension" of the extradition bill in mid-June was considered untrustworthy by the protesters. When the formal withdrawal of the extradition bill came on 4 September 2019, state-society confrontations had greatly intensified. The 21 July "Yuen Long incident," during which civilians were randomly attacked by pro-regime groups on public transport, created a moral shock. 68 The injury sustained to a female medic's eye on 11 August, perceived to be caused by a beanbag round, added fuel to the fire. The rapid increase in civilian-police confrontations and political suicides then deepened ties through common experience.
Among the protest protocols, "climbing mountains using your own efforts" was used to justify the need for diversity in protest actions; "don't split, don't dissociate, and don't snitch on anyone" was used to contain and restrain the ideological differences among different protest camps; and "go up and down together" was used to produce a sense of collectiveness among different protesters. These themes and slogans produced affective ties between those who were once strangers but nonetheless participated in the collective actions and shared similar identities and agendas. Other sufferers are perceived to reinforce group solidarity and produce a moral obligation to sustain one's reciprocal actions. 69 Table 3 shows the emotions across different groups of protesters at a community rally on 4 August. Anger was the most common emotion among different genders, generations, classes and educational backgroundsmore than 80 per cent of respondents reported experiencing this emotion in the week preceding their interview. A greater proportion of degree-holders reported feeling anger when compared than their non-degree-holder counterparts. A greater proportion of young people felt anger, whereas fewer senior citizens felt anger. More female participants felt fear and worry than among the male participants. Overall, more than 60 per cent of respondents felt worried and only 10 per cent felt hope, but they nonetheless participated. An almost identical pattern was found in a student-initiated sit-in on 16 August.
The respondents' anger was predominantly attributed to the actions of the authorities. Yet it remains unclear whether they were worried about their own well-being, other people's suffering, or the city's future. Our further analysis revealed that their anger was triggered by police actions and that their worry was directed towards fellow protesters who were arrested or injured. A statement of inter-group solidarity, "peaceful protesters owe a debt to militant protesters," was affirmed by 79.5 per cent of respondents. A more personalized statement, "I feel guilty when I see them [the protesters] being arrested," was affirmed by 91.9 per cent. The sense of guilt caused many protesters to believe that they were not contributing enough to the movement. The affective ties developed among different groups of protesters sustained their collective actions.
Another way to unpack group solidarity is to see how collective identity unfolded. Figure 4 shows our analysis of the ranking of keywords on the LIHKG forum over time. We measured three sets of identity keywords, namely, Authors' on-site survey on 4 Aug 2020 with a sample size of 717. Entries are the percentage of respondents who reported experiencing that feeling about recent events. The X 2 values were derived by cross-tabulating the age group and gender with protesters' emotions during the protest. * p < 0.05; **p < 0.01; ***p < 0.001; #p = 0.001.
"civic-nonviolent" and "militant," "yellow ribbon" and "yellow," and "the righteous" and "the hands and feet." Overall, none of these terms attained prominence until the largest mass rallies in June. Once they emerged, the terms "civic-nonviolent" and "militant" were widely used and closely associated. They were mostly used together to maintain the narrative of "no splitting" (bugexi 不割席). Meanwhile, although "yellow ribbon" was supposed to be an identity label for all protesters, the expression was not commonly used on the forum. "Yellow" was used more widely only in the later stage of the movement, when discussions of the "yellow economic circle" (referring to pro-movement businesses) arose. Most visibly, the term shouzu (手足), literally meaning "the hands and feet," and conceptually denoting comradeship or siblingship, became the top-ranked phrase on the forum. 70 The term was often associated with frontline activists who had been injured, allegedly disappeared, or who had committed suicide during the protests. These individuals were also called "the righteous," falling somewhere between the status of crusader, vigilante and martyr. These narratives indicated that the protesters were considered one group due to their common movement experience.
Group solidarity, in this regard, mediated the participants' perception of threats following intensified repression. The shared emotions also explained why the protesters persisted in seeking justice for their comrades despite the regime's partial concessions. Ranking denotes the word's rank among the most commonly occurring keywords and key phrases on the LIHKG forum from June 2019 to 31 January 2020.
70 Discounting phrasal verbs such as "how" and "why" and key noun phrases of a particular week, "hands and feet" always ranked first on the forum from August 2019.

Conclusion
Hong Kong's Freedom Summer was an explosive and resilient mobilization that arose when the political environment was utterly unfavourable to mass mobilization. This eruption worked against the core assumptions of mainstream theories that widening political opportunity structures or abundant organizational resources are prerequisites for collective actions. The movement then continued for several months, featuring a diverse range of innovative repertoires and alternative spaces for contention without generating divisions among different protest camps and without alienating public support. Its capability to construct a collective identity and to overcome tactical inertia is exceptional in contemporary social movements.
Relying on a range of original data, this paper posits that attending to the interplay between latent informal civil society networks and threat framing is conducive to explaining the origins and the dynamics of the movement's total mobilization from below. Without the abeyance networks that helped preserve civic values and identities the ramifications of the extradition bill would not have been translated into widespread perceptions of threat, which in turn generated fear, despair and anger. Digital media platforms functioned in a manner that was interdependent with on-the-ground civil society networks, which served to communicate protest frames and enforce action protocols. A bottom-up division of labour was created, through which hundreds of thousands of people participated, enabling each of them to make sense of their contribution and simultaneously create innovative repertoires. The movement's ripple effects were evident in the opposition's electoral victory in the 2019 District Council elections, the unionisation efforts under a state-corporatist structure and the resilience of political consumption.
Theoretically, this analysis has thus presented a case of how threat perceptions can kick-start a wave of protest mobilization despite a lack of apparent political opportunities. Yet it certainly does not mean that threat perceptions can explain the size and scale of the protest movement by itself. They were a starting point, whereas the subsequent evolution of the protest movement was constituted by the relational dynamics of society-wide digital communication, social mobilization and regime responses. This paper also distinguishes between systemic and repressive threats. Notably, systemic and repressive threats may influence each other. On the one hand, the perceived systemic threat arising from the extradition bill led to protest actions, and repressive threats were encountered following assertive approach of protest policing. On the other hand, the repressive threats arising could also be understood by citizens as carrying the systemic threat of authoritarian encroachment if protest policing was not constrained.
Comparatively, this paper also examines Hong Kong's Freedom Summer to reveal the turns of protest claims and actions during democratic backsliding. The relational dynamics in networked movements exhibited a tendency to institute contradictory forces: endorsing innovative repertoires that aimed to contain internal splits while preserving reformist demands that aimed to preserve the preexisting status quo. An illustrative paradox was the "be water" and "guerrilla war" playbook transmitted from Hong Kong to the protests in Chile, Thailand and Myanmar versus the cross-sectoral support over establishing the independent commission of inquiry into police conduct. Unlike the early waves of democratization in the late 1980s and early 2000s that emerged out of liberalization and widening political opportunities, the protests in the late 2010s and early 2020s were often reactions to suddenly imposed threats to existing freedoms and institutions. Despite the radical claims and tactics, the majoritarian consensus aimed to restore institutional integrity, be it the rule of law, competitive election, parliamentary oversight or impartial investigation. These suggest the defensive nature of this wave of movements.
However, the intensity of total mobilization from below severely disrupted the once static equilibrium between government and opposition in Hong Kong's hybrid regime. 71 Although the movement's potential to spill over to mainland China had been effectively contained, its mobilizing and transformative capacity severely impaired the governing coalition in Hong Kong. In parallel, although the protesters' main demands mostly fell within the constitutional framework, the networked protest structure inevitably gave way to transgressive slogans and violent actions. The central authorities saw these developments as subversive acts challenging its sovereignty, to which any compromise would be seen as a sign of regime weakness. The movement's ideological and tactical radicalisation, along with the strength it imparted to the democratic opposition, also posed ample threats and justifications for the authorities. They provoked the higher powers in Beijing to revamp the semi-autonomous city's political landscape by imposing the National Security Law in July 2020 and instituting electoral overhaul in March 2021. This series of social re-engineering policies is likely to alter Hong Kong's state-society relations dramatically.