China's control and censorship of the internet have attracted the attention of scholars in political science, communications and area studies who are interested in how governments monitor and eliminate critical voices online.Footnote 1 Despite such a large body of research, there remain some gaps. First, research has largely focused on China's internet regulation practices at the central level, such as censorship on sensitive issues,Footnote 2 or propaganda and disinformation.Footnote 3 Cyberspace administration at the local levels, such as at the prefectural, county and lower levels, has received relatively less attention.Footnote 4 Second, partly because of the disproportionate focus on the central government, previous research on China's cyberspace administration has often perceived and portrayed the state as a Leviathan. The common impression is that a simple telephone call can result in the removal of all posts about a scandal or videos about a protest. The image of an omnipotent state may be true for the censorship machine at the central or national level, but it does not apply to the local-level governments in China.
This article adds to the literature by focusing on local-level cyberspace administration and analysing how it responds to environmental pressure in a context of limited resources.Footnote 5 Based on fieldwork and interviews at a county-level cyberspace administration office (wang xin ban 网信办, CAO hereafter) in the east of China, we find local-level CAOs to be “toothless” but “tireless” tigers. They are “toothless” in that they often lack the technology, power and manpower required to maintain adequate levels of surveillance and censorship. Nevertheless, they work tirelessly and deploy unconventional strategies that have not yet been institutionalized to navigate the challenges. By adopting practices such as outsourced surveillance, mass reporting and personnel secondment, the local CAOs manage to censor unwanted voices online.
These three coping strategies represent typical responses of bureaucrats faced with pressures from above and technological, power and resource limitations. We refer to CAOs’ coping strategies and practices as “extra-institutional governance” (EIG). Our use of this term is inspired by the criminological works on “extra-legal governance” or “extra-legal protection,” which look at how state actors resort to unofficial channels and manoeuvres to achieve governance goals and how non-state actors serve semi-official functions (for example, gangsters providing protection and social order).Footnote 6 Organizational researchers have established that institutional practices should be formal rules, with typified identities and actions.Footnote 7 Following their arguments, we define EIG as governmental actions that transgress the “identity” and “means” of the original institutional arrangement. Transgression of “identity” means that the people undertaking EIG do not have the official identity to do the job; transgression of “means” refers to the fact that the measures they take are not formally authorized. In our research, outsourcing surveillance and personnel secondment fall into the first category; mass reporting falls into the second.
We argue that EIG practices may emerge when institutional means are unavailable or inconvenient to use. EIG may also emerge when upper-level offices place too much pressure on lower-level bureaucracies or when local offices and staff members have their own interests and agendas that cannot be fulfilled within the institutional framework. As these practices are not formally institutionalized, they are often informal, unconventional, atypical and extra-legal – but not necessarily illegal. EIG is useful when bureaucrats are faced with challenges, such as shortages in their budget, headcount or time (as our case illustrates), or when they try to avoid possible moral risks or legal consequences – such as the use of thugs-for-hire in governance.Footnote 8 To sum up, EIG provides local bureaucrats with additional tools to help them respond to political or administrative pressures and circumvent obstacles.
Our study contributes to the literature on China's control and censorship of the internet by turning to the local cyberspace administration, delineating the CAOs’ daily practices, and explaining their logic and rationale. The notion of EIG can shed light on how China's censorship machine works, why the space for online expression is shrinking, and how bureaucracies respond to institutional constraints. The rise of EIG in today's China reflects a classical tension described in the organizational literature, namely that of conforming to the institutional structure versus pursuing efficiency.Footnote 9
Internet Control and the Cyberspace Administration System in China
The Chinese party-state pays close attention to cyberspace administration, as it sees the internet as an “ideological battlefield” (yishixingtai zhendi 意识形态阵地).Footnote 10 It perceives online critical expressions as “negative voices” and potential threats.Footnote 11 After all, the internet has the capacity to influence public opinion, fuel discontent and mobilize mass protests.Footnote 12 Thus, controlling the internet is an important task for the party-state. In the 1990s, when the internet was still a new phenomenon in China, the party-state was uncertain on how best to manage it.Footnote 13 Fortunately (for the regime), there were not many internet services, and access was limited to a small group of social elites – mostly young, well-educated, urban Chinese. However, China's internet users and platforms grew quickly in the 2000s, and netizens had more channels through which to express themselves. Some Chinese netizens were able to obtain overseas information, which often differs from or even contradicts the official political doctrines. The growing netizen communities and booming internet services posed new challenges for the party-state, which in turn had to adapt and evolve new responses.Footnote 14 The party-state became more experienced in controlling the internet with infrastructural projects (for example, the Great Firewall) and administrative efforts (for example, real-name registration). It now actively influences online content, maintains the domination of official ideology and mutes dissenting voices.Footnote 15
Studies on China's internet control have concentrated on censorship at the central government level, given its high visibility and significance.Footnote 16 The common understanding is that a central state apparatus crushes all critiques of the legitimacy of the CCP regime; this, however, may not reflect the full picture of censorship. First, in China, web control is decentralized instead of centralized and is mainly conducted by local-level offices rather than a national-level CAO.Footnote 17 Second, although the censorship of legitimacy challenges or scandals involving top leaders makes an eye-catching story in the news, most cyber control tasks are trivial in importance but massive in quantity.Footnote 18 Local offices often have to deal with complaints about late-night noise and neighbourhood pollution rather than national-level scandals. In recent years, some studies have focused on local censorship and web management in China. These studies are either based on textual analysis of leaked official records and emails or else based on the computational analysis of web-scraped data.Footnote 19 This line of research has illustrated the fragmented nature of the state and described the consequences of local cyberspace administration, including censored web content, the shrinking space for online expression and extended social control. However, to gain a more comprehensive picture of China's internet control, we suggest it is necessary to consider how control of China's cyberspace is conducted on a daily basis by local bureaucracies.
Institutional context of local CAOs
Managing the internet in China is too complex a task for any single agency. A number of departments and offices at both the central and local levels are involved, including the propaganda department system (also known as the Publicity Department, xuanchuan bumen 宣传部门), public security system (gong'an xitong 公安系统), state security system (guoan xitong 国安系统), Ministry of Industry and Information Technology (gongxinbu 工信部) and the National Press and Publication Administration (xinwen chubanshu 新闻出版署). In 2014, China established a new system dedicated to regulating the internet.Footnote 20 At the national level, this newly established office is called the “Cyberspace Administration of China” (guojia hulianwang xinxi bangongshi 国家互联网信息办公室, CAC hereafter).Footnote 21 The CAC is a ministry/provincial ranking office under the leadership of the State Council.Footnote 22 The CAC is both a state office and a Party office, following the norm of “one institution with two names” (yige jigou liangkuai paizi 一个机构，两块牌子). It falls under the supervision of the CCP Central Cyberspace Affairs Commission (zhongyang wangluo anquan he xinxihua weiyuanhui 中央网络安全和信息化委员会, CCAC hereafter), which is conventionally headed by the deputy head of the CCP Publicity Department.
The CAO system starts at the central level of the CAC and has branches at lower levels of government, including the provincial, prefectural and county levels. The county level is the lowest level in the formal CAO system. The CAOs at each level are subject to the dual supervision (shuangchong lingdao 双重领导) of their upper-level CAO and local Party committee (difang dangwei 地方党委), which is common practice among other government departments, such as the police. The balance of responsibility under this dual leadership system is skewed. For local CAOs, the supervision provided by upper-level CAOs is occasional and limited to professional guidance. Their daily operations, including staffing, budgets, appointments, promotions and punishments, come under the control of the corresponding CCP committee, in particular its publicity department, as illustrated by Figure 1.Footnote 23
Figure 1. Organizational Structure of the CAO System at the Central and Local Levels
This dual supervision echoes the “fragmented state” literature in China studies and organizational research.Footnote 24 Past work on the fragmented state has noted that the national and local governments may diverge in their goals and interests.Footnote 25 Similarly, the vertical and horizontal supervising offices may have different priorities.Footnote 26 Fragmentation is also evident in the cyberspace administration system. For example, there is always a tension between what the upper-level CAO orders and what the local Party committee demands. A recent study on online commentators, which was based on leaked emails from Jiangxi province, found that the local government prioritized tasks mandated by higher authorities while concurrently maintaining a close vigilance over negative reports pertaining to local issues.Footnote 27 Local governments are agents for the central government in implementing censorship, but they are more often motivated by the need to maintain a positive self-image. In contrast, the upper-level CAOs may tolerate some degree of criticism of local governments, for the sake of information and transparency.Footnote 28 Local CAOs work for their vertical, professional supervisors within the CAO system and also for their local leaders within the prefecture or county. The discrepancy between their designated tasks and their actual work agenda represents a classic principal-agent problem.Footnote 29
Local interests incentivize local CAOs to use EIG, as they work to their own agenda. Furthermore, like most Chinese bureaucracies, local CAOs and their officers face the constant pressure of evaluation and competition in a system often called the “promotion tournament model.”Footnote 30 Leaders of local CAOs and corresponding publicity departments have extra incentive to mobilize their staff to control the internet and preclude any negative voices within their territories, as any failure in the peer competition would be seen as a setback in their overall performance. Under these circumstances, local CAOs are constantly striving to improve their performance. Thus, when faced with excessive upper-level and peer pressures, as well as their own interests and limitations, local CAOs often turn to EIG rather than continue to work within institutional boundaries.
The case: the CAO in District T, City S, east China
We conducted fieldwork in a county-level CAO in District T of City S. City S is a prosperous and populous city with approximately 10 million residents. It is located in one of China's eastern coastal provinces. District T, as an urbanized and developed county within City S, has more than half a million residents, a lively economy and multiple universities. Consequently, the task of the local CAO in District T to maintain online social order and monitor unwanted voices is a tricky one. As Figure 1 shows, the county-level CAO is officially the lowest administration unit in the CAO system, leaving it with limited financial resources, personnel and power. The District T CAO has only seven formal employees and is quite understaffed. Theoretically, it falls under the supervision of upper-level CAOs, but it actually reports to the publicity department of District T's Party committee, exemplifying the imbalanced dual leadership discussed above.
After obtaining consent from two gatekeepers in District T's publicity department and the local CAO, we embarked on three months of fieldwork in the local CAO office in the summer of 2022. During the fieldwork, we observed the daily operation of the District T CAO. Detailed field notes were taken every day. Besides participant observation, we interviewed 42 current and former staff members, including those seconded to the office. The interviews were conducted towards the end of our fieldwork, after we had become familiar with the staff. We used the interviews to supplement and triangulate the data collected through observation.
Limitations in Technology and Outsourcing Surveillance
Managing the internet requires a group of professionals with IT skills. China's lower-ranked CAOs often lack the technological capabilities to conduct real-time surveillance. First, the CAO system is a relatively new agency in China's state apparatus. Many newly established local CAOs operate with minimal staff. Second, Chinese public employees predominantly have backgrounds in the humanities, arts or social sciences, rather than in STEM fields.Footnote 31 A small number of officials may have IT training, but their skills often become rusty after years of work in unrelated fields. In the field, we observed that the CAO staff members relied heavily on WeChat for communication, even sending sensitive data and classified documents via the WeChat group chat without encryption. This is partially because WeChat is free to use, user-friendly and already installed on everyone's smartphones. If persuading middle-aged bureaucrats with no IT background to switch to professional collaborative platforms, such as Teams, Slack or Alibaba DingTalk, is difficult, getting them to conduct tech-savvy tasks such as surveillance and censorship is even more challenging.
The problem of limited technology is becoming more evident as China's virtual economy continues to grow. China's thriving internet results in more platforms, users, user-generated content, critical voices and, consequently, an increased workload for local CAOs. Although District T can be considered a “small” county in east China, the web content generated every second in District T is unimaginable. Yet the District T CAO, with its seven employees, is expected to complete the following tasks: monitor more than 60 social media platforms, detect unwanted voices in various formats (for example, text, image, video, audio), collect, clean and analyse the data, and respond quickly – a “mission impossible” for a county-level CAO. For larger counties and prefectures, this challenge would only be greater.
There is another issue facing all bureaucracies: the pressure from above in an authoritarian system. Adding to the severity is the urgency: since any topic on the internet could go viral in minutes, local CAOs are expected to be on constant alert and respond immediately when an emergency occurs. The upper-level offices expect them to submit hourly reports and daily summaries of any detected unwanted voices. The worst scenario for local CAO officers is when their supervisors catch the bugs that they have missed. Consequently, CAO officers try to move fast to resolve problems before they appear on their superiors’ desks. However, it is an almost impossible task to perform effectively, especially when competing against those with more power and better resources. In addition, all CAOs are under the constant pressure of peer competition: within a province, no prefectural CAO wants to be rated as having the worst performance; within a prefecture, no county-level CAO wants this either. In response to these pressures, the District T CAO seeks external technical support by outsourcing surveillance to a private service provider, the “Public Opinion Assistant” company (yuqing zhushou 舆情助手, POA hereafter).
Coping strategy 1: outsourcing surveillance to private service providers
The POA is an internet-based software-as-a-service (SaaS) platform designed to monitor critical voices online for government and enterprise clients.Footnote 32 In the contemporary Chinese landscape, there has been a notable increase in the availability of services and products akin to those offered by the POA in response to the growing demand across government tiers. The promotional strategy used for these products is to offer what clients, often government officials, need the most. First, the POA boasts a user-friendly and foolproof user interface, catering to individuals with limited IT backgrounds. Figure 2 is a screenshot of the POA interface. Its users intuitively understand what they should pay attention to: District T CAO staff members reported that it was easy to acquaint themselves with the POA interface.
Figure 2. User Interface of POA
The POA plays a key role in assisting the local CAO to do its job. During an information session, the CAO director emphasized the importance of the POA in daily surveillance:
Every morning, the first thing you do is to initiate the POA interface and browse their pushed messages on your timeline. The job is quite easy: whenever you find a suspicious post highlighted by the POA, click it, and read it thoroughly. When you feel it needs our action, you can forward the link to your teammates to report and take it down.Footnote 33
Second, the POA provides value-added services, including generating daily, weekly and monthly reports for its clients. This feature significantly eases the workload of District T CAO officials. Reports are carefully tailored to mimic bureaucratic documents, and they are almost ready-to-go whenever upper-level CAOs request a debrief. The reports show summarized and visualized statistics of detected negative voices. They contain detailed information on trending topics and hashtags in any given period, and they can also provide policy suggestions for propaganda department leaders. In essence, the product meets the local CAOs’ technological and bureaucratic needs.
The POA, similar to many other private service providers, carefully designs its product to meet the needs of the local government and promotes it accordingly. Figure 3 is an example of the POA's pricing structure, taken from its promotional documents. It shows that the company primarily targets government clients, with its products and plans simply labelled: “District/county service pack” (quxian ban 区县版), “Prefectural service pack” (dishi ban 地市版) and “Provincial service pack” (shengji ban 省级版). Different service packs cover different numbers of authorized devices, sizes of keyword lists,Footnote 34 hours of customer services and training sessions, and levels of computational and surveillance capacity. In addition to tailoring bureaucratic reports, the POA provides other value-added services, such as emergency operations for special events or important dates that are expected to trigger political unrest online. This illustrates how the local CAOs’ lack of up-to-date technology provides an opportunity for private service providers like the POA, and how such companies tailor their products to meet the needs of their government clients and to turn a profit.
Figure 3. Price Quotations for Different Service Packs, Provided by the POA Company
After outsourcing its surveillance tasks to a private service provider, the workload of the District T CAO became more manageable. The POA acted as a radar for the local CAO, automatically detecting threats and alerting the CAO. Now, instead of waiting for orders from upper-level offices, the CAO staff could be proactive and get ahead of their tasks. While receiving tips from the upper levels is convenient, it comes at a cost, as a reliance on one's superiors signals incompetence and could lead to criticism or punishment; obtaining information from the private sector avoids such repercussions. However, discovering unwanted voices is just the first step in maintaining online social stability. The local CAOs still need to remove such posts before they go viral on social media. The next section discusses how they address and censor these captured negative voices.
Limitations in Power and Organized Mass Reporting
How do local CAOs respond to negative voices? China experts and foreign journalists tend to view the state apparatus as omnipotent in censorship. Our fieldwork tells a different story. While national-level CAOs have the capacity to control social media platforms, and the police can silence discontented citizens using coercion, local-level CAOs do not have sufficient resources or authority. Instead, local CAO staff manually report each negative voice, one by one, using the “Report” page on social media apps.
In China, all social media platforms and mobile apps must provide a “Report” button for users.Footnote 35 This button represents efforts by social media to protect individual rights from violations such as defamation, scams or spam, and to reduce threats to social stability from acts such as terrorism, extremism, racism or rumours. Figure 4 shows a typical report and feedback page. The “Report” page asks users to identify which post they are reporting and why; the “Feedback” page informs users of the outcome of the complaint. After receiving tips from users, social media administrators or automated algorithms review the complaint and decide whether to uphold or dismiss it based on the platform's terms and conditions, regulations and laws.
Figure 4. Screenshots of the User-reporting Interface and Feedback Page on TouTiao.com (Jinri toutiao 今日头条)
CAO reports often fail because social media are incentivized to be prudent. Today's social media platforms depend heavily on user-generated content. They cannot delete this content without a valid reason, or else they risk frustrating users, discouraging content generation and possibly driving users to other social media platforms. In addition, most CAO requests lack a solid basis. The CAOs often flag low-sensitivity posts. Highly sensitive content is handled by publicity departments, public security bureaus or social media platforms long before local CAOs can respond. Moreover, after years of censorship and propaganda, there are few dissenting voices remaining on social media in China. As a result, local CAOs mainly target complaints about local issues, which are numerous but less conspicuous and harder to capture. It is also harder to justify their removal, as they are often trivial and apolitical. Furthermore, accounts that frequently report others without valid reasons lose their credibility and are downweighed by social media platforms in future reporting.
We noticed this when we began our fieldwork. The CAO staff warmly welcomed us, and they seemed sincerely happy. We soon discovered why: an experienced CAO officer told us that they needed our smartphones. To be more accurate, they needed our social media accounts on different platforms (for instance, Bilibili, Kuaishou, Sina Weibo, TikTok, Zhihu) to help them report negative voices. Our social media accounts were valuable, as they were “fresh accounts.” Accounts operated by CAO staff are used extensively for reporting posts, and social media platforms consider this as an abuse of the reporting function. Consequently, their accounts are downweighed and their reports are taken less seriously. Fresh accounts – such as ours – are more effective when mass reporting.
This shocked us – it was surprising to discover that local CAO workers were using their own personal social media accounts and cellphones to report unwanted voices, just like ordinary users. This was not what we had imagined before entering the field and not how the media portrays China's censorship machine. It was also shocking to learn that social media platforms would downweigh CAO workers’ accounts if they overused the reporting function. To be fair, these platforms do not know the accounts belong to CAO officials, and that is precisely what the CAO wants. The CAO staff members choose to look and act like civilians; they do not use official power and, indeed, do not have any. In this sense, local CAOs are not powerful at all, unlike their headquarters in Beijing, the national-level CAC. The CAC is an impactful agency; its former head, Mr Lu Wei, was once called China's “internet czar.” However, the CAC's authority and influence do not trickle down to offices at the prefectural and county levels. Upper-level offices in Beijing and provincial capitals have direct leverage over social media giants.Footnote 36 County-level CAOs, however, have no authority to demand that companies block an account or a video.
Despite their lack of power, the local CAOs are still expected to mute unwanted voices before they spread.Footnote 37 When local scandals occur, the CAOs in the corresponding county or prefectural city must seek help from upper-level CAOs or publicity departments to curb the flow of information. The director of the District T CAO explained the power limits as follows:
We cannot compel media companies to remove negative information directly. All we can do is follow the municipal leaders, visit the provincial publicity departments and upper CAOs regularly, and build some connections for future use. My primary duty as a CAO director is to maintain these ties with upper offices and social media companies. You won't have access to the social media CEOs or administrators at People.com.cn 人民网, iFeng.com 凤凰网, Sina Weibo 新浪微博, NetEase 网易, XinhuaNet.com 新华网, ThePaper.cn 澎湃新闻 and TouTiao.com 今日头条 unless you visit the provincial capital or Beijing.Footnote 38
As this comment illustrates, local CAOs do establish connections with social media companies, but they cannot use those connections frequently. Instead, they reserve them for critical situations. CAO officials know they cannot make hundreds of telephone calls a day to more than 60 companies just to block humorous video clips about a teenager smoking at school. Their upper-level offices will not authorize these daily tasks, as most of the content does not threaten national interests or social stability. However, local governments want to remove posts that expose uncivilized behaviour or tarnish the public image of their county or city. As these demands cannot go through the conventional, official channels, local CAOs attempt to resolve these issues instead.
However, as mentioned, local CAOs have a limited toolkit. They cannot squander their useful ties with social media companies on trivial matters and they do not want to appear incompetent in front of their superiors, so they try not to ask for help from higher-level offices. Unlike law enforcement agencies, such as the police or national security agents, local CAO officials are not authorized to use detention or arrest to silence citizens. Instead, they resort to reporting negative voices in the manner of any regular user. However, their frequent reporting leads to downweighed accounts and little success.Footnote 39 So how does a toothless tiger bite? The next section describes the strategy used by the District T CAO: “organized mass reporting.”
Coping strategy 2: organized mass reporting
In response to the failed reports and downweighed accounts, the local CAO we studied deployed a new strategy: “mass reporting.” This involves mobilizing dozens or sometimes hundreds of accounts to report a social media post. To ensure the successful removal of a post, the CAO members often coordinate their reporting actions to occur within a short time frame. Their connections with social media platforms mean that mass reporting is effective. Most social media platforms use certain algorithms to determine which posts to take down; posts reported by multiple users within a short time alert the algorithms. The posts are then taken down either automatically or by human arbiters.
The District T CAO established a WeChat group to coordinate the mass reporting. It comprised nearly 100 members and was nicknamed the “Cavalry Team.” Some members were CAO formal employees, but most were temporarily seconded workers (discussed further below). The group earned its nickname for its manoeuvrability – it would be on stand-by, move fast and resolve problems upon request. The team members constantly exchanged information about newly captured negative voices through WeChat group conversations. Figure 5 shows a screenshot of a typical conversation in the Cavalry Team WeChat group. Members used the chat to coordinate mass reporting, evaluate the performance of each member, debrief leaders on outcomes and provide a daily summary.
Figure 5. Screenshots of a Typical Conversation in the Cavalry Team WeChat Group
After receiving tips about unwanted voices, Cavalry Team members would report the posts in a coordinated fashion. Sometimes, they called upon their friends and families for help, as the more people reporting a post meant a higher likelihood that the post would be removed. In an interview, Wang, a CAO officer, even told us that if enough people acted simultaneously, the justification or reason for reporting the post did not matter:
You do not have to carefully pick a reason for reporting [a post]. You can select any option. Nobody in the social media administrative team cares. You can just click on “false information,” “defamation,” or whatever you like. If we flag these posts a massive amount in a short time window, these social media companies will skip the review process and directly remove them. Nobody wants the bother or to get into trouble, right? But you may want to write something long in the box on the reporting page – just filling in accusations such as “threatening social stability” or so. Based on our experience, long complaints are more likely to succeed. After the post is removed, please confirm, and debrief the office with a screenshot of the feedback page for records.Footnote 40
In addition to organizing mass reporting and “writing longer complaints,” the District T CAO had other tips for newly recruited Cavalry Team members. These included “reverse-engineering” the algorithms of social media companies by, for example, using new accounts or different devices, visiting from different IP addresses, or reporting with high intensity within five to ten minutes of the offending material being posted. Using such tactics helped the District T CAO's organized mass reporting to succeed: according to a weekly working summary, the CAO team targeted and reported 301 different social media posts (based on our field experience, this means several thousand user attempts to report a post), and 286 were removed, yielding a success rate of 95.01 per cent. Such high success rates are typical according to our field observations, explaining why social media users in China have felt more confined in recent years.
Limitations in Manpower and Personnel Secondment
It takes more than technology and power to censor the internet; the local CAOs need foot soldiers to flag up and report negative posts, to communicate with social media platforms and other offices, and to do the daily tasks required by the bureaucracy. However, there were only seven formal employees at the District T CAO. In an interview, Xing, a District T CAO officer, complained about the workload generated by a single scandal at a local hospital:
This scandal generated 5,226 critical messages within 24 hours. There were 124 KOLs [key opinion leaders] retweeting it and, later, more than 20 newspapers followed up by reporting on it. When it goes viral that fast, there is nothing we can do about it. That is why we always prefer muting it before it gets any attention.Footnote 41
The shortage of personnel presents a real challenge for CAOs as they attempt to keep the lid on scandals and information in the age of internet and smartphones. The nature of social media requires CAO members to be on constant alert as any topic could go viral within seconds. CAOs receive urgent requests every day, and they must respond within hours or even minutes. During our fieldwork, we observed how the District T CAO managed a tough situation when many netizens from various platforms poked fun at the new mayor of City S, particularly mocking his taste in clothing. The head of the publicity department of City S telephoned the District T CAO and demanded immediate action. The director of the District T CAO was very apologetic to his leader on the phone. Immediately after the call ended, he messaged all members and asked them to report the negative posts, summarize the incident statistics and debrief the higher-level office every hour:
Now, let's all put away all other tasks for a while and do our best to monitor the negative content about our mayor. Let's also monitor the statistics and categorize them based on where such content is posted, and what type of negative information [the posts] contain. Give me those numbers in half an hour. Tomorrow, I want an event summary on my desk about our progress in removing this content.Footnote 42
As this example suggests, the District T CAO officers are under constant pressure, especially when negative voices target their direct leaders in District T and City S. Rumours linked to local leaders are given top priority and are sometimes handled with more urgency than scandals involving provincial and national leaders. The story also illustrates the fragmented nature of local governments: they often act in accordance with their own best interests, rather than pursue their designated goals (for example, “maintaining online social order”), as specified by the centre and formal rules. Finally, the example offers an explanation as to why local CAOs cannot and will not use formal channels to manage such negative voices: they are the only ones who perceive these voices to be negative. Neither the CAC nor the social media platform cares about the public image of a local official.
The daily routine and bureaucratic burdens exacerbate an already problematic shortage of personnel. China's bureaucratic machinery is notorious for its complex, time-consuming and strict procedures and workload. Public employees are confronted with daily tasks such as liaising with many different agencies and drafting a variety of documents, including meeting summaries, working logs, reports and speeches for leaders. In addition to these duties, the seven employees at the District T CAO were tasked with censoring the critical voices on social media platforms in District T, a populous county with half a million residents. All the CAO workers we spoke to complained about the shortage of staff and incessant workload. Some middle-aged staff had started to slacken off, or “lie flat” (tang ping 躺平), once they had lost any hope of promotion, leaving a bigger workload for their younger CAO colleagues, who were still hoping for career advancement.
Although the local CAO had insufficient personnel to manage its workload, it was unable to hire additional staff. In China, all bureaucracies are tightly regulated in terms of budget and staff count. This is known as the “headcount cap” under the nomenklatura, or bianzhi 编制 system.Footnote 43 To combat the problem of insufficient labour and circumvent the headcount limit, the District T CAO used its informal power to mobilize public employees from other offices, taking advantage of the “personnel secondment” system.
Coping strategy 3: personnel secondment from other government offices
As already mentioned, the District T CAO only had seven formal employees, but it had a Cavalry Team of around 100 members. These additional members were all temporarily summoned to the group through an informal personnel resource reallocation system called jie diao 借调, or personnel secondment, which is a flexible arrangement used by organizations with labour shortages.Footnote 44 Generally, across China, local governments and their departments are delegated an enormous workload by upper-level offices but are given limited power and resources to complete their tasks.Footnote 45 They compensate for this lack by seconding staff members from other, less busy and less influential lower or same-level offices.
During our fieldwork, we noticed that the seconded individuals were typically young civil servants, who were usually more motivated and more familiar with social media. They often came from subdistrict or street offices (jie dao 街道) or other lower-level government or Party offices. The local CAO cannot mobilize them directly, but it can request the local CCP committee (especially the publicity department) leaders to authorize their mobilization. In other words, the CAOs themselves are not intrinsically powerful; however, when local leaders value online social stability, the CAOs can be temporarily empowered. During periods of heightened sensitivity when there is a need for increased cyber vigilance, such as during the Covid-19 pandemic (2020–2022), the CAOs are able to turn to personnel secondments from lower tiers to help address online critical voices.
As well as alleviating staff shortages, personnel secondment also facilitates inter-office collaboration and coordination. Many of the local CAOs’ duties demand a collective effort by various agencies and departments (for example, public security bureaus, educational systems, environment protection forces). If a social issue consistently triggers online criticism, the local CAO might have to tackle it offline, either by persuading the individual to keep silent or by genuinely resolving the issue. Such issues could include complaints about water pollution in a local river, late-night noise in a neighbourhood or a labour conflict in a nearby factory. In these instances, the local CAO can take advantage of the connections a seconded staff member might bring from another department. Thus, personal secondment offers the possibility of alleviating two problems simultaneously – staff shortages and limited capacity.
Although personnel secondment can help the local CAO to resolve such issues, it entails a degree of exploitation and alienation. We observed that, unlike the CAO staff, the seconded individuals were not getting paid. They had their own jobs at other agencies (for example, street offices, education bureaus) and yet had to find additional time to work towards achieving censorship targets. They were always on standby on WeChat, including during evenings and weekends. Sometimes, when an emergency occurred, the CAO would pressure its staff and seconded members to mobilize their families, relatives, friends and acquaintances to file complaints and censor the negative voices together. Moreover, the social media account usernames and passwords of the staff and mobilized individuals were saved in an Excel file, which was circulated within the CAO WeChat group – again, unencrypted – so that the accounts could be used without the consent of the individuals when they were off work. Such practices violate privacy rights, property rights, information security, and more. And while they are a convenient means for the bureaucracy to govern the internet, they are an imposition on the individuals involved, not to mention the people being governed.
Discussion and Conclusion
We sought to understand China's censorship machine by turning to lower-level governments and examining how a county-level CAO in east China operates on a daily basis. We found that the local CAO officials have to go above and beyond their remit and their capacity. They are constantly faced with insufficient technology, power and staff, and the urgency of internet regulation increases the pressure on them to act quickly, thus rendering them toothless tigers. In comparison to police officers, who can detain citizens or demand that social media companies enforce regulations, local CAO officers have little power. However, the District T CAO we studied responded well to these challenges and accomplished more than its capacity allowed. The CAO staff members were not constrained by the power and resources that were officially granted to them; instead, they instigated practices that were not institutionalized, recognized or authorized – what we call extra-institutional governance (EIG).
According to our definition, EIG oversteps institutional boundaries mainly around two issues: identity and means. The former involves people external to the government (in this case, outsourcing surveillance) or government employees from outside the office of interest (personnel secondment). The latter refers to using means other than those formally authorized (mass reporting). Although the EIG practices of local bureaucracies transgress institutional rules, the upper-level offices acquiesce because of the high priority given to controlling the internet by the party-state. That being said, the higher-level authorities have neither formally endorsed nor institutionalized such practices.
An increasing number of studies have remarked on the existence of EIG in contemporary Chinese society. Scholars have noted that practices similar to those described here are being widely adopted throughout China and becoming implicit norms or unwritten rules (qian guize 潜规则). These EIG practices include hiring thugs for demolition tasks, employing contract-based auxiliary police officers in daily social control, mobilizing students to serve as online commentators, and so on.Footnote 46 Both the central and the local governments have incentives to tolerate but not institutionalize EIG: the central government needs tasks to be completed with the minimum headcount and budget; the local governments want the flexibility and effectiveness of EIG to get around institutional, legal and ethical constraints. Moreover, since the EIG practices can be immoral and hard to justify, all governments want the option to deny and avoid consequences when something goes wrong. In other words, both the local offices and the upper-level offices need these practices to be extra-institutional.
The concepts of institutional governance and extra-institutional governance are at two ends of the governance continuum. More often than not, governance practices fall somewhere in between these extremes. As organizational research has noted, in any organization, there is a constant tension between institutional conformity and efficiency. Bureaucrats and offices all have the incentive and agency to ignore the boundaries and achieve their own goals.Footnote 47 The case of District T CAO reinforces this line of research. Furthermore, since the two belong on a continuum, EIG and institutional governance may develop more links over time, and sometimes evolve into one another. In recent years, some EIG practices have become institutionalized. For example, local governments have officially announced and formally acknowledged the existence of “voluntary online commentators.”Footnote 48 The possibility of institutionalization may not weaken EIG. In fact, it could even encourage local governments’ use of EIG as they may initiate a practice as a fait accompli expecting it to be endorsed by the upper-level governments one day.
EIG does help local CAOs to “succeed” in their daily operations. Unfortunately, such “success” may not be good news for Chinese citizens. Local CAOs’ efforts reduce the space for online expression. Ironically, their short-term success is a burden in the long run. Since the CAOs have been doing well in censorship, there are fewer posts that can be easily identified as negative voices. The remaining critical voices, if any, are less evident and harder to capture.Footnote 49 The local CAOs are now chasing after posts that are neither “radical” nor “sensitive.” The bureaucratic concern about “false negatives” (real dangers slip away) has led to a tremendous number of “false positives” in censorship, whereby non-sensitive posts are deleted and non-dissidents are punished. Today, the local CAOs are even reporting TikTok videos about teenagers smoking at schools or newlyweds complaining about the price of real estate.Footnote 50 At the District T CAO, these absurd tasks took up most of the staff's time. We therefore conclude that the censorship machine creates a problem for everyone. Furthermore, such high-maintenance practices may no longer be viable when local governments are tightening their belts and staff members are “lying flat.” Future works may follow up with the changes in local government practices as the fiscal crisis escalates.
Our findings add to the understanding of how China's censorship machine works on a daily basis; they also shed light on the general interpretations of how bureaucratic agencies respond to environmental constraints. Therefore, this study contributes to the research on local governance and cyber control in China and to the organizational research in general. There are several directions that can be considered for future research. First, future research could take a more comprehensive and comparative scope. As we studied a CAO in an average-sized district/county in east China, our findings are likely typical for most mid-tier Chinese cities and counties. Future researchers may want to consider different scenarios, such as rural China where the CAOs’ resources are even more limited, and metropolitan areas with highly developed IT sectors and social media giants (for instance, Beijing, Shanghai, Shenzhen, Hangzhou). The working mechanisms of local CAOs could be different given the different constraints faced by different local governments.
Second, we focused on a local CAO. The relationship between local CAOs and other government departments could be a topic of future research. For instance, researchers could examine how local CAOs work with the police or upper-level CAOs. Third, we explored how the local CAO worked as an institution. Others could explore the experiences of the people who work in the institution. How do they perceive their censorship work, and how do they cope with or deal with the moral dilemmas, if any, arising from their work? We believe the pursuit of this line of research could enrich our understanding of China's censorship machine and, more broadly, the complex nature of the bureaucratic organizations.
Acknowledgements
The three authors contributed equally to the present article and the family names follow a reverse alphabetical order. The authors are grateful to Professor Paul Rock and members of the weekly “Sociology and Criminology” workshop at the University of Macau, three anonymous referees and the editors at The China Quarterly for their helpful comments. All errors remain our own. This study was funded by two multi-year research grants provided by the Faculty of Social Sciences, University of Macau (Grant Number: MYRG2022-00085-FSS and MYRG2022-00199-FSS).
Conflicts of interest
None.
Tony Huiquan ZHANG is an assistant professor at the department of sociology at the University of Macau. Dr Zhang received his PhD in sociology from the University of Toronto (2018). His research focuses on political culture, political communication, diversities and inequality, and politics in China/East Asia. His works have appeared in the British Journal of Sociology, Journal of Occupational Health Psychology, Chinese Sociological Review and Social Indicators Research, among others.
Jianhua XU is an associate professor at the department of sociology at the University of Macau (UM). He is also an adjunct associate professor at the Center for Macau Studies, UM, an affiliate at the Institute of Advanced Studies in Humanities and Social Sciences (IAS), UM, and an honorary fellow in the Center for Criminology at the University of Hong Kong. His research interests include sociology of crime and deviance, policing, victimology, urban sociology, and Macau studies.
Jinjin LIU is a PhD candidate in the department of sociology at the University of Macau. Her research interests include public opinion, social media, organizational studies, and public administration in China. Her works have appeared in peer-reviewed journals such as the Journal of Women and Aging and the Canadian Journal of Criminology and Criminal Justice.
Open access
