On July 16th, 1990, the front page of the Fuzhou Evening News carried a poem by Xi Jinping, then head of a local party committee. The subject of the poem was Jiao Yulu, an official who had worked himself to death in the drought-stricken Henan countryside in the 1960s.
Jiao was far from that decade’s only political martyr. Plenty of party cadres died with their boots on during the Great Leap Forward and its aftermath. But Jiao Yulu was celebrated for attributes beyond idealistic commitment to the country: he was a pragmatist and a risk-taker, and he could operate independently of party bosses. He implemented an innovative paulownia tree-planting campaign to fight desertification that he arrived at through on-the-ground experience and by tapping local expertise, not from a command on high. This made him a model party member for Xi to memorialize during the Reform and Opening Up era, when individual politicians and bureaucrats were empowered to run local authorities as they saw fit.
The most significant reform carried out in China after 1978 was one of systematic decentralization. Politics at the top could grind to a halt thanks to factional and ideological battles. But, down below, local leaders had the authority and resources to make their own decisions. Neither a clear constitutional framework, nor an unambiguous basic law governed Deng Xiaoping’s experiment. Instead, he drew on a principle first established by Mao during the early years of the People’s Republic of China:
Our territory is so vast, our population is so large and the conditions are so complex that it is far better to have the initiative come from both the central and the local authorities than from one source alone.
The Two Initiatives
When Mao and the Communists took power in 1949, the first requirement had been unification. The previous century had seen various powers struggling to do the same. The ailing Qing dynasty had struggled with projecting imperial power out to the periphery. The nationalist prophet Sun Yat-sen and his revolutionaries managed to topple the old regime but found themselves outfoxed by the warlord Yuan Shikai and his Beiyang government, China’s first republican regime. General Chiang Kai-shek, Sun’s successor, ultimately pushed aside the Beiyang government but found himself contesting territory with warlords, Communists, and Japanese invaders.
By the time that Mao and the Communists themselves defeated the Nationalists, they had seen multiple successive regimes try and fail to unify China’s expansive territories. In order to knit together a country, the Chairman and his lieutenants placed the new People’s Republic of China under six administrative regions, each closely overseen by the central government. Their aggressive centralization of power allowed them to rapidly carry out policies like land reform.
But by the middle of the 1950s, the limits of this approach were obvious. The power of the administrative regions could threaten the center. These massive regions required competent leadership. Looking at the names of their leaders—men like Liu Shaoqi, Lin Biao, Peng Dehuai, Gao Gang, Rao Shushi, and Deng Xiaoping—we see all of Mao’s rivals and potential successors. Allowing those men to build a power base was risky. By 1954, both Gao Gang and Rao Shushi were purged. Gao killed himself that same year, and Rao died in prison in 1975.
The potential for these regions to become bases for opposition was not their only problem. Running policy through six administrative regions was just not efficient. A single policy from the top would need implementation by administrative heads that were overseeing sometimes vastly different territories. What worked in tropical Hainan or fertile Jiangsu could not always be applied in landlocked Qinghai and or arid Shanxi. Given their immense size, conditions between localities could differ greatly, even within a single administrative region.
Apart from climate, there were social cleavages between and within administrative regions. The Northeast of China, for example, might appear fairly demographically homogeneous, but its population included groups as different as the sizable urban proletariat and the herders living on the Soviet border, dozens of distinct ethnic groups, adherents of multiple religions, and speakers of many languages. The party might resort to violent excesses in pursuit of goals like land reform, but relying solely on sheer brute force was not feasible in the long term.
In response, Mao became increasingly critical of the centralized Soviet approach and began to move away from it. Taking policy initiatives from central and local authorities was the solution. Under what became known as the Two Initiatives approach, local governments would have the autonomy to introduce their own policy, and would have affordances in how they interpreted central policy.
This idea of Two Initiatives originates in Mao’s key philosophical works of 1936 and 1937, “On Contradiction” and “On Practice.” His interpretation of dialectics in the former work differentiates between “antagonistic contradiction”—the real, irreconcilable conflicts based on competing material forces—and “non-antagonistic contradiction,” which can be resolved in a new synthesis. The latter work emphasizes the importance of praxis over theory. A year after the speech that gave us the Two Initiatives, Mao returned to this core philosophy in “On the Correct Handling of Contradictions Among the People” in 1957, stating that non-antagonistic contradictions might remain between the “between the leadership and the led,” but that they could be resolved through struggle.
Individuals might be the elementary particles of society, but in Maoist theory, they came together to produce the mass line: the collective popular will which the party had to stay close to and learn from. Mao pointed to the millions of party cadres as representative voices of the mass line. Local authorities were considered qualified to interpret the mass line. Their decisions would be fine-tuned through theoretical debate and tested ultimately by practice.
The participatory nature of Maoist horizontality embodied Mao’s own criticism of the verticality of the Soviet command economy and of what he saw as Stalin’s distrust in the mass line. “Those bureaucrats who are afraid of great democracy,” Mao warned, “must study Marxism hard and mend their ways.” Democracy in this sense had less to do with voting and more with active involvement in the construction of a new Chinese society.
But Mao’s criticisms of centralization were mostly a theoretical response. The Great Helmsman, as his cult of personality dubbed him, had a habit of steering into rough seas. The Two Initiatives came during a brief period when the party was relaxing its control. But, during the collectivization drive of the Great Leap Forward between 1958 and 1962, guidance on agricultural collectivization and rural industrialization again came from the top. Even if there was theoretically more autonomy for local authorities under what Mao began calling democratic centralism, the loyalty of cadres was paramount and enforced with violence. There was no formal devolution of powers to the localities, even though they remained in control of agricultural land, industrial capital, and labor.
Under a reformist line proposed by Liu Shaoqi in 1962, there was again a brief flowering of local experimentation and Maoist horizontality—and this is when Jiao Yulu went to work in Henan. But all of this was quickly repudiated in the Socialist Education Movement of 1963, which led directly into the Cultural Revolution, launched in 1966. It’s likely for the best that the risk-taker and pragmatist Jiao Yulu died two years shy of the Cultural Revolution. Veneration of him as a martyr ceased once it began, his handpicked successor was removed from office, and radicals tormented his family and desecrated his tomb.
Whatever Mao’s theoretical musings, the winning strategy was what senior official Ke Qingshi described in 1959 as “the whole nation as one coordinated chess game.” Experimentation would have to wait.
When the smoke cleared and Deng Xiaoping inherited the leadership from Hua Guofeng in 1978, he was in a position not unlike Mao Zedong in 1956: even before the factional leaders of the Cultural Revolution were arrested, the country had been stabilizing. But it was also starting to pull in different directions.
Deng Xiaoping had watched Mao’s successor Hua Guofeng, as well as his allies in the oil-focused Petro Clique, fail at an ambitious top-down scheme to develop heavy industry. Meanwhile, there was an increase in the speed and scale of unauthorized experimentation that had been taking place in the wake of political disorder at the top.
Deng seized on the Two Initiatives. Maoist horizontality had its advantages for Deng, even as he abandoned concepts like the mass line and the dictatorship of the proletariat.
Unlike other decentralized federal systems, there was no formal constitutional or legal framework to govern the Two Initiatives approach. Party-state theoreticians like Jiang Shigong defend Mao’s introduction of the Two Initiatives in his 1956 speech as an interpretation of the 1954 Constitution. But it’s an ambiguous, possibly contradictory interpretation. “According to our Constitution,” Mao concedes, “the legislative powers are all vested in the central authorities.” However, he advises, “local authorities may work out rules, regulations, and measures in the light of their specific conditions and the needs of their work.” But the lack of a binding legal framework was functionally unimportant. In practice, it was the practices and proclamations of the party—what thinkers like Jiang today call the “unwritten constitution”—which decided the norms of government, especially those of the party leadership.
Deng wrote Mao’s Two Initiatives into the 1982 Constitution, but there is no precise statement about the division of powers. The ambiguity of the written constitution and the power of the unwritten one meant that very little was expressly forbidden.
Letting the localities take the lead was a way for Deng Xiaoping to avoid conservative opposition at the top, circumvent powerful factions linked to the state sector, and avoid committing to any singular model for reform. The fact that localities collected the bulk of taxes also meant that it was a cheap solution for the central government. Reform and Opening Up was led from below.
The Cultural Revolution even ended up having a silver lining for Deng’s reforms. Against party wishes, it had provided several years of very much undirected decentralization. Mixed messages from the top and the paralysis of local politics undermined the years of aggressive collectivization. Local cadres in Guangdong stood idly by as markets flourished in Qingyuan and Puning. A report from the provincial archives in 1975 showed that Guangzhou was receiving hundreds of thousands of packages from abroad a year, full of clothes, cooking oil, and medicine, ready to be shifted into the black market. There was no need to plan for private entrepreneurship, only to manage it.
While all land was still legally administered by the collective, farmers had begun secretly dividing it up amongst themselves. They grew what they needed to fulfill government quotas and they were free to dispose of the surplus as they saw fit. This system of contracting plots of land to individual farmers got central government support in 1979, but it took a while to spread. Contracts between households and the state gained support in 1980, and the household contract responsibility system was well-established by 1985.
In Jiangsu, Anhui, Zhejiang, and Guangdong in particular, commune and brigade enterprises—companies controlled by collectives, usually engaged in light industry—had been thriving since the Cultural Revolution. The landmark 1978 Third Plenary Session of the Eleventh Central Committee of the Party provided political justification for these local experiments. In the 1980s, their ownership passed from communes to become the property of townships and villages. This model spread quickly. Light industry rapidly developed in the Yangtze River Delta and Pearl River Delta.
The eventual and gradual privatization of township and village enterprises also began as a local experiment. Ideas came from managers and cadres on the periphery, rather than from the party leadership. And instead of censure, those coming up with them were reaping rewards.
Through the 1990s and 2000s, the private sector began to grow and the power of the state began to shrink. Local experimentation was crucial to this process. Local appointees became distributed ganglia. The arm could move without the brain giving a command. They were not simply interpreting the mass line and carrying out the dictatorship of the proletariat but securing the multivalent interests of individuals, grassroots organizations, enterprises, and the party itself. Deng had adapted Maoist horizontality to a newly depoliticized age.
Deng’s incorporation of market logic into this devolution didn’t lead him to give up the party’s monopoly on cadre promotions. While enterprise managers and private entrepreneurs could now exercise some leverage over local authorities, the market mechanism was still ideally subservient to the power of the state. The party remained the backbone of the system and had the power to appoint and remove cadres, as well as promote and demote them. Economic incentives were secondary, at least in design. No matter how much money a cadre made for himself, for local enterprises, or for influential individuals in their jurisdiction, promotion was only possible through the party.
The system existed to test those within it. Posting party officials to the provinces made the best use of their creativity and intelligence. Rather than pushing paper in the capital, the future leaders of the nation were more likely to be toiling in party posts in underperforming regions. Despite being the heir to the political legacy of Xi Zhongxun, revolutionary hero and originator of the Special Economic Zones, even Xi Jinping was compelled to vacate a cushy position in the capital to prove himself in the provinces. The story of how he turned Zhengding into a “testing ground for a slew of bold initiatives” has become part of his own carefully-crafted political legend.
Policy experiments created both elite and grassroots knowledge. Experiments continued even if they did not accord exactly with central government policy. With the party looking for policy alternatives, any intelligence and practical experience were highly valued. Local elections conducted in Sichuan in the 1990s and Jiangsu in the 2000s were not widely replicated, but they provided the foundation for electoral experiments carried out by Guangdong party boss Wang Yang to disarm a lengthy dispute over land seizures in Wukan in 2011.
Even if the central government did not take on board policy innovations, those who came up with them could still be rewarded. The central government has not expressed much interest in holding elections, local or otherwise, but Wang Yang was promoted to Vice Premier a year after he oversaw a vote in Wukan. Wang pushed mid-2000s Chinese liberalism to its logical conclusion, pruning the state to make space for civil society and the private sector. Despite Xi Jinping’s statist turn, he still raised Wang Yang to the Politburo Standing Committee in 2017.
The ideal experiment is one that can be modified, scaled, and repeated. The downfall of Bo Xilai’s populist campaign in Chongqing has many roots, but one of them was that his experiments were based on personalized authority and risky bank loans. Whether or not it fits the tastes of Xi Jinping, Wang Yang’s Guangdong Model can be safely replicated, while attempting to repeat the Chongqing Model would risk political and financial instability.
Bo Xilai’s bank loans also pointed to a major vulnerability of the scheme, which is that experimentation had to be mostly self-financed. The central government was happy to turn a blind eye to practices like local governments selling state-owned land out from under residents. This helped pay for the functions of local authorities but could also lead to corruption and reaching out to organized crime for the necessary muscle.
Even though Bo Xilai was running a huge municipality directly under the administration of the central government, the issues in his leadership were only uncovered when the party’s Central Commission for Discipline Inspection began investigating Wang Lijun, Chongqing’s second-in-command. One defect of the authoritarian meritocracy is that vertical decentralization reduces the ability of the state to supervise its workers. Bo Xilai’s gaming of the system with financing backed by risky loans occurred many times over at lower levels of government.
Factional politics also negatively impact the system. When factional politics are introduced, local cadres can cheat by making use of intra-party connections to increase growth or hit other performance goals; the right relationships can conceal poor leadership or even corruption.
The Dengist horizontal strategy had helped to make China rich. Its flaws and abuses paved the way for another vertical, centralizing reaction.
The End of the Two Initiatives
Xi Jinping’s centralization of power since 2011 has been a response to the rise of factional politics in the wake of Deng Xiaoping’s death. Hu Jintao was never able to gain a firm hold of the party-state, root out powerful factions, and centralize power. Xi, in turn, made it a priority to secure a strong personal leadership position.
Dengist horizontality is giving way to a new model of verticality. The party under Xi has not explicitly repudiated Deng Xiaoping’s interpretation of the Two Initiatives, but the Great Leap Forward-era theory of “the whole nation as one coordinated chess game” has been enthusiastically revived.
Local experimentation has slowed. The party’s Central Commission for Discipline Inspection has begun targeting local officials for “irresponsibility, lack of achievement, irresponsible conduct, and faking results,” suggesting a lack of hunger for further innovation. This inspection drive is about cleaning up the kind of corruption that flourished under Jiang Zemin and Hu Jintao, but it is also directed at “slack governance:” risk-averse local politicians just going through the motions. Fearing censure in the new inspections, they jumped into action and subsequently had to be warned against engaging in hasty experiments.
Another key drive by Xi Jinping has been to take the ambiguity out of the legal and constitutional framework governing local experimentation. In January of this year, the Party Central Committee issued their “Plan on Building the Rule of Law in China.” It is part of a larger scheme to diminish the role of the unwritten constitution in daily life.”
This presents an immediate problem for the Dengist style of horizontal administration. The strictest interpretation of even the reformist 1982 Constitution could have seen Special Economic Zones deemed illegal. The dividing up of land among individual tenants could be ruled to be unconstitutional, too. The commentators at state newspapers decrying the 1998 Sichuan experiments in township elections as unconstitutional were likely dismayed that the central government did not step in. These experiments were allowed to continue because of the haziness of the Two Initiatives and the overall constitutional and legal framework governing China.
The move towards tightening the rule of law now looks to get China caught up in one of the more successful fads to come out of the world of Western policy wonks: digital governance. Building the capacity of so-called “fourth industrial revolution” technologies, including artificial intelligence, is one of the key tenets of Xi Jinping Thought on Rule of Law, which states that China should:
Make full use of big data, cloud computing, artificial intelligence and other modern technological means to comprehensively build a ‘smart rule of law’, and promote the digitalisation, networkisation, and intelligentisation of the rule of law in China. Optimize and integrate various information, data, and network platforms in the field of rule of law, and promote the construction of the nationwide rule of law informatization project.
The plans for AI-assisted judges would go nowhere without a waterproof legal and constitutional framework. The Shanghai AI Assistive Program for Criminal Cases and pilot programs in other centers are test runs for turning more significant constitutional cases over to software.
The clear interest among the party leadership in the promises of big data, surveillance, and artificial intelligence is one of the reasons that they have de-emphasized local innovation.
The expertise, money, and private sector connections required for high-tech innovation mean that larger subnational governments have outshone the periphery. The Tianjin AI pilot zone and the Suzhou municipal government’s combination of pandemic tracking and social credit management in a mobile application are good examples. It’s not coincidental that Alibaba’s City Brain was rolled out in Hangzhou, where the company has its headquarters.
Smaller centers are mostly imitating the models of Internet Plus regulation and developing “social credit cities.” Yantai in Shandong was praised by the National Development and Reform Commission at the Summit for Credit System Construction of Chinese Cities, but it’s unclear what they have done that sets them apart from larger centers. A recent report from Rongcheng in Shandong reveals the details. The idea of issuing citizens with an “integrity identity” that helps administer a scheme to give them local government benefits (like waiving fees, speeding up access, or offering loans) based on healthy civic behavior is not particularly innovative but combines ideas from other localities.
The drive is to collect and centralize as much data as possible. One of the key problems that the state has faced is “unreliable and incomplete basic data, and incompatible datasets and systems.” Big data analysis requires integration; surveillance works best when it is networked. The central government will become the hub.
Algorithmic regulation will first “augment human decision making,” but the goal is for data-driven social governance technology to eventually put an end to corruption and a tighter leash on local experimentation by reducing reliance on human judgment.
Tu Zipei, the former Alibaba executive and theorist of social governance, has called the proposed Chinese model “single-particle governance.” The model integrates data from government and commercial sources into individual master files that become the “elementary particle”. This idea doesn’t come from Maoist egalitarian politics or Dengist market horizontality, but from online shopping and social media platforms. Fittingly, it will run on software developed by commercial digital technology firms, like Huawei, Tencent, and Alibaba.
Such a system would fundamentally change the political culture. Horizontality requires not just individual autonomy, but also a sense that local communities or interests can be organized and exercise some kind of collective agency, as the peasants of Henan did under Jiao’s direction and as Chinese companies did throughout the Deng era. Xi’s verticality only requires a population that can be effectively managed. The state can fulfill the interests and express the will of a broad population of abstracted social individuals without having to rely too much on the human judgments of local cadres at all.
Despite the party’s centralism, it has always operated through a huge matrix of institutions: schools, planning committees, workers’ organizations, cultural groups, trade boards, and many others. A person’s political identity was linked in part to the collective bodies in which they participated. But with an urbanizing population that is increasingly integrated into service economies instead of life-long economic or social roles, the bases for these collective expressions of political identity are disappearing.
As the population atomizes, the government seems intent on creating a stronger civic Chinese identity and wants its citizens to politically relate primarily to the national government. Their aggressive cultural assimilation policy in Xinjiang is one example of this. But so is the new rhetoric about data-driven governance: it presumes a population where the individual is a data-generating automaton whose activities are input for the state to work with, with few or no intervening social structures.
The logic of big data governance at its highest scale appears horizontal in flattening the inputs into decision making. It de-emphasizes the importance of political, economic, and intellectual elites but also of local government. It also increasingly removes the possibility of a cadre-managed collective autonomy in goals and decisions.
Xi has based his position on continuing China’s claim to follow an independent ideological path from the West. Part of his impetus for a re-centralization of party-state authority has been a desire to avoid convergence with the liberal atomization associated with Western decline.
But for all of both Chinese and Western bluster about not wanting to be like the other, the structural realities are shockingly parallel. Both have undergone extensive social atomization in this period. Since the 2010s, both have seen new crackdowns on local autonomy and independent discourse using commercially-derived surveillance and social control technologies.
Xi’s invocation of the “fourth industrial revolution” puts him in a path of explicit convergence with the surveillance individualism that has come to characterize Western liberal democracies. The causes and rhetoric differ, but the result seems remarkably similar: an increasingly powerful national state, ideologically and structurally centralized around the national elite, governing an increasingly atomized population. In this new ideal of digital governance, the individualized population is rendered into legible, data-generating citizens strictly governed by a single centralized discourse of possibility.
Western commentators have worried that the Western elite is quietly copying the Chinese surveillance model, but their Chinese counterparts could as easily worry that the party is simply copying the Western model.
What is in danger under centralization of power with these technologies is the radical heterogeneity made possible by horizontality. Save for the direct intervention of the party, a local cadre could act in the best interests of the local authority and its citizens; the decisions a local cadre made did not have to take a strict interpretation of official ideological orthodoxy.
Vertical management can only act in the interests of and by the logic of its central management; the consumer desires of its elementary particles are researched, quantified, and then minimally satisfied. The key difference is that the centralized model has much less room for handling the complexity of local collective problems. Everything is done through the lens of the center. What the party stands to lose is the collective temperament and culture of a cadre base inculcated in the pragmatic, experimental, and innovative environments of the last few decades. This was the environment that reared up a generation of Chinese leaders, including Xi himself.
This homogenization of methods by a single set of central actors is apparent from poverty reduction schemes, which are conducted usually through private-public partnerships, leveraging the expertise and equipment of big tech. In Anhui, the firm that led the AI and big data poverty reduction project was iFlytek, which has been sanctioned by American authorities over their involvement in law enforcement in Xinjiang.
The Building New Rural Communities initiative is another sign of what the future could bring. Rural residents living in fragmented communities are shepherded into central municipalities as a means of providing better governance. In one “resettlement zone,” outside of Jinan in Shandong, planning and building was assisted by China Unicom, turning it into a pilot project for radical digitalization: a “full life-cycle management system” built for “next-generation networking” with the “Internet of Things.” It’s the sort of stuff that makes tech enthusiasts and mechanism designers go wild. What’s not clear is how much it will deliver. But even in the event that the promises of these technologies go largely unrealized, the cultural change they cement might well endure.
Imagining the model cadre Jiao Yulu back in the Henan countryside, perhaps such technologies could have provided him with superior solutions. Big data integration and IoT-ready light poles would have saved him from tramping around Lankao County. And back in the 1960s, without big data, it would have been impossible for anyone to quantify how many lives were bettered by Jiao’s innovations. He was celebrated, though, because he carved autonomy for grassroots innovations from higher levels of government.
In Xi Jinping’s 1990 ode to Jiao, he wrote: “Wherever we are appointed / We must bring prosperity, / Unchanging purpose in life.” The new model of verticality informed by visions of AI-augmented decisions might take human error out of that purpose, but also human inspiration.