How to Digitalize Confucianism on a Planetary Scale

In his 1795 booklet Perpetual Peace: A Philosophical Sketch (Kant 2003), Immanuel Kant articulated his vision for achieving peace on a planetary scale through the concept of “Rechtsstaat,” or the “constitutional state” in English. In Kant’s view, the chances of planetary peace would be greatly enhanced if a majority of nations were organized in a constitutional manner, as this would make it easier to predict the consequences of complex international interplays. It is not hard to see how Kant’s vision of perpetual planetary peace influenced figures such as U.S President Woodrow Wilson, who proposed the establishment of the League of Nations as a form of "organized common peace" after World War One, and Francis Fukuyama, the American political scientist, whose end-of-history thesis argued for the adoption of the Kantian constitutional model by all governments as means of preserving global peace.

This Kantian picture of planetary political structures, though commendable in many respects, highly simplifies the reality of planetary economic, cultural, and political systems. It overemphasizes the constitutional forms of nation-states at the expense of sub- and supra-national elements. Sub-national elements include the power of ethnic loyalties and religious sectarianism (for religious disputes confined within the boundaries of nation-states). The corresponding supranational elements include the power of capital, as well as cultural and religious influences. As a result of this oversight, Kantians fail to do justice to how peace was gradually built over the long course of history. For instance, when the notion of Westphalian sovereignty—considered the forerunner of the Kantian notion of perpetual planetary peace—was established in 1648, the only nation-state that could be considered remotely liberal or democratic was the Republic of the Seven United Netherlands. For Kantians, this raises a question: Why was the peace that ended the Thirty Years' War (1618–1648) and the Eighty Years' War (1568–1648) achievable in Westphalia, given the absence of a large number of democratic states? The explanation is quite simple: a balance of military power prevailed among different sub- and supra-national warring actors making the additional inputs of war unprofitable for all sides. Whether these actors were constitutionally organized was largely irrelevant to the central issue. The participants in the Thirty Years’ War are seen as either sub- or supra-national actors owing to the complexity of the Holy Roman Empire, which itself can be regarded as either a single state, a confederation of states, or a super-state wielding supranational influence. More specifically, the Habsburgs, the ruling house of the empire, was the provider of the main supranational bond in the war by virtue of links between Austria and Spain, alongside its religious influence on its Catholic allies, which was mediated by the papacy. However, certain Holy Roman principalities’ participation in the Protestant campaign greatly undermined the integrity of the empire as a whole and thereby turned certain local feudal organizations into prototypes for more full-blooded nation-states awaiting wider recognition. Consequently, a unified criterion for categorizing the different campaigns of the Thirty Years' War is difficult to identify due to the fact that France, though standing with the Protestant side, was a Catholic country. Nonetheless, the performance of French regular troops compared to Spanish elite infantrymen in the Battle of Rocroi (1643) serves to strengthen my argument that it was the balance between the competing military campaigns, rather than each side’s constitutional form, let alone their religious motivations, that brought about the Westphalian peace. In this light, Kantian peace is merely a post hoc embellishment of what, first and foremost, was achieved through bloodshed. In summary, planetary peace cannot prevail with a balance of power.

However, such a balance of power should not be construed solely in military terms; soft power also needs to be taken into account. The type of soft power that I wish to address here is the power over the control of data, or more broadly, information. Now, based on the preceding discussion, a proposition concerning data-related soft power can be easily derived, which is as follows: In the digital age, planetary peace cannot be achieved without a balance of soft power in terms of data-controlling capabilities. Moreover, given that the large corporations capable of collecting and using big data today operate at either the sub- or supra-national level, the Kantian framework—originally conceived for nation-states—may turn out to be even less appropriate in this age, owing to its inherent disregard for the nature of the balance issue.

The position of this article is straightforward: Just as in 1643, when France relied on its regular troops to counterbalance the elite Spanish infantry, today’s anti-big data activists, like Cathy O'Neil (2016), also deserve forces of their own to countervail the insatiable appetite for data among today’s corporate giants. These “forces” are envisioned as “small dataism,” which technically addresses subnational elements in order to maximally preserve cultural diversity, without which prosperity-generating innovation would be under threat. This diversity-sensitive aspect of my approach further highlights the limitations of the Kantian approach, which typically focuses more on idealized constitutional forms at a relatively abstract level.

The non-Kantian philosophical narrative I have elaborated for my small-dataism approach can also be termed “digitalized Confucianism,” which describes a reimagined form of Confucianism that retains minimal fidelity to the original intentions of pre-Qin Confucians and is maximally adaptable to the digital age on a planetary or even extraplanetary scale. The roadmap for my argument will proceed as follows:

1. Reconstructing the debate between Confucianism and Legalism using a data-based narrative.

2. Justifying the preferability of a Confucian approach to data control in a Star Wars-like interplanetary scenario.

3. Conducting a non-technical discussion on implementing a Confucian and small data-based approach through large-scale reverse engineering of human cognitive architecture on an appropriately abstract level.

1. Disentangling Confucianism from big data-backed Qin-style governance

How can we make Confucianism relevant to contemporary data science, a branch of natural science? At first glance, connecting the two appears to be no easy task, given that Confucianism has traditionally been regarded as a school of thought that overlooks the importance of science (Needham 1956). And yet, isn’t it true that the official ideology of Japan prior to the Meiji Restoration was heavily influenced by Confucianism? This leads us to a further question: Why, despite having great respect for Confucius, did Shibusawa Eiichi (1840–1931), recognized as the “father of Japanese capitalism,” not only achieve immense industrial success but also advocate the popularization of science in Japan? One explanation could be the lack of an Imperial Examination for Selecting Candidates for the State Bureaucracy (IESCSB), as well as the persistence of feudal elements in pre-Meiji Restoration Japan. Together, these elements allowed Japanese intellectuals to reside in a political reality similar to the one inhabited by Confucius around 24 centuries before. This reality included a decentralized political system that operated under a ruler with the title of the “Son of Heaven.” Such a system bears affinities with the Edo Era, where the semi-autonomy of local feudalist lords—known as Daimyo—was maintained by the Shōgun. The semi-autonomy of local cultural regions, which precluded an IESCSB system, enabled Japanese intellectuals to flexibly manage their private time, leading to the emergence of some remarkable theories, including Takebe Katahiro’s (1664–1739) version of calculus, which was developed independently of the teachings of either Isaac Newton or Gottfried Leibniz. In contrast, in ancient China, the IESCSB system developed into a centralized mechanism for managing the time of intellectuals nationwide, especially after it became firmly entrenched during the Song dynasty (960–1279). In particular, the IESCSB’s three levels—prefectural, metropolitan, and palace examinations—were designed to unify candidates’ comprehension of a select number of Confucian classics. This left limited space for Katahiro-like talents to pursue their own goals, which were very likely unrelated to IESCSB requirements. However, while the IESCSB may bear responsibility for the stagnation of science in traditional China, Confucianism itself may be exempted from similar blame, given its compatibility with political realities that did not include the IESCSB, including ancient Japan in the Edo Era, pre-Qin China, and the Han Empire when the Song Dynasty-style IESCSB had yet to emerge. In this sense, if we are to make Confucianism fit for purpose in the digital era, we must execute a “disentanglement maneuver” in order to extricate it from the IESCSB context.

It should be noted that this disentanglement process does not end with the IESCSB, for it is merely one example of a broader category of institutions, namely, any centralized system that aims to unify the management of all the resources of those under its control.More specifically, the target for disentanglement will vary depending on the resource type involved. As such, if we were talking about time resources, the IESCSB would be the target; if we were discussing economic resources, the target would be economic policies, such as the state monopolies on salt and iron; if we were considering tax sources, then the target would be the household registration system, which served to severely restrict domestic migration by registered subjects.

By completing this disentanglement maneuver, we can make Confucianism more suitable for a data-based dialogue based on a simple inference: any resource that can be submitted to a centralized system can also be digitalized insofar as it is technologically possible. Consequently, if a centralized data-management system is the target for disentanglement, a Confucian approach to data management should be decentralized by being aligned with small dataism.

To show that this judgment is based on credible evidence rather than sheer speculation, we can examine the functioning principles of the “Well Field System” (WFS), which Confucians have long admired as an ideal model of economic management.

The WFS was a land redistribution system that existed from the ninth century BC (late Western Zhou dynasty) until around the time of the Warring States period (475–221 BC). To avoid an extensive historical discussion, I will provide a high-level but data-based description of the WFS. The basic motivation for introducing the WFS during the Zhou dynasty was to facilitate tax collection within the constraints of a limited bureaucratic system. The small bureaucracy meant that it was not possible to register individual taxpayers or to accurately measure the area of taxable land. A smart alternative was to shift the focus away from “who pays” and “how much” to the simpler question of “where to collect.” This gave rise to the WFS, which worked in the following way: Land was divided into nine equal square sections, which roughly resembled the shape of the Chinese character “井” (jing, which means “well,” hence the name “Well Field System”). The eight outer sections were individually farmed by peasants, while the central section was collectively cultivated on behalf of the landowning aristocrats. In return, these aristocrats organized a local government to provide public services, including security, to the peasants.This system allowed landowning aristocrats to collect taxes by claiming the harvest from the central section, thereby eliminating the need to gather economic data for the other eight sections. The WFS was founded on both ecological and ethical assumptions. Ecologically speaking, given the proximity of all nine sections to each other, it was likely that the success or failure of the harvest would be uniform across all land plots. Ethically speaking, the mutual benefits and shared bonds of kinship between the aristocrats and peasants meant that private landowners were unlikely to neglect their duty to cultivate the central section. In an ideal scenario, these two factors would reinforce each other so that a crop failure in the central section would be ethically acceptable if the surrounding sections also suffered a similar fate due to some unavoidable ecological change. Hence, with the aid of the WFS, mutual accommodation could be achieved between aristocrats and peasants, and Confucian harmony could thereby be ensured as an important moral outcome. Accordingly, Confucians’ endorsement of the WFS, which is based on small dataism, can be smoothly connected to their standard ethical narrative, which most of us are familiar with.

The overarching Confucian term for practices like the WFS is “Zhou-style governance,” as opposed to “Qin-style governance.” View through a data-centric lens, Qin-style governance—named for the Qin Empire (221–207BC), China’s first true empire—was a forerunner to the big dataism favored by large corporations today. The Qin government’s recipe for seizing resources was to standardize the formats of data across all conceivable aspects of social life. This included everything from measurements and language to more practical details, like the width of chariot axles. Under Qin-style governance, every aspect of social life was so tightly regulated that very little information could be hidden from the government. As a result, families were left with little time and economic resources to pursue individual goals unrelated to the emperor’s concerns of defeating “barbarians” and building palaces. The stagnation of science in China seemed inevitable with the rise of Qin-style governance, which proved to be a powerful weapon against the diversity essential for scientific progress. Regarding the dynamics of the evolution from Zhou-style governance to Qin-style governance, Zhao (2015) suggests that it was largely driven by the expansion of militarism among vassal states. A rapid increase in the regular troops of each vassal state helped bolster their respective tax systems. This inevitably created an insatiable demand from the government for data concerning all aspects of the economy.

However, from a Confucian and small data-based perspective, while these Qin-style measures delivered military success in the short run, they ultimately depleted the energy of society over time, as the big dataism exploitation of social resources provided little space for the patience needed to replenish populations and other economic resources. Accordingly, in the standard Confucian view, the fleeting success of the Qin Empire fails to provide a justification for Qin-style governance. Rather, as Yi Jia articulates in his work On How to Surpass the Qin Empire, the Qin’s ephemerality only serves to highlight the fundamental flaws of these methods. Jia further asserts that the only way for the Han dynasty, the Qin’s successor, to avoid such tragic brevity was through the realization of the Confucian principle of benevolence in conjunction with Zhou-style governance.

Nevertheless, the imperial architecture that the Han inherited from the Qin was so centralized that a general restoration of the WFS was out of the question, particularly given the bloated bureaucracy and the fact that its members served the emperor rather than local feudal societies. Furthermore, the existence of an outsized bureaucracy provided the government with a justification for strengthening the large data-based tax mechanism, which it relied on to support itself. Ironically, then, Qin-style governance was able to re-emerge in the guise of the Han, albeit in a more moderate form, due to the rhetorical victory of Confucian ideology. During the Song dynasty, the camouflage disguising this revival became even more effective. Notably, the imperial court’s rhetorical respect and financial support for Confucian scholars has led many careless observers to mistakenly regard the Song as a golden age for Confucianism, when it was in fact a golden age for Qin-style governance. So vast were the central government’s economic resources that it was able to maintain an unprecedented regular army whose scale (1.2 million soldiers out of a population of 90 million) was outmatched only by its incompetence on the battlefield. Alas, the military burden eventually became so great that the threat of bankruptcy loomed. A program of disarmament intended to focus a smaller military budget on the most capable troops appeared to be the most sensible solution, and this is precisely what Fan Zhongyan attempted with his Qingli Reforms. However, when the reforms ran aground because of a backlash against the forced retirement of officers and officials, the impatient imperial court decided to change tack. This led to the introduction of new chancellor Wang Anshi’s Green Sprouts Law, along with other measures carried out during the Xining Reforms. In essence, the purpose of this law was to expand the national budget through a nationwide loan plan that was semi-compulsorily applied to nearly all taxpayers. The resulting interest from the loans was intended to serve as profit for the government. In reality, this plan was just the latest reincarnation of Qin-style governance, which had traditionally aimed to facilitate the expansion of government power. However, its true nature was cleverly concealed with Confucian camouflage, as the Green Sprouts Law was intended to financially support the poor who had little chance of obtaining low-interest loans elsewhere. But this camouflage did little to fool sharp Confucian observers like Sima Guang, Ouyang Xiu, and Su Shi, who strived to shoot down Wang’s plan to “Make the Song Great Again.” They argued that the intrusion of state power into the private economy would rapidly corrupt the virtues of the entire bureaucratic system, whose members were supposed to be the people’s teachers, not their moneylenders. This ethical criticism can be easily translated into a small-dataism narrative, which takes the following form: one of the virtues of Confucian officials was supposedly their ability to suppress any impulse to collect economic data about the people, even when such data could be to their benefit. As such, the act of gathering data about who needed urgent loans—as required by the Green Sprouts Law—could not be tolerated. In this sense, the economic omniscience required by Wang’s program should be judged as a moral flaw.

Despite the strong criticism of Wang’s Green Sprouts Law by many of his Confucian contemporaries, the emperor’s support meant its implementation was a fait accompli. While the law created some short-term profits, it gradually drove large numbers of peasants into debt. The resulting famine and social unrest allowed reform opponents to temporarily halt the program, but the pause came too late to save the Northern Song Empire, which fell during the Jurchen invasion of 1127. This analysis shows that the Song’s tragic failure should be attributed not to Zhou-style governance, which advocated restrictions on government data-gathering, but rather to Qin-style governance practices, which Wang skillfully masked as Confucian measures to aid the poor. This understanding makes the abovementioned disentanglement maneuver even more valuable, as it removes Qin-style governance to allow for a faithful depiction of the nature of Confucianism, even though its full-blooded application in practice was all too easily thwarted by imperial power. Ironically, however, the prevalence of such imperial interference reinforces the point that the small-data nature of Confucianism is better suited to a decentralized political system—something rarely seen in China after the Qin Empire unified the country in 221 BC.

This concludes our discussion on the disentanglement maneuver intended to save the small data-based Confucian position from contamination by Qin-style governance. But the question as to whether such a maneuver is truly relevant to the big dataism of today still persists. I argue that it is, and I will now proceed to explain why.

2. Applying Confucian small dataism to the world of today

When we examine the abovementioned disentanglement maneuver through an economic lens, its contemporary relevance becomes readily apparent. Contemporary Western libertarians might find common cause with Confucian critics of Wang’s reforms since both groups are hostile to any form of state intrusion in the private economy. Moreover, Confucian objections to Wang’s policies also resemble certain U.S. Republican criticisms of the Affordable Care Act (ACA), otherwise known as “Obamacare.” The ACA, which mandates the purchase of medical insurance by all Americans, bears some affinity to the semi-mandatory loans of the Green Sprouts Law. However, while these parallels are intriguing, they risk oversimplifying the contemporary libertarian-liberal debate by framing it in terms of the opposition between supporters of Zhou-style governance and their Qin-style foes. This simplification is problematic for the following reason: unlike many contemporary libertarians, Confucians do not fundamentally oppose social authority. On the contrary, they believe that between state-sanctioned legal frameworks and individual liberties, considerable space should exist for soft power to establish a sophisticated hierarchy of social authority. This is best exemplified by the Confucian ritual system. That said, the operation of this ritual system is still grounded in a small-data approach since economic data required for the holding of, say, solidarity-maintaining ceremonies would circulate only within clans or other forms of local communities without being passed on to central authorities. Within such communities, geographic and consanguineous ties between the providers and users of local economic resources minimize the potential for exploitation by the data-managing side. This is similar to what we have seen in the WFS system during the Zhou dynasty.

It is notable that the divergence between Confucians and contemporary libertarians does little to narrow the gap between Confucianism and liberalism. Confucians, like libertarians, worry about the unlimited expansion of state power. Liberals, on the other hand, tend to endorse such expansion in pursuit of Rawlsian distribution justice. When confronted by Confucian concerns about such expansion, contemporary liberals often attempt to dismiss them with the optimistic assumption that a Kantian constitution—which was most definitely absent from ancient China—can safeguard people’s interests, including those concerning their private data. Based on this logic, the state power required to enforce the ACA can also be justified, given the legitimacy of the act itself. However, from a non-liberal perspective, this optimism is unwarranted. The Kantian principle of “rule of law” is based on the mistaken assumption that lawmakers can fully comprehend the part of reality for which they are legislating. In today’s digital environment, this assumption proves problematic because the technical complexities of the data-driven industry often exceed the understanding of “outsiders”—including many congressmen and senators. “Insiders” can easily mislead them with technical jargon, much as Wang Anshi intended to deceive his colleagues by reinterpreting the Confucian texts. Unfortunately, the Qin-style governance built into today’s big dataism is more difficult to expose than its Song dynasty counterpart, as the underlying algorithms are far more complex than any reform proposal ever posited by Wang. This dynamic gives rise to a paradoxical situation in the typically liberal societies of the West: By maintaining exclusive control over both the knowledge of big-data systems and the mechanisms through which they work, large corporations can easily apply Qin-style-governance practices over people who believe themselves to be politically free, even though these corporations may be blissfully unaware of how the Qin government actually operated.

The following demystification of algorithms, inspired by Cathy O’Neil (2016), among others, may help to more clearly illustrate the parallels between Qin-style governance and today’s big dataism.

Firstly, an algorithm's technical complexity can conceal from users the embedded intentions of its designer. This feature has something to do with the technical meaning of “training” as it relates to deep learning systems. In simple terms, training a deep learning system largely resembles Ivan Pavlov’s conditioning of dogs: Trainers guide the system’s behavior via a series of rewards and punishments until it achieves the designer’s goals—for example, recognizing Brad Pitt in stills of his new movie. So far, this process is unproblematic, as no one would want an algorithm that would misidentify Brad Pitt as Angelina Jolie. However, the situation changes entirely when we move from perception-based concepts like “being Brad Pitt-like” to more value-laden concepts like “whether someone is handsome or not.” Whereas identifying Brad Pitt can be done objectively, evaluating whether someone is handsome must be guided by a set of values. Similarly, whether a conversation is provocative or politically incorrect will also be determined based on certain values. The crucial question is, whose values? Inevitably, it is the values of the algorithm’s developers—the “insiders” who can easily embed their values into the system without needing to inform the public about the criteria they have used for selecting their training data. As Shibani Santurkaret al. (2023) reveal, ChatGPT, the popular artificial intelligence chatbot, exhibits significant systemic political biases. Exposing such biases requires expertise beyond the capacities of ordinary users. Consequently, with this ability to quietly inject specific values into digital products, developers can easily realize a more discriminating version of Qin-style governance in the digital era. The key difference between this version and its original ancient Chinese counterpart is that it maintains the illusion of freedom. Unfortunately, due to the aforementioned knowledge deficit of lawmakers about the reality they legislating for, exposing such a situation remains a challenge, even within a Kantian constitutional framework.

The illusion of freedom felt by today’s algorithm user also highlights another parallel between today’s Qin-style governance and its historical counterpart: the standardization of data formats. During the Qin Empire, such standardization, which, as mentioned, concerned issues like language, currency, and measurements, was strictly enforced, and any non-compliance was met with severe punishment. In contrast, modern Qin-style governance tends to rely more on Pavlovian rewards rather than punishments. Consider, for example, the current popularity of short video platforms, which attract users by providing instant gratification through superficial content. However, these rewards come at the cost of one’s critical thinking abilities. Platforms exploit the psychological principle of random reinforcement: The endless video stream is addictive in itself, simply because we expect to get a reward (in the form of a funny video, followed by a shot of dopamine) on an almost continuous basis (White 2024). This addiction greatly diminishes users’ ability to focus on longer content, such as a 15-minute video or a longer article such as this one. Capitalizing on humanity’s natural aversion to the kind of hard thinking required for time budgeting, short video algorithms standardize the “optimal” video length at 20-30 seconds. Such brevity simply does not allow for the unpacking of any sophisticated argument, even though the patience required to follow such arguments is essential for citizens to meet the Kantian criterion of “daring to know.” Ironically, without resorting to coercion, today’s short video platforms easily accomplish through endless video consumption what Han Feizi (one of the philosophical architects of Qin-style governance) intended by means of state-mandated labor: to keep the populace weak both economically and intellectually by keeping them preoccupied.

Another parallel between the original Qin-style governance and its modern digital counterpart is this: Both undermine communal solidarity by atomizing the individual. They do so by aligning them with unified data-submission formats that are cleverly designed to cut each one who submits off from his or her own community. The TV drama The Chair, though fictional, effectively illustrates this phenomenon in contemporary campus life. When Bill Dobson, a professor in the English department at Pembroke University, satirically imitates Hitler during a class, students record the act on their smartphones, edit it, and then circulate it online. This leads to widespread accusations that Dobson actually is a Nazi, almost destroying his teaching career. What is striking about this storyline is the absence of a traditional Confucian respect for teachers. A minimal degree of such respect would have prompted students to re-think the context and intention behind their professor’s actions. It also would have given them pause to consider why Charlie Chaplin’s imitation of Hitler was politically acceptable. Even by more relaxed Confucian standards that allow students to criticize teachers in offline settings and behind closed doors, basic respect would still have prevented them from sharing the edited video online—much as a son with minimal respect for his father would not upload a video of his father behaving improperly. Ultimately, the students shared the video online because they valued the potential virtual rewards offered by the internet more than their real-world offline relationship with their teacher. This mindset is particularly destructive for communities because it undermines local community leadership. And without this, individuals cannot be effectively organized to resist modern-day Qin-style totalitarianism, which is built on the technical foundations of big dataism.

An immediate consequence of this process of authority erosion is the elimination of cultural diversity. According to Linda Zagzebski, “We do not have criteria for goodness in advance of identifying the exemplars of goodness” (Zagzebski 2004: 41). In other words, the subtle moral information embodied in exemplars’ behaviors cannot be captured by any Kantian ethical criterion—only immersion in moral hero-emerging contexts can effectively guide the conduct of followers. Given that such immersion must occur in local communities where these subtleties can be preserved, the requirement for practically knowing what to do is merely minimal loyalty to the local community, and especially to its spiritual core in the form of the local cultural hero or exemplar. It should be noted that respect for cultural diversity goes hand in hand with respect for the local cultural hero since a large number of these heroes are recorders and developers of local culture. Consider, for example, the distinctiveness of Czech literature—in contrast with, say, German literature. These unique qualities might never have been widely recognized if local Czech writers hadn’t shown sufficient respect to local cultural heroes, such as Jaroslav Hašek. In other words, Hašek’s followers had to appreciate all the nuances unique to Czech. If not, the dominance of German in the Austro-Hungarian Empire—where Hašek himself served as a soldier—would likely have forced many Czech writers to write in German. This would have made the later success of Milan Kundera, who initially wrote in Czech before fleeing to France, improbable. To grasp just how essential this local authority transmission chain is for preserving diversity, it may help to imagine a possible world where cultural resistance to Qin-style standardization was not tolerated—where Dante’s choice to write the Divina Commedia in Italian rather than Latin, or Martin Luther’s translation of the Bible from Latin into German, was deemed shameful. Such worlds would undoubtedly be less culturally diverse and poorer in the absence of cultural phenomena like Italian literature and the Protestant Movement. Fortunately for our world, the diversity we inherited from the pre-digital era has given us Protestant thought and literature written in Italian, German, and Czech. Most unfortunately, however, it appears that this diversity-producing mechanism will now struggle to function due to the authority-eroding effects of big dataism. To illustrate, let us consider how the typical short video platform operates: the commercial value of a content creator depends on the click rate of his or her videos. Such a rate depends on meeting the expectations of viewers, most of whom are not living in the same community (or even the same country) as the creator, who consequently does not qualify as a cultural hero. Driven by click rate-optimizing algorithms, creators have little choice but to cast aside their local culture if it fails to generate engagement and boost clicks. In this sense, Hašek was fortunate to have published The Good Soldier Švejk in the 1920s, before click rates ever existed. His digital-era counterpart will very likely be ignored by the world for his anachronistic insistence on using Czech rather than English, the primary language used for training today’s ChatGPT.

What might Kantians say about this picture? In light of such extensive cultural destruction without any visible breach of a Kantian constitution, what solutions would they propose? While it may seem arbitrary to completely deny the existence of a Kantian remedy, I will still contend that a Confucian approach can provide a more effective solution.

3. The Kantian Remedy vs. Its Confucian Counterpart

As mentioned in the introduction, in many cases, harmony can be achieved only by means of a balance between competing forces. Accordingly, in light of the dominance of large corporations backed by big data, an easily conceivable countermeasure is to rely on public power, which itself must be constitutionally established. Following this train of thought, a Kantian remedy for handling big data-related issues involves establishing data-regulation laws, like the General Data Protection Regulation (GDPR), a regulation on information privacy within the European Union (EU) and the European Economic Area (EEA).

The GDPR can be considered Kantian in two ways: firstly, it is designed mainly to protect personal data, which are non-physical entities. This echoes the Kantian view that the possession relationship is itself non-physical. As Kant asserted, “I can say that is it mine only if I can assert that it is my possession even if I leave the place in question,” (Kant 1965: 54.). This implies that my possession of my personal digital data is valid even when I am offline. Secondly, the GDPR features an institutional apparatus called the “One Stop Shop Mechanism” (OSSM). This mechanism requires data handlers active in multiple EU member states to liaise primarily with the supervisory authority in the location of their main establishment. This ensures a Kantian/Westphalian respect for the sovereignty of each EU member state.

But would a Confucian find this solution satisfactory? To recall, Confucians typically oppose excessive bureaucracy, as its tax-consuming nature evokes Qin-style governance, even under the label of democracy. Lamentably, the legal procedures involved in applying the GDPR are both time-consuming and costly due to the large number of potential legal disputes arising from the issue of whether a data handler’s practices meet GDPR criteria, not to mention the OSSM-induced over-bureaucratization stemming from the multiplicity of EU members. For example, in 2021, Luxembourg's National Commission for Data Protection (NCDP), backed by the OSSM mechanism, claimed that Amazon's handling of personal data violated GDPR rules. This led to a hefty fine of $886.6 million on Amazon, though it vehemently disputed the accusation. The main question at hand is whether, regardless of the potential legitimacy of any such charge, such a heavy fine would likely trigger consequent appeals leading to a lengthy legal process (Due to space limitations, I won’t delve into the details of the ongoing legal battle between the NCDP and Amazon. For further discussion, see Burgess 2022). Such a situation would naturally lead a Confucian to question whether there might be a Confucian alternative to the Kantian remedy.

To recall, according to the Confucian ideal, the ritual system plays a less resource-intensive role in regulating the relationship between government and individuals. However, from a liberal perspective, the functioning of such a ritual system requires the validation of patriarchy, which left-wing liberalism seeks to challenge. Moreover, from a scientific standpoint, the Confucian ritual system is too rooted in agricultural civilization and thus too remote from the technical realities of the digital era. In essence, many feel that the anachronistic aspects of any Confucian narrative are fundamentally at odds with an era in which everything can be digitalized.

However, I will still insist on the conceivability and even the feasibility of the notion of “digitalized Confucianism.” Undoubtedly, in order to vie with big dataism at a practical level, such a concept must go beyond merely reconstructing Confucianism using a data-based narrative, as we have seen in Section 1. My proposal is to introduce a new AI approach that embodies the principle of small dataism. In principle, this approach does not involve large-scale data collection, thus eliminating any need for non-local data management. The localized data processing of small dataism would enhance the security of personal data without the need for state intrusion. Consequently, GDPR-like regulations would be less necessary in any potential world where small dataism was preferred.

But what makes this approach Confucian? It is Confucian primarily because it aims to support robots designed to enhance the capacities of families, which are the building blocks of local communities and without which the edifice of Confucian ideology would likely crumble. By “robots,” I mean embodied artificial assistants that help humans with cooking and caring for the sick, along with other family-related tasks that are located in the real world. This is in contrast to disembodied programs like chatbots or machine-learning conversational engines that can be easily deployed to undermine communal solidarity through misleading comments. From a Confucian and communitarian point of view, the ideal small-data robot would be more like a family member in the sense that it would be loyal enough to prioritize the guidance and interests of its master. Loyalty of this type would allow these robots to amplify the positive consequences of their masters’ actions due to their physical durability compared to the human body. Thereby, their masters, as family leaders, would be able to seize more resources to counterbalance the formidable power of Qin-style governance, whether that be in the form of the Qin Empire of old or the high-tech big-data firms of today. This point can be illustrated using the famous science fiction series Star Wars, which contains more Confucian elements than many realize. Consider the following question: How does Han Solo, the captain of the Millennium Falcon and his greatly outnumbered Rebel Alliance, effectively oppose the giant Galactic Empire, which could be considered an extraplanetary version of the Qin Empire? It is clear that he could not have succeeded without loyal robots and droids like R2-D2, who, as an expert in both repairing and destroying electronic and mechanical devices, plays an indispensable role in the team’s escape from the clutches of the empire, as well as C-3PO, whose fluency in over six million forms of communication helps bridge the cultural gap between Han Solo and inhabitants of other planets. In other words, it is the amplification of Han Solo’s power and will by these robots that establishes him as a hero, in much the same way that Confucius was established as an enduring cultural icon thanks to loyal disciples who amplified their master’s voice. As such, it is possible to imagine a world where Confucius was not only supported by human disciples like Zilu or Yanhui but also by droids like R2-D2 or C-3PO. In such a world, Confucius’ will and power could be amplified to the extent that it could cut off the evolutionary path leading to Qin-style governance. Moreover, due to the family-centered nature of these droids and the multiplicity of families, the proliferation of such technology would prevent the emergence of another form of Qin-style governance. What would emerge in its place is a balance between different players—one that would help create a culture dish for nurturing Confucian virtues such as integrity, promise-keeping, and honor.

But why is it impossible for robots of the preceding function to be underpinned by big dataism? In my view, big dataism is a non-starter in a Star Wars-like extraplanetary environment. This is because such an environment fails to satisfy certain essential criteria for the application of big dataism: (1) Sufficient data for training algorithms, especially about unexplored planets; (2) adequate electronic power for running ChatGPT-like Large Language Models on a Millennium Falcon-stylespaceship, where power is already limited; (3) a WIFI-system that could function on an extraplanetary scale. Even on our own planet, big dataism would prove less useful for domestic robots that are expected to possess artificial general intelligence (AGI). This is because the unique details of a specific home environment—a rare dialect used by the master, the unusual arrangement of furniture, or the complex health conditions of residents—can easily be overlooked by big dataism, which is intended to serve “normal users.” This is not to mention the risk of a household’s private data being exposed to tech companies. If big data-backed robots were imposed on households, it could lead to the loss of family culture diversity through a Qin-style process of standardizing everything. In a worst-case scenario, it may lead to a digital version of the “panopticon” so severely criticized by Michel Foucault (1975).

This wraps up the discussion on the ethical preferability of Confucian small-data robotics over their big-data equivalents. As to the preferability of the Confucian approach over its Kantian counterpart, which is represented by GDPR-like regulations, we can draw the following conclusion: Since the primary aim of Kantian regulation is to limit big dataism, once big dataism is inevitably proven to be a white elephant, the transition to small dataism will quickly render such regulation obsolete. Ultimately, therefore, the development of small dataism will prove to be a game-changer.

Kantians may contend that, as such a game-changer is as yet unavailable, GDPR-based privacy protection remains valuable at present. To some extent, that is true. However, generally speaking, the framing of these protective regulations is too negative, such that AI developers have little information on how to positively enrich cultural diversity in a digital context. The most these regulations can do, therefore, is slow down the process of de-diversification. In other words, they serve merely to postpone the coming of digital Qin-style governance rather than prevent it outright. In this sense, the Confucian solution to issues posed by big dataism may well prove to be invaluable.

Some may still doubt the technical viability of a Confucian alternative. Yet, it is not as unfeasible as it first seems. Consider, for a moment, the “Expert System” (ES), a once popular computer system designed to emulate the decision-making ability of human experts. ES can be considered small dataism, as it only needs to encapsulate the knowledge of experts rather than the everyday expressions used by ordinary people online. It is also somewhat oriented toward Confucianism, given its elitist implications, which echo Confucian respect for social authorities. Importantly, it also ensures user security since data processing largely occurs on local ES devices. Although ES is notorious for its incompetence in acquiring new knowledge and making inferences in ambiguous contexts, its existence shows that big data is not the only game in the town. Although ChatGPT, which is big-data based, may appear to be much more adept at reasoning in ambiguous contexts, according to the Turing Prize winner Yann LeCun, it can only capture statistical dependencies between input and output without actually engaging a common knowledge base, much less truly understanding the external world. LeCun himself proposes a non-big data alternative called “Image-based Joint-Embedding Predictive Architecture” (I-JEPA, cf. Meta 2022). This proposal would allow systems to achieve representations at a high level of abstraction of certain input images without requiring extensive training data. Additionally, AI scientist Pei Wang has long been engaged in the development of a general-purpose intelligent system, namely, the Non-Axiomatic Reasoning System (NARS), which is expected to exemplify the same principles that guide human thought at an appropriately abstract level (cf. Wang 2013). Therefore, the marriage between AGI and digitalized Confucianism is not as implausible as it seems.

Conclusion

Due to its conservative outlook, Confucianism has long been dismissed as irrelevant to contemporary digital technology. However, when the Confucian doctrine of benevolence is brought down to earth through low-tax policies and concomitant economic measures, the dimension of “digitalization” naturally emerges in a Confucian context. Such a context allows for the reinterpretation of two competing political-economic positions:

1. Confucian Zhou-style governance, according to which the local clan’s right to economic privacy must be preserved to defend local economic vitality. Without this, the local political authority of Confucian grand clans would quickly lose its economic roots. Since such a position limits the central government’s ability to gather data on civil economic activities, it can be labeled a primitive form of small dataism.

2. Anti-Confucian Qin-style governance, which promotes the expansion of a central government’s right to collect data concerning civil economic activities via, say, a household registration system. This ensures a government may gather enough resources to realize its short-term national goals without concern for any negative social impact in the long run. Such a position can accordingly be labeled a primitive form of big dataism.

In its modern guise, big dataism is typically embodied in deep-learning algorithms. It has been completely updated, becoming both digitalized (in a literal sense) and capitalized, to the point where the in-built disciplinary power of its algorithms is eroding the cultural diversity and internal solidarity of local communities at an accelerating rate. In this new historical context, Confucian small dataists are unlikely to prevail in their millennia-long battle against big dataism without a radical technological enhancement of their weapons, which must be powerful enough to mitigate the threat of big dataism. Small data-based home robots that embody AGI can find a place in a Confucian-oriented arsenal in the following ways: First, data-processing for these robots can be localized to minimize the risk of privacy breaches; second, in aging societies with declining birth rates, these robots can empower family members by facilitating the physical realization of their will. Such an empowering mechanism would further safeguard cultural diversity, which is deeply woven into the offline activities of community members. High-level regulations constraining big dataism, like the GDPR, can supplement digitalized Confucianism. Yet, the approach I propose is indispensable, as GDPR is incapable of providing a new technical roadmap for meaningful competition with big dataism. This inability relates to GDPR’s formalist Kantian assumptions, which result in a failure to provide positive guidelines for shaping new technical realities. In this sense, Confucianism, rather than Kantianism, may be considered a preferable culture dish for nurturing a more promising digital life on a planetary scale.

References:

Burgess, Matt. “How GDPR Is Failing.” Wired, May 23, 2022. https://www.wired.com/story/gdpr-2022/.

Kant, Immanuel. The Metaphysical Elements of Justice: Part I of the Metaphysics of Morals. Translated by John Ladd. Indianapolis: Bobbs-​Merrill, 1965.

Kant, Immanuel. To Perpetual Peace: a Philosophical Sketch. Translated by Ted Humphrey. Indianapolis: Hackett Publishing, 2003.

Meta. “Yann LeCun on a Vision to Make AI Systems Learn and Reason Like Animals and Humans.” Meta AI Blog, Ferbruary 23, 2022. https://ai.meta.com/blog/yann-lecun-advances-in-ai-research/.

Needham, Joseph. Science and Civilisation in China: Volume 1, Introductory Orientations. Cambridge: Cambridge University Press, 1956.

O'Neil, Cathy. Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy. New York: Crown, 2016.

Santurkar, Shibani et al. “Whose Opinions Do Language Models Reflect?”
arXiv:2303.17548 [cs.CL], 2023. https://doi.org/10.48550/arXiv.2303.17548.

Wang, Pei. Non-Axiomatic Logic: A Model of Intelligent Reasoning. Singapore: World Scientific Publishing Co., 2013.

White Caleb. “TikTok Brain: Short Videos Suit Our Attention Spans but Are Killing Our Ability To Concentrate.” The Science Times, June 07, 2024. https://www.sciencetimes.com/articles/50587/20240607/tiktok-brain-short-videos-suit-attention-spans-killing-ability-concentrate.htm.

Zagzebski, Linda. Divine Motivation Theory. New York: Cambridge University Press, 2004.

Zhao, Dingxin. The Confucian-Legalist State: A New Theory of Chinese History. Oxford: Oxford University Press, 2015.