[author]Sun Xiaoxia

[content]

How Digital Power Shapes the Rule of Law? --On the Logic and Mission of the Digital Rule of Law

Sun Xiaoxia

Professor and Director of the Institute of Digital Rule of Law, Zhejiang University

Abstract: Why should jurisprudence expand from the study of algorithms to the study of digital power? Observing the impact of digital technology in both commercial and state applications, it can be found that digital private power and digital public power have been formed, and the universal social risks they bring are threatening human rights and reshaping the rule of law. As a new type of power, digital power is legal in nature from both factual and normative perspectives, and it should be included in the scope of jurisprudence, and consequently in the legal conceptual system. While being controlled by the rule of law, digital power can also overcome the inherent shortcomings of the rule of law, improve human rights and the rule of law, and benefit mankind. It is based on this logical starting point, the digital rule of law has a clear and special mission - the use of digital power to shape a future rule of law that can "empower and promote goodness".

Keywords: digital power; digital private power; digital public power; digital human rights; digital rule of law

In the traditional social order, state power has always been the only dominant force. Therefore, the "power" in law refers to the state power, i.e. public power, and the so-called power order in traditional theory is the order of division of state power. In ancient Greece and Rome, there was the idea of controlling power by law. Since the modern revolution, the concept of the rule of law and human rights, in order to check and balance power, to protect human rights, the rule of law came into being. Thus, human rights, public power and the rule of law have become the three elements of power in the order of governance. Since the Industrial Revolution, despite the progress of science and technology, there has never been a scientific and technological force that has been able to exert a dominant influence on this triangular order. In the era of the first rise of the Internet, people only used it as a social communication tool, and sociologists were worried that the Internet would reinforce social segregation and atomization. For example, Giddens discussed the impact of globalization in the first edition of his Sociology published in 1989, and added a discussion of the new impact of the Internet and information technology in the fourth edition. Although Giddens treats the Internet only as a media and communication tool like television, he still worries about "losing our identity in cyberspace," asking the question, "Will computer technology rule us rather than us controlling it?" This view is certainly prescient, but does he see the future of technological development as it is today?

Today, humankind has entered the digital age, where human freedom is lost to the hegemony of technology. Facilitated by contemporary emerging science and technology, represented by digital technology, the digital economy, digital society and digital government have emerged, which, while bringing benefits to social life, also have a certain dominant impact on all three: human rights, public power and the rule of law. In other words, technology is shaping new powers and new rule of law. The central thesis of this paper is twofold: first, why should digital power be incorporated into the law as a juridical and legal concept of legal nature and significance? And two, what is the logic and mission of the digital rule of law? These two theses, the former is the logical premise of the latter, and the latter is the ipso facto conclusion of the former. Around these two theses, this paper intends to discuss the following five aspects: First, how does the application of algorithms bring algorithmic power and expand into "digital power"? Second, how do the two forms of digital power participate in and affect the triangular order of human rights, public power, and the rule of law? Third, from a risk perspective, how are the two forms of digital power threatening human rights and shaping the rule of law? Fourth, from a jurisprudential standpoint, can the rule of law only adopt an attitude of apprehension and a stance of passive regulation towards technology? Does digital power have the potential to shape the rule of law for good? Fifth, what is the mission of the digital rule of law? What is the possible vision of the digital rule of law? This paper attempts to "wide-angle" perspective panoramic examination of the positive and negative impact of digital power on the digital rule of law, and discuss the significance and mission of the digital rule of law.

1. Transfer: the application of algorithms brings algorithmic power and expands into "digital power".

From algorithm or algorithmic power to digital power, tracing its research lineage, we can see the significance and necessity of digital power research. The research on the relationship between digital technology and power in China's academic circles probably started at the end of last century. 1999, an article introducing the digital technology issues in the American media world talked about the power structure and commercial interests brought by digital TV. The use of the concept of "algorithm" in the domestic natural science community may have begun in 2001, when some scholars keenly put forward the concept of "digitization of power", which refers to the digitization of public power. 2007, some domestic social science scholars published an article on "technological power", which mentioned that the relationship between digital technology and power may begin at the end of the last century. In 2007, a domestic social science scholar published an article on technological power, which mentioned that "technological power has existed relatively independently in contemporary times", and in 2010, there was a paper in the field of media studies with the keyword "digital power". At that time, the legal profession did not respond in a timely manner to this topic of "digital power" raised by disciplines outside of law. It was not until 2018 that the legal profession began to study the power generated by algorithms. Like the media academy, the legal academy focused more on and studied "algorithms" rather than "numbers". At this time, the legal profession began to call "algorithms" "power", such as Zheng Ge proposed that "intelligent algorithms on the basis of big data have created a new form of power, and also called for a new mode of rights protection". "In 2019, Zhang Linghan explicitly used the concept of "algorithmic power", arguing that "the law should configure independent algorithmic interpretation power to equalize the asymmetric power relationship between automated decision-making users and their counterparts". Searching the China Knowledge Network, so far in 2018, there have been more than 50 journal articles with the keyword "algorithmic power". Among these 50 articles, only individual scholars have discussed algorithms from the perspective of philosophical subjectivity and rights protection, and the number of articles advocating the governance of algorithms as power is gradually increasing in the academic community. These legal articles on "algorithmic power" generally discuss legal issues such as algorithmic black box, algorithmic discrimination, violation of personal privacy, and procedural justice of algorithms.

However, it must be recognized that legal scholars or judges are powerless in the face of the algorithmic black box, and can only look at these issues in accordance with the traditional path of jurisprudence - the so-called procedural thinking, i.e., converting the technical issues into issues of due process to be scrutinized. In recent years, some scholars keenly algorithmic power is located in due process issues, from the jurisprudence of the "best way to intervene" to limit the jurisprudence of the problem awareness and "issues", thus narrowing the scope of jurisprudence in the study of this issue, objectively reduce the jurisprudence of the "burden" - the "burden" of jurisprudence, the "burden" - the so-called procedural thinking to look at these issues. The "burden" - scientific and technological issues have the weight that legal scholars can not bear. For example, according to Chen Jinghui, "if one realizes the importance of the power of algorithms, then algorithms are no longer speech or trade secrets, but a kind of due process; accordingly, breaking the black box of algorithms is not the best way to intervene, but rather should intervene in algorithms with the basic requirements of due process". Algorithms "not only challenge the fundamental attribute of procedural limitations on arbitrariness, but also impact the intrinsic values of procedural justice such as participation, openness, neutrality, and fairness," and therefore, "procedural justice theories and institutional connotations need to be renewed and advanced on the basis of traditional foundations."

It is possible to regulate the issue of algorithmic power by limiting it to due process. But the question is, can jurisprudence only focus on and regulate digital technology issues in terms of procedural due process? This is questionable. Because it may lead to the omission of some huge risk areas in the social application of digital technology. As the range of applications of digital technology expands, the concern is no longer just about the algorithmic technology itself, but also about whether and how it should be applied in a particular area - something that even algorithmic engineers have trouble deciding. For example, when applying algorithmic technology to automated crime prevention surveillance, engineers are responsible for the accuracy and transparency of the algorithms, but not for decisions about which intersections or houses to use for automated surveillance. Because algorithms are only one aspect of digital technology, algorithmic power is also only one aspect of digital power, which covers a much larger and broader area and has more complex manifestations. Therefore, it is necessary to shift from the technology itself to the angle of social impact, comprehensively and deeply analyze the actual changes brought by the application of technology to society, so as to re-examine the phenomenon of "power" known as "digital power".

First of all, social sciences should focus on the effects of the social application of digital technology, that is, the influence of digital technology in the field of social application. The problem of digital technology is only a technical problem in the final analysis, and only by taking the application of digital technology as the object of social science research can the research purpose be reflected. From both domestic and international perspectives, the concept of "algorithm" is more often used in related research, so when we refer to algorithmic dominance, we can continue to use the concept of "algorithmic power" as a common practice. However, with the wide application of algorithmic technology in social life, the connotation of the concept of algorithm has changed. 2019, some scholars of philosophy and social sciences mentioned "digital power" again, arguing that in the "microscopic, digital power dominates the behavior of actors and users of the digital network and transforms the traces of their activities into a big data network. traces of their activities into big data networks. At the macro level, the big data network composed of traces of user behavior has become a third form of capital, digital capital, over and above industrial and financial capital". 2020 to 2023, there is an increase in the number of papers that refer to the concept of "digital power", with at least five of these papers From 2020 to 2023, there is an increase in the number of papers that mention the concept of "digital power", with at least five papers using "digital power" as a keyword in their titles.2023, some legal scholars use "digital power" as a keyword in their research on the application of digital technology on the 12345 government platform. The "digital power" mentioned here is the special administrative power generated by the government's use of digital technology to provide services in the administrative context, but it does not show the full face of digital power. When looking horizontally at the field of social sciences, social science scholars' observations are more comprehensive. For example, one article analyzes and argues the concept of digital power from the perspective of political science or political sociology, arguing that "digital power has been born as a new form of power and has further changed the organizational and legal forms of the original power". An increasing number of social science scholars have shown a tendency to use the concept of "digital power" instead of "algorithmic power", and this situation is expected to continue to grow.

Have foreign scholars in the social sciences used the concept of "digital power" or similar concepts? The answer is yes. The original "Data Power" (Data Power) is an old concept that originated in the 1980s and was used in the field of CD-ROM storage of library materials. In recent years, there are still foreign legal scholars using similar concepts, such as in 2019, Orla Lynskey in the famous "Responding to Data Power: A Normative Push from Data Protection and Privacy" article used the concept of "Data Power". So, does the concept of "data power" in this context refer only to the power of big data? Apparently not. One scholar, Isabel Hahn, makes it clear that "'data power' does not mean big data power, and while big data certainly helps to build data power, not all companies that use big data technologies have data power. In other words, data power is not just about using new technologies to gain new insights into data; it's about the control that some companies have over the flow of data between participants in the digital environment." "It is not enough to focus on concepts such as 'gatekeepers' or 'digital platforms' because they are too broad and do not address the notion of control over data, which poses potential regulatory problems. The focus on data power is therefore an attempt to emphasize the personal and broader digital ecosystem issues that may result from such control." It is very insightful that this discourse rationalizes the transformation of "data power" into "digital power," transforms a purely technical issue into a social one, and sees a "digital ecosystem" that is not visible only through the power of algorithms. "Isabel Hahn also points out three defining characteristics of data power: first, data power is ubiquitous in digital environments; second, massive and diverse data leads to data power's control over users; and third, data power refers to the ability to aggregate data across different data sets. That is to say, digital power has a universal control with extremely wide coverage and powerful data aggregation, is this not enough to constitute a kind of digital power that rivals traditional state power?

Secondly, digital technology is based on algorithms, but algorithms are not the whole of digital technology. With the continuous development of digital technology, more emerging technologies may arise in addition to algorithms. Based on the existing knowledge in the past, arithmetic power is computing power (Computing Power), i.e., the computing power to process information data and output predetermined results. Reasoning is computing, so algorithm is the method of reasoning. If the given premise is wrong, then the reasoning result will also be wrong. In recent years, with the emergence of AlphaGo, AlphaFold, especially after the introduction of Chat-GPT, such as the big model of language, artificial intelligence appeared "Emerge" (Emerge). According to Prof. Wu Fei, an AI researcher from Zhejiang University, the characteristics of AI can be roughly summarized as follows: it possesses the ability of "deep learning", it can utilize recurrent neural network models to deal with sequential data such as natural language, it can carry out "sequential learning", and it possesses the ability of "deep learning", and it has the ability of "deep learning". Transformer architecture with self-attention as the core, and supported by powerful computational resources, thus showing strong content synthesis ability. From this, we can see that the "model" created on the basis of the algorithm has the ability of content synthesis, which is not just the so-called computational or reasoning ability in the past. Therefore, Prof. Wu said that Chat-GPT pushes AI to a high level of content synthesis, "shapes a new paradigm of content production, and becomes a powerful means of intelligent digital interactions", and "promotes breakthroughs in areas such as language generation and conversational AI". ". Unlike the original algorithms or reasoning, Chat-GPT is able to reinforce its own learning techniques from human feedback, i.e., it "inputs human feedback on the content synthesized by the model as a kind of supervisory information in the communication, fine-tuning the model parameters to improve the authenticity and fluency of the linguistic patterns of responses." It is evident that Chat-GPT will engage in human-like learning, thinking, and conversations. Some have even boldly predicted today that AI has the power of comprehension (Grokking) and shows the ability to generalize to unseen data. In simple layman's terms, Chat-GPT does not only (rely on) algorithms, algorithmic technology is not the whole of digital technology, and thus algorithmic power is not the same as digital power.

Jurisprudence, as a conservative social science discipline, has always lagged behind technology. Although jurisprudence cannot synchronize with technology, it can predict the effect of technology application in advance. If predicted from a meta-universe perspective, the technology is not limited to digital technology, but also includes brain science and technology, cognitive science and technology and neurotechnology and many other technologies. Neurotechnology combined with smart technology will be a serious threat to human rights and freedom. Biotechnology and neurotechnology may not all be algorithmic technology, but biotechnology will definitely involve digital technology. It can be seen that algorithms are not the only technological power that poses an impact on the order of power in this world. In a way, we should still return to technology in the broad sense that the power brought about by the application of technology is all technological power. However, the thesis of this paper only examines digital technology.

Finally, before the algorithm is applied to the social scene, it is only a kind of technology, and does not constitute "algorithmic power". As some scholars have pointed out, the algorithm is obviously not power, the algorithm is a tool for people to exercise power, the algorithm needs to be combined with the purpose to form the algorithmic power. Only when the algorithm is applied to the relationship between people, with social attributes, it evolves the power with the characteristics of "power", and then form a digital social ecosystem. At this point, the "algorithm" is no longer pure technology, but transformed into a "digital power" in the core force, when the algorithm is mainly rely on the law to constrain. Although the algorithm is the core technical force in the digital power, but the digital power is not equal to or limited to "algorithmic power". Some scholars define algorithmic power as a procedural issue in law, which is reasonable. Because the opacity of an algorithmic black box is clearly a due process issue; the invasion of personal privacy can barely be considered a due process issue. However, some administrative law scholars have shown that algorithmic power is not just a procedural issue, but also a substantive issue. He argues that "algorithms are not instructions made by computers themselves, nor are they codes written by computers themselves; from the realization process, algorithms require human beings to set up rules, and then delegate technicians to write codes for code realization; therefore, algorithms have a personal nature." As early as 2001-2002, some scholars pointed out that the operation of the "digital government" has two main technical characteristics: the operation of the government's rigid procedural and rigid mode. However, the overall artificial operation of the "digital government" makes the government will still trample on or escape the "digital government" inherent rigid regulation and abuse of power. When the algorithm and administrative power depth combination, need to be regulated by law is not the object of the algorithm technology itself, but should be the algorithm of administrative power, that is, should be from the "power constraints" rather than "technical regulation" point of view of the study of algorithms. "Administrative algorithms need administrative organs to set rules, and then entrust technology enterprises or specific personnel to formulate algorithms, the subject of the behavior and the subject of responsibility behind the behavior are administrative organs." It can be seen that the algorithmic power formed by the combination of algorithms and state power is no longer just a procedural issue. If we only consider digital power as algorithmic power, and thus equate algorithmic power with procedural issues, then we will ignore the risk potential of many substantive issues, including whether the rules set in advance by administrative algorithms (assigning powers, rights, obligations and responsibilities) are justified, the monopoly created by AI in the financial and consumer fields, the infringement or control of public and personal security interests by algorithms, the usurpation of state power by algorithms and the sharing, bias in the administration of justice caused by algorithms, and miscarriages of justice in judicial decision-making.

Algorithmic power is only one part of digital power, which is stronger than the technological power of the past. Although the technological power in the era of the steam technology revolution had an impact on political power, it was far from the extent of the impact of digital power on social and political power today. In short, the author uses the term "digital power" instead of "algorithmic power" for the same reason. So, what kind of new type of power has digital technology brought? Some scholars in the review of British scholar Jamie Saskander's book "The Power of Algorithms: How Can Humans Survive Together? book, notes that "algorithms, as a machine formula and language, are shaping the underlying logic of codified law and digital new kinds of power." Digital technology is by far the technology that has enabled the greatest extension of human capabilities, and it is being used in a wide range of scenarios in life and production, profoundly changing human life. Digital technology has been most widely used in business operations and state management, and under the influence of digital technology and driven by commercial and public interests, commercial power and state power have been greatly extended. The combination of the "power" of digital technology and the administrative power of government governance has given rise to a new governance technology and model, which some scholars call "digital governance".

The designers and users of digital technology will substitute their own motives, purposes and values into the technology to make it a tool that meets their requirements, and then strengthen their influence and control over other subjects through the technology. Thus, the tool itself and the process of using the tool together synthesize digital power. Digital power is a de facto power that is self-created and self-empowered through the use of digital technology, and it is currently only a concept in the general sense of social science, not in the sense of jurisprudence, and even less in the sense of strict statutory power in the sense of positive law. However, science and technology, as a dominant force of a non-state nature, has become so powerful that, like power of a state nature, it can not only coercively dominate individuals, markets and societies, but also rival state power, and even the state is forced to utilize this power from science and technology. It can be seen that digital power has already possessed most of the characteristics of "power" in the legal sense, such as dominance, coercion, expansion, exclusivity and authority. At this point in the article, we can define "digital power": it is a one-way coercive power centered on digital design, computation and application generated by digital technology, and a universal technological ability to enforce the implementation of binding obligations on members or units of a collective organizational system through digital technology, and lead to the transformation of the digital ecosystem into a rule-of-law order. It is a universalized technological capacity to enforce binding obligations on members or units of a collective organizational system through digital technology and lead to the transformation of the digital ecosystem into a rule of law order. Digital power is participating in and deeply affecting this order of human rights, public power and law.

As technological applications enter society and "algorithmic power" becomes "digital power" with legal significance, the focus of jurisprudence has shifted, and I argue for the inclusion of "digital power" within the scope of the law. Therefore, I advocate that "digital power" should be included in the scope of law. To judge whether a "new power" in the sense of social science should be included in the scope of the power of jurisprudence, whether the power has a legal nature, and whether it can become a legal concept at the level of statutory law, the key is to see whether it is necessary to take more intensive special regulatory measures to regulate this new power, i.e., to elevate the usual regulation of enterprises to a higher form of regulation. The key is whether it is necessary to adopt more intensive special regulatory measures to regulate this new power, i.e. to upgrade the usual regulation of enterprises to a higher form of regulation, so as to establish a more optimized structure of the rule of law order. Specifically, the judgment is made on two levels: factual and normative. From the factual level, does digital power have the characteristics of power? Does it have a generalized social impact and not just on an individual? From the normative level, does digital power involve new relations of rights and obligations and the distribution of responsibilities? Is there a new definition of legal accountability? I will analyze these questions later.