[Chengze Observation · Platform Economy 40 Evaluation 33] Yang Ming: Algorithm recommendation should be matched with the platform responsibility

Zhongxin Jingwei September 8th. Question: Algorithm recommendation should be matched with the platform responsibility

Author Yang Ming, Professor of Law School of Peking University

Introduction: The essence of platform competition is data -driven. The algorithm can play an important role in platform competition through embedded business models. In the information ecosystem supported by data and algorithms, the platform responsibility should be determined, and people are still controversial. The view that "recommendation is the output of pure algorithms and is a collaboration" obviously cannot explain the operation of different information ecosystems. It should follow the principle of "risk and income matching" to achieve the mutual match between algorithm recommendation and platform responsibility.

We have deeply appreciated the digital survival what Nig Popci said, and the most touched one is the convenience, wealth, and uncertainty of technology brought by technology, which has caused human society to be cognitive and cognitive and cognitive and Separation between progress. The reason why people hate risks is because of fear of unknown. In the past, risks exist objectively, but whether it happened and when it was unknown. Even so, people would have hesitated during behavior decision -making, lest they made a choice that caught themselves into the most unfavorable situation.到了数字时代，风险是否客观存在都具有不确定性了，这是因为，人们对于隐藏在商业活动背后的技术甚至都难以察觉，所以，一旦这些技术被揭示出来，无论是否会带来风险，人们It is criticized for hidden (not technology).

Algorithm recommendation is a focus of people's concerns about uncertainty in the field of platform economy. The controversy of the newly -received attention is that the video platform uses the algorithm to recommend videos containing specific content, which has aroused a complaint about the relevant copyright owner to launch copyright infringement on the platform. For example Copyright violated the lawsuit on the grounds of "today's headlines". The core controversy of the case is what the video platform has the obligation to pay attention to the compliance of the audiovisual works that spread. Although there are still a few people who claim that the platform should not be "overweight" on the issue of copyright governance on the grounds of technical neutrality, more people have realized that it is meaningless The application layer to analyze whether the platform has a technical neutral status, and then discuss whether the platform can be relieved of responsibility because of "technology cannot".

To discuss whether technology is neutral, that is, we are familiar with the economics about network neutrality. data. In this mode, the content provider and the platform operator cannot be differentiated to treat different content through transactions. So naturally, in addition to the linity of the platform with the obligation to ensure operational security, the platform does not have the integration of the content of the content spread on the platform. Obligations of regular burden. Therefore, network neutrality should not be associated with the platform. If the platform has the ability to control the content of the content, it does not have a technical neutral position.

Regarding the obligation to protect the copyright in terms of copyright, the Supreme People's Court adopted the "Regulations on Several Issues of the Application of Laws of Civil Disputes on Civil Disputes on the Civil Disputes of the Information Network Communication Rights in November 2012" clearly stipulated: " If the network service provider induces words, promotion technical support, reward points, etc., and encourage network users to implement the right to infringe on information network dissemination, the people's court shall determine it to constitute an institutional infringement. "However, the" promotion technology "in the regulations shall Can it be recommended by algorithms? There are still controversy in all walks of life. It should be seen that this clause uses "induction and encourage network users to implement infringement behaviors" to limit the use of "promotion technology", which makes the interpretation of "promotion technology" cannot simply use the abstraction, strong tolerance of the concept of "technology" concept Sex to expand at will. Judging from the era of judicial interpretation, at that time, the business model of the Chinese platform had just had an application recommended by the algorithm, and the recommended algorithm based on deep learning was blooming in the platform economy. The problems caused by the annual judicial interpretation cannot handle the recommendation of algorithms. In fact, we define the definition of "Algorithm Recommendation Technology" in Article 2, paragraph 2 of the "Internet Information Service Recommendation Management Regulations "-" use the generating synthetic class, personalized push class, sorting selection, searching and filtering categories , Schedule decision -making and other algorithm technologies to provide information to users "-the is not difficult to understand this.

However, it is worth thinking about that if the application recommended by the algorithm does not give the platform more attention obligations, will this have the effect of reverse incentives due to violation of the principle of "risk and income matching"? Not only is copyright infringement liability, but inviting other property rights, as well as privacy, personal information rights, etc., the platform responsibility system should be configured with the same logic. The author believes that the thinking of platform responsibilities must closely follow the relationship between "control of Internet activities" and "allocation of legal obligations". In other words, the stronger the operator's control of consumers to participate in business activities, the higher the legal obligations it should be. In the algorithm world recommended by the collaborative, on the surface, network users enjoy greater decision -making freedom because of more information provided by the platform, or it can improve decision -making efficiency, but in essence The shaping is more and more solidified. We think that our needs are increasingly responded to platform operators. We do n’t know that we are actually caught in the platform business model and cannot extricate ourselves. Whether it is an information cocoon room or a digital training, it is a subtle description of this platform control. In view of this, the basic principle of platform responsibility is inappropriate to adhere to the traditional "shelter" rules. There is no doubt that people are always in a specific information environment to determine their behavior. In other words, people's behavior is shaped by the information ecosystem they are in the information ecosystem. The network effect we are familiar with the concept of economics is an excellent interpretation of the relationship between behavior and information systems. The data -driven platform is both a "activity" place for users and the information ecosystem that its decision -making. The high and low decision -making efficiency have a significant effect on the guidance of user traffic, and traffic is the fundamental source of platform profits. Therefore Competition has become a more capable of locking users for the information ecosystem that competes. In the context of information overload, using algorithms to match the preferences or heterogeneous needs of network users has now become an important way for platform competitors to compete for traffic, but we must also see that data drivers and algorithm matching and user heterogeneous nature The inter -interacting relationship is not just one -way. While the deviation and alienation guidance algorithm continue to evolve, the algorithm is also shaping the user deviation and alienation, that is, the algorithm and the user are "mutual". The algorithm recommendation is recommended It is essentially an artificial control. Therefore, the view that "recommendation is pure algorithm output and a collaborative" obviously cannot cover the information ecosystem of different operating mechanisms in practice.

Economic thoughts that "match risks and income" can also find theoretical mapping in traditional civil law in mainland law. The absolute empowerment (one of the neighborhoods), except for the value of the constitution and the value of the private law, their inherent logic is also the matching of the relationship between control and rights and obligations. As a result, the key to the problem becomes a specific algorithm embedded in the business model, whether it means that it has a leading power on the commercial activities on the platform, or whether it is the organizer of those activities. The questioner believes that if the platform does not purchase content and does not produce content, even if it participates in the sharing of content dissemination income, it is still not appropriate to determine that such a platform is an organizer of content dissemination activities and has no control of communication behavior; from the technical level; from the technical level, In terms of the current algorithm used by the platform, the collaborative algorithm is recommended. Technically, it is a weighted recommendation and eventually manifested as user collaboration.

In response to this question, we might as well start with the essential theory recommended by the algorithm: the so -called recommendation refers to the behavior of the probability of establishing a contacted (or interaction) to enhance the consumer to enhance the consumer; the algorithm; the algorithm; the algorithm; The recommendation is to train through the original data, and then manually label to complete the sorting, and finally show the results. The manual labeling method and result depends on the characteristic vectors and similarity set by the algorithm. It can be seen that the application of the platform's recommendation of the algorithm aims to help network users solve the problem of information overload, but it is obviously solved in the direction preset by the platform. The more preset feature vectors, the higher the requirements of similarity, the deeper the platform participation in the communication activities. Therefore, it can be said that the algorithm has reshaped the relationship between people, which has led us to even understand the algorithm from the level of social structure change.

Of course, the artificial control of the platform operator through algorithm recommendation is positively related to the lock effect of the platform information system to the platform information system, and the algorithm is designed. It is the result of artificial control. In view of the dynamic change of platform control and its lock effect, the attention obligations that the platform has the attention should not be fixed. "Rules of Wind Harbor". Clarified this point, we can conclude that through the analysis of the algorithm recommended by the platform in the case on its "control of Internet activities", we can conclude that the platform uses the algorithm to work hard to promote traffic tilt. When the effect, it needs to bear the corresponding attention obligations for this reason. The greater the influence of the algorithm on the flow of flow, the heavier the attention obligations of the platform. Therefore, the measurement of the platform's attention obligations in the case involves the analysis of the algorithm itself, so as to evaluate the algorithm's recommendation to how much the traffic tilt has played. In order to meet the needs of judicial practice, it is imminent to implement algorithm audits. Conclusion:

The development of the platform economy is to continuously cause contradictions between the publicity and exclusion of the platform, while promoting the continuous improvement of social governance structure and system to resolve these contradictions. The current algorithm era is very different from the popularity of the Internet industry, which is popular in the "Rules of Wind Harbor", and adheres to the configuration of the platform responsibility system of that era. The application of the platform's recommendation of the algorithm is to reduce the cost of information searching while also improving the quality of people and things. This is not only a problem of innovative business models. The business model is stripped out and evolved into a new platform. If we see it without seeing it, we always think that the platform does not participate in the business activities that occur on it, lack of management and leadership. On the surface, it inspires the development of the platform to promote the development of algorithms. Coordination of regulation. The significance of the information ecosystem mentioned repeatedly in the previous article is to highlight this cognition. (Zhongxin Jingwei APP)

This article was selected by the Sino -Singapore Jingwei Research Institute. The copy of the work produced by the selected work, the copyright of the work, is not authorized by any unit or individual. The views involved in the selected content only represent the original author and do not represent the view of the Sino -Singapore Jingwei.

Editor in charge: Song Yafen

[Chengze Observation · Platform Economy 40 Evaluation 32] Yu Changhua: How does the data economy promote economic growth?

[Chengze Observation · Platform Economy 40 Evaluation 31] Xu Yuan, etc.: evolution of antitrust theory and platform governance

[Chengze Observation · Platform Economy 40 Evaluation 30] Huang Zhuo: The future direction of the innovation and development of China Fintech Platform

Pay attention to the official WeChat public account of JWVIEW (JWVIEW) to get more elite financial information.

- END -