On January 4, China's algorithm recommendation policy was clearly implemented. The Provisions on the Administration of Internet Information Service Algorithm Recommendation (hereinafter referred to as the "Provisions") promulgated by the NFC China Official Account, jointly formulated by the Cyberspace Administration of China, the Ministry of Industry and Information Technology, the Ministry of Public Security and the State Administration of Market Regulation, were deliberated and approved by the Cyberspace Administration of China on November 16, 2021, and promulgated with the consent of the Ministry of Industry and Information Technology and other four ministries and commissions, and will come into force on March 1, 2022.
The Provisions require that algorithm recommendation service providers should provide users with options that do not target their personal characteristics, or provide users with a convenient option to turn off algorithmic recommendation services. Where a user chooses to turn off algorithm recommendation services, the algorithm recommendation service provider shall immediately stop providing the relevant services.
A number of industry insiders interviewed by the first financial reporter said that as the first departmental regulation focusing on algorithm governance in China, the landing of the "Provisions" gives users more independent choices and corrects the tendency of maximizing the commercial benefits of the platform in the past. Behind it is the continuation of the Anti-Unfair Competition Law, the Data Security Regulations, the Personal Information Protection Law (hereinafter referred to as the Individual Protection Law), and the Law on the Protection of Minors, which also means that in 2022, China will enter the first year of algorithm supervision.
The first year of algorithm regulation
In recent years, while the application of algorithms has injected new momentum into political, economic and social development, the problems caused by the irrational application of algorithms such as algorithm discrimination, "big data killing", and inducing addiction have also profoundly affected the normal communication order, market order and social order, and brought challenges to the maintenance of ideological security, social fairness and justice, and the legitimate rights and interests of netizens. The introduction of targeted algorithm recommendation rules and regulations in the field of Internet information services is the need to prevent and resolve security risks, and it is also necessary to promote the healthy development of algorithm recommendation services and improve the level of supervision capabilities.
The "Provisions" make it clear that the application of algorithm recommendation technology refers to the use of algorithm technologies such as generative synthesis, personalized push, sorting selection, retrieval and filtering, and scheduling and decision-making to provide information to users. Zhang Xiaorong, president of the DeepIn Science and Technology Research Institute, told the first financial reporter that whether it is a website or a mobile APP, it belongs to the scope of algorithm regulations. Focusing on the situation, those greatly affected by the "Regulations" include e-commerce such as Pinduoduo and Taobao, content categories such as Toutiao Douyin, Baidu search and other industries. The algorithm provisions cover information service specifications, user rights and interests protection, supervision and management, legal responsibility and other aspects, and have played a positive role in promoting the orderly development of the industry to safeguard user rights and interests.
Guan Yuqing, founder of Lens Company Research, told the first financial reporter that the "Rules" is the first algorithmic regulatory rule in China with clear boundaries and operational guidelines. Although there have been provisions in the industry such as not allowing big data to be killed, it is relatively general and there is no clear regulatory rules to support. In the view of Quan Yuqing, this development law is the inevitable result of the application of the big data industry to a certain extent - the law of industry development is that commercialization is before supervision, and this is the case in any innovative industry.
Wang Xinrui, a partner at Shihui Law Firm, called 2022 the "first year of algorithm supervision", saying that both China's Personal Information Protection Law and the EU's GDPR have made principled provisions for automated decision-making, and laws and supporting rules such as the E-commerce Law and the Anti-Monopoly Law have also made principled restrictions on the killing of big data and the abuse of algorithms to eliminate competition. However, the above-mentioned legal rules, whether they regulate acts or results, are currently very abstract and not strongly operable.
In the past few years, algorithm technology has actually been applied to the commercial aspect, serving the interests of the platform side and the enterprise side, and even continues to have problems such as riders trapped in the algorithm system, big data killing, rice circle chaos, information cocoon room, and user privacy leakage. Feng Qi, founder of Marteker, told the first financial reporter that the algorithm, as a technology itself, has not played a real role in "science and technology for good", and even caused harassment to consumers. The landing of the "Provisions" is essentially a good thing, allowing advertising and marketers to "dance with shackles", allowing the industry to grow from barbaric to having laws to follow, allowing the platform to focus more on protecting the rights and interests of users, and launching the content and services that users really need.
Of course, the platform needs a certain adaptation period, Feng Qi said, for example, through the federal learning technology multi-dimensional reference, extract commonalities from the group; in addition, through content marketing, KOL marketing, advertising implantation, etc., it also has better results.
Be recommended, don't be hegemonic
A person who has long been concerned about the Internet industry told the first financial reporter that the key choice of supervision is at the algorithm level, a major reason is that the algorithm corresponds to the rights and interests of users - the content of the "Provisions" mentions that the Network Information Department will establish an algorithm classification security management system together with relevant departments such as telecommunications, public security, and market supervision, and at the same time recommend services according to the public opinion attributes or social mobilization capabilities, content categories, user scale, The degree of data sensitivity processed by algorithm recommendation technology and the degree of intervention in user behavior, and the classification and hierarchical management of algorithm recommendation service providers, to a certain extent, reflect the legislative concept of accurate governance.
Among them, the person said that the "public opinion attribute" includes user comments and speeches, and "social mobilization ability" refers to the user base served by the platform. The biggest feature that distinguishes Internet marketing from conventional marketing is the user portrait, and the purpose of the algorithm is to find the most accurate measurement standard.
Previously, the industry had the view that "algorithms have no values", which triggered a debate. However, with the expansion of the scale of the platform, the algorithm values are also passively attached to the tendency of the platform. "There are tendentious goals first, and then there are tendentious algorithms. Good-for-good services provide 'bad' algorithms," said the person who said.
In recent years, taking the financial industry as an example, in order to obtain financial services from platform companies, Chinese consumers often need to provide them with personal information, which in turn creates the problem of excessive data collection, such as the purchase and sale of borrower information during the rapid growth of "cash loans" from 2016 to 2017.
Previously, at the third Bund Finance Summit, Zhang Xiaohui, senior researcher of the China Finance Forty Forum and dean of tsinghua University's Wudaokou School of Finance, said that the complexity of algorithms and the deliberate concealment of algorithm users make it impossible for the vast majority of people to understand the working principle of algorithms, resulting in large technology companies as algorithm users, especially those that hold almost all digital platform companies related to personal life behavior, to be in a de facto dominant position, forming "algorithm hegemony". This seriously endangers the legitimate rights and interests of the algorithm counterpart, that is, consumers.
Li Keshun, president of the Guizhou Data Treasure Product Research Institute, told reporters that a more obvious impact of the landing of the "Provisions" is that some videos or commodities are exposed through algorithms with high intensity, so that consumers' excessive consumption and even the behavior of making wrong consumption decisions are curbed. Bad content or goods will not be able to obtain algorithmic dividend bonuses, and the growth rate of non-"compliant" revenue of the platform will slow down.
Of course, Li Keshun believes that personalized recommendation is only one of the aspects of algorithm recommendation, from a technical point of view, the algorithm is divided into two categories of strong interpretability and weak interpretability, the former is easier to review, so at this stage there is actually no very perfect algorithm standard transition program, so that consumers grasp the initiative of whether personalized recommendation is one of the solution options.
Wang Xinrui said that the algorithm itself is not a new technology, but some social problems caused by the application of algorithms have attracted increasing attention from all walks of life in recent years. The Provisions respond to these hot social issues, especially for algorithm discrimination, big data killing, and inducing addiction. Algorithms are more technical than data, less easily described and perceived, and therefore more difficult to develop and enforce rules.
What does the industry "count"?
The negative impact of algorithm abuse on the industry and consumers has long been secretly growing.
Previously, Weibo issued an announcement that it would increase efforts to deal with the hot search list hot topic list brushing behavior, and punish the direct beneficiaries of the list brushing behavior. In order to ensure the authenticity and credibility of the hot search list and the hot topic list, the algorithm of the product itself is constantly upgraded, and an anti-brushing system based on big data recognition is set up. For behaviors with cheating characteristics, the system will identify and intercept them in real time.
However, according to the reporter's understanding from the entertainment industry and people close to Weibo, there are a large number of supplier enterprises outside Weibo, which can guide public opinion through methods such as trumpet brushing.
At that time, the core of Weibo adjustment was to introduce the mode of manual intervention by editors, take algorithm mining as the basis, abandon pure algorithms in sorting and selection, and introduce manual intervention by editors on content that violates relevant laws and regulations, information with negative social energy, and information that is excessively entertaining.
In addition to Weibo, the e-commerce and short video industries are also making self-adjustments during the gradual landing of algorithm supervision. In October 2021, Tmall and JD.com encrypted sensitive information involving individuals in consumer orders during the delivery process, and software service providers and merchants could no longer obtain recipient privacy information from the Taobao development platform or order promotion services.
After the promulgation of the "Data Security Regulations", WeChat updated the "Privacy Protection Guidelines", adding "video number, public account, take a look, search and search personalized recommendation or display mechanism" to increase the personalized independent control path. Platforms including Douyin, Kuaishou, Today's Toutiao, B Station, and Xiaohongshu all completed the update of their privacy policies in early November 2021. The reporter's experience found that both Douyin and Kuaishou set the personalized advertising selection button in a prominent position.
Focusing on the short video and e-commerce industry mainly affected by the algorithm policy, Li Keshun told the first financial reporter that in short videos and other scenarios, the algorithm will be biased towards the big V account, rather than the high-quality content trumpet, forming a traffic monopoly behavior within the platform, which will directly affect the user's fair access to information and the right to know.
At the same time, the traffic of the platform will also be biased towards some of the head e-commerce numbers, and the products of these account merchants may not be of high quality, but under the continuous bombardment of content algorithms, users are easily guided and then make abnormal purchase behaviors, and even form a common impulse consumption.
He Yanzhe, director of the review department of the Information Security Research Center of the China Electronics Technology Standardization Institute, told the first financial reporter that short videos are pushed according to user preference data, and it is necessary to evaluate the push content, whether it will lead to real addiction. He Yanzhe believes that it is time to put on the brakes on the push of short video information, prevent the information cocoon effect from being too heavy, replace more space for consumers' choice, and guide the transparency of algorithms and the right of choice of users.
He Yanzhe said that previously the platforms have been revised according to the "Personal Protection Law" on the personalized recommendation function, combined with the current "Provisions", the platform needs to give users more choices, the past recommendation personalization is often to maximize commercial interests, rather than the real stand to the user's point of view to provide personalized high-quality push services.
The Provisions clarify the requirements for the protection of the rights and interests of algorithm recommendation service providers, including ensuring the algorithm's right to know, requiring users to be informed of their provision of algorithm recommendation services, and publicizing the basic principles, purposes and intentions of the services, and the main operating mechanisms; to ensure the right to choose algorithms, users should be provided with options that are not tailored to their personal characteristics, or the option to conveniently close algorithm recommendation services. Where algorithm recommendation services are provided to minors, the elderly, workers, consumers, and other such entities, the Provisions require that algorithm recommendation services must not be used to induce minors to become addicted to the Internet, and that the elderly shall be facilitated to safely use algorithm recommendation services.
However, in the face of algorithms, how individuals supervise and present evidence is still weak, and the improvement of the industry itself is also slightly.
![Algorithms and pricing: The underlying logic of "big data killing" 丨 Wei Yan Yi[fig]](data:image/gif;base64,R0lGODlhAQABAIAAAP///wAAACwAAAAAAQABAAACAkQBADs=)