The Central Cyberspace Administration of China supervises platforms such as Douyin, KUAISHOU-W, and WeChat Video Account to optimize and adjust algorithmic recommendation chaos
The Central Cyberspace Administration announced that since the launch of the special action "Clear and Bright - Governance of Typical Issues in Algorithmic Recommendations on Online Platforms," it has addressed risks such as the amplification of vulgar information through algorithmic recommendations, the exacerbation of "information cocoons," and increased viewpoint polarization, which have been strongly reflected by netizens. The Central Cyberspace Administration has urged key platforms to optimize their information recommendation algorithms and adjust their algorithmic rules accordingly. Key platforms have actively responded by signing the "Algorithm for Good" Nanning Declaration, improving content review for algorithmic recommendations, establishing dedicated websites, channels, or accounts to publicly disclose algorithmic principles, and developing innovative features such as "Cocoon Assessment" and "One-Click Break Cocoon," enhancing user interest preference management services, and increasing the diversity of algorithmically recommended content.
Recently, platforms such as Douyin, Xiaohongshu, Weibo-SW (09898.HK), KUAISHOU-W (01024.HK), Tencent's WeChat Video Account (00700.HK), and Bilibili-W (09626.HK) have systematically optimized and improved multiple functions around core aspects such as weighted recommendations for positive content, safeguarding user autonomy, optimizing the diversity of recommended content, and enhancing algorithm transparency.
In terms of enhancing algorithm transparency, they have publicly disclosed algorithmic operating rules, promoted transparent display work, and ensured users' right to know. Douyin has explained its recommendation logic, intervention mechanisms, and governance effectiveness to the public through the establishment of a "Safety and Trust Center" website and open day events. Weibo has improved the transparency of its trending search algorithms by publicizing listing rules and data rules, launching trending heat tags to indicate the driving factors behind the heat. WeChat Video Account has used simple and understandable graphics and videos to inform users, releasing "Understanding WeChat Video Account Algorithm Recommendations at a Glance" and "Algorithm Break Cocoon Series."
To break the "information cocoon," innovative features such as "Cocoon Assessment" and "One-Click Break Cocoon" have been launched to help users guard against the risks of "information cocoons." Douyin has fully upgraded its "Usage Management Assistant," innovating content preference assessment functions to visually present users' recent browsing content. Xiaohongshu has set up "Content Preference Assessment and Adjustment" and "Explore More" functions, allowing users to easily browse more diverse and rich recommended content with one click. KUAISHOU has increased the exploration of "positive," "useful," "warm," and "trustworthy" content through its positive energy algorithm, fully presenting it in algorithmic recommendations.
In terms of improving the review mechanism for recommended content, they have continuously enhanced the review mechanism for recommended content, strengthened the promotion of positive content, and prevented the recommendation of vulgar and harmful information. WeChat Video Account has improved its dual mechanism of "Friend Recommendations" and "Algorithm Recommendations," continuously iterating and upgrading its identification and crackdown models to strictly prohibit typical harmful information such as vulgar and lowbrow content from entering the recommendation pool. Douyin has innovatively launched a verification mechanism for hot topics to prevent malicious dissemination behaviors such as staged fakes, imitation for hype, and splicing.
In terms of user empowerment, they have continuously optimized interest preference management and content negative feedback functions, making it easier for users to autonomously adjust and optimize algorithmic recommended content.
A relevant person in charge of the Central Cyberspace Administration stated that the governance of information recommendation algorithms on key platforms has achieved certain results, but there are still issues such as the unsatisfactory effectiveness of related functions and the insufficient quality of recommended content, which fall short of the expectations of netizens and society at large. Algorithm governance is a long-term and systematic project. The cyberspace authorities will conduct regular inspections to supervise information recommendation algorithm platforms to continuously optimize algorithmic operating mechanisms and management rules, improve the quality of recommended content, explore diversified "break cocoon" paths, innovate user empowerment functions, and effectively safeguard the legitimate rights and interests of netizens
