个性化推荐算法的多维治理

Multi-Dimensional Governance of Personalized Recommendation Algorithm

个性化推荐算法引发干扰用户自主决定、影响未成年人健康成长和“回音室”效应等诸多问题。作为因应之道的算法治理应当在三个维度上展开:针对算法使用人与相对人的个人自治之维、针对算法内在设计的行为规范之维、针对算法应用依赖的平台数字生态系统的主体规范之维。《个人信息保护法》确立以权利义务为核心的个人自治规范框架,通过保障个人的知情权与决定权,增强个人的自我管理能力。在个人自治失灵的场景中,法秩序应当直接规范算法程序,在算法设计层面保护算法相对人的意思形成自由与人格发展自由。个人自治之维与行为规范之维的顺利展开离不开外在的监督与管理。守门人规则开拓了主体规范维度,通过为平台赋权并对平台施加必要的监管,要求平台尽最大努力监管平台内的算法活动。

Personalized recommendation algorithm has triggered several problems,such as interfering with user’s free choices,affecting minors’healthy growth as well as“echo chamber effect”.As the response,algorithm should be regulated in three dimensions:the dimension of personal autonomy for algorithm users and counterparts,the dimension of behavioral regulation for the inherent design of algorithms,and the dimension of participant regulation for the digital ecosystem of platforms on which algorithms are applied.Personal Information Protection Law establishes the normative framework of individual autonomy with rights and obligations as the core and enhances individual self-management ability by ensuring individual’s right to know and decision-making.In several scenarios where personal autonomy fails,the legal system should directly regulate algorithm itself,protecting the freedom of mind and personality development at the level of algorithm design.The development of the personal autonomy dimension and the behavior regulation dimension cannot be achieved without external supervision and management.The gatekeeper rule helps to develop participant regulation,and requires the platform to do its best to supervise the algorithm activities in the platform by empowering the platform and imposing necessary supervision on the platform.