REGULATING ADDICTIVE ALGORITHMS IN CHINA:LEGALITY AND PRACTICALITY  

算法沉迷规制的中国路径

作　　者：Peng Chun 彭錞(Peking University Law School)

机构地区：[1]Peking University Law School

出　　处：《China Legal Science》2022年第6期114-133,共20页中国法学（英文版）

基　　金：supported by the 2020 Grant from National Social Sciences Fund of China(No.20CFX019)。

摘　　要：In the Rule on the Management of Algorithmic Recommendation by Internet Information Services of 2022,China has established a general ban on addictive algorithms,which is for the first time in the world.This ban can be formally justified with reference to superior legal rules that require online service providers to respect social morality and substantively justified on the legal ground given its risk and harm towards individual users’wellbeing,social welfare and democratic government.A parallel can be observed between addictive algorithms and addictive drugs and gambling that are prohibited in China’s mainland.However,the ban on addictive algorithms shall not be construed as a ban on all algorithms because algorithms are now the bread and butter of the information economy.It is therefore imperative to draw a line between addictive algorithms and attractive algorithms.However,this important distinction cannot be demarcated in the abstract and ex ante,for the reason that addictiveness is highly subjective and that the government is deficient in sufficient technological capacity to catch up with the rapidly changing world of algorithms.Instead of a traditional command-and-control approach,meta-regulation that relies on technical sludge and legal accountabilities is proposed.The algorithmic recommendation service providers shall be asked to put in place addiction-reducing mechanisms in their recommendation services to help the users to combat addiction.And the providers shall be held accountable through the users’right of refusal,the duty to explain and potentially administrative penalties such as warnings and fines.根据2022年出台的《互联网信息服务算法推荐管理规定》,中国在全球首次设立对成瘾性算法的一般性禁止。此项规定形式上可基于上位法关于网络服务提供者遵守社会公德的要求而获得合法性依据,实体上则可基于成瘾性算法对个体用户、社会公益以及民主政体的风险和危害而获得正当性证成。成瘾性算法可与毒品、赌博相类比,后者在中国被全面禁止。但对成瘾性算法的禁止不应解释为对算法的全面禁止,因为算法如今已成为信息经济正常运转的命脉。因此,有必要在成瘾性算法和吸引性算法之间划出界线。这一重要界线无法在事前、抽象给定,因为成瘾性本身高度主观,而且政府欠缺充足的技术能力应对高速发展的算法技术。与其固守传统的命令-控制型管理模式,不如采用依靠技术助推和法律问责的元规制治理模式。算法推荐服务者应设置防沉迷机制以帮助用户避免成瘾,同时也应借由用户的拒绝权、可解释权以及行政处罚等方式实现法律问责。

关 键 词：行政处罚 互联网信息服务 网络服务提供者 成瘾性 社会公德 民主政体 社会公益 上位法 

分 类 号：TP391.3[自动化与计算机技术—计算机应用技术] D922.16[自动化与计算机技术—计算机科学与技术]