信息通信技术嵌入公共组织给其组织制度以及个体行为带来了深刻影响。在已有的文献中，研究者已经对组织制度的变化进行了广泛的探讨，但对个体行为的研究则相对欠缺。特别是针对技术嵌入对街头官僚自由裁量权的影响，现有成果的观点存在明显的争议。同时，本研究也观察到，技术嵌入一方面提高了街头官僚的工作效率，另一方面也导致了街头官僚灵活性降低、行政执法合法性缺失、公平性欠缺等问题。因此，研究街头官僚自由裁量权在技术嵌入的过程中所受到的影响，是一个具有重要性和紧迫性的议题。为了应对实践中的挑战，也为了回应理论上的争议，本文围绕“技术嵌入如何影响街头官僚的自由裁量权”的研究问题，开展了一项实证研究，尝试以街头官僚的自由裁量权为切口，解读街头官僚在数字时代受到的影响。

本文以技术嵌入理论为基础建构了一个整体分析框架，该框架包含“结构现状”“社会互动”和“结构重塑”三个阶段。其中，“结构现状”分析了现阶段技术嵌入街头官僚自由裁量权的现状，即技术嵌入模式的分类；“社会互动”分析了技术嵌入如何影响街头官僚的感知自由裁量权以及他们的自由裁量权行使意愿；“结构重塑”分析了重塑后的街头官僚自由裁量权。基于这个分析框架，本文分别采用了案例研究、问卷实验和归纳思辨的研究方法。

该框架涉及到两个重要概念。一个是“技术嵌入”。此前的相关研究往往采用“技术应用”的概念并将其视为一个二分的分类变量，因此这些研究者只对比“无技术应用”和“有技术应用”之间的差别。本文受西蒙有限理性理论的启发，以“技术嵌入”取代“技术应用”作为自变量，并创新性地将此概念操作化为一个包含了无技术嵌入、数据型嵌入、方案型嵌入和偏好型嵌入四种技术嵌入模式的变量。另一个是“自由裁量权”。本文基于技术嵌入理论，将自由裁量权解构为“感知自由裁量权”和“自由裁量权行使意愿”两个层面的可操作化变量。

本研究通过来自文献的证据，将研究场景确定为警务行政执法，并且对北京市两个区以及河南省一个市的警务行政执法工作进行了实地调查，最终将问卷调查的场景确定为交通执法，将街头官僚确定为交通警察，同时结合交警在交通执法工作中的技术嵌入实践，将技术嵌入模式分类调整为无技术嵌入、数据型嵌入和方案型嵌入。基于交警真实的日常工作内容，本研究设计了一个组内调查实验的问卷，其中包含三个实验场景，分别代表三种不同的技术嵌入模式：无技术嵌入、数据型嵌入和方案型嵌入。该问卷被首先进行了预调研（N = 41），经过修改完善后进行了二次预调研（N = 62）并开展了正式调研（N = 190），调研获得了三种技术嵌入模式下街头官僚自由裁量权的有效数据并被用于后续分析。

数据分析结果验证了部分研究假设，包括：（1）技术嵌入对街头官僚的感知自由裁量权存在先递增后递减的作用，即感知自由裁量权出现了先升高后下降的趋势；（2）技术嵌入对街头官僚的自由裁量权行使意愿也存在先递增后递减的作用，即自由裁量权行使意愿也出现了先升高后下降的趋势，且感知行政问责在其中具有部分中介效应。其中，技术嵌入对中介变量感知行政问责同样存在先递增后递减的作用，感知行政问责与自由裁量权行使意愿之间存在正向关系；（3）技术嵌入对街头官僚的感知流程复杂性具有负向作用。

基于以上问卷实验结果，本文进一步探讨了街头官僚自由裁量权所受到的来自技术嵌入“弧形效应”（即先递增后递减的作用）的影响及其影响机制。“弧形效应”的影响印证了目前部分研究者的观点，即数据型嵌入能够赋能街头官僚的感知自由裁量权，而方案型嵌入会削弱街头官僚的感知自由裁量权，同时也补充了研究者对于街头官僚自由裁量权行使意愿的认知。在分析技术嵌入对街头官僚感知自由裁量权的影响机制时，本研究发现，技术嵌入模式并非是解释感知自由裁量权变化的唯一要素，即技术嵌入模式和技术嵌入决策程序的方式共同决定了技术嵌入对街头官僚感知自由裁量权的影响机制。同时，本研究也提出认知中介模型解释了街头官僚自由裁量权行使意愿是如何受到感知行政问责影响的。

基于技术嵌入后街头官僚自由裁量权的变化，本研究发现，街头官僚的工作方式正在转向“行政执法代码化”，而街头官僚也向兼任政策执行者和决策审核者转变。“行政执法代码化”意味着行政执法过程通过技术嵌入对现实世界的编码来实现，即街头官僚通过影响一个“编码的世界”来治理“真实的世界”。这一方面提高了治理效率，另一方面也面临着“编码的世界”机器执法和“真实的世界”人性化执法的矛盾。“街头官僚兼任政策执行者和决策审核者”意味着由于街头官僚的决策权及其信息控制能力受到技术嵌入的挤压，因此街头官僚受到的垂直控制在加强，其作为政策创新贡献者的角色在削弱，并逐渐向政策执行者和决策的审核者转变。

The application of information and communication technology in public organizations has had a profound impact on their organizational systems and individual behaviors. In existing literature, researchers have conducted extensive analysis and discussion on the changes in organizational systems, while research on individual behaviors is relatively lacking, especially on the impact of technological embeddedness on the discretion of street-level bureaucrats and there is a clear dispute among researchers. At the same time, in social governance practice, we have also observed that technological embeddedness in grassroots governance has improved the work efficiency of street-level bureaucrats on the one hand, but on the other hand, it has also caused problems such as reduced flexibility of street-level bureaucrats, lack of legitimacy in law enforcement, and lack of fairness. Therefore, studying how discretion of street-level bureaucrats is impacted in the process of technological embeddedness is an important and urgent issue. In order to respond to the challenges faced in practice and to integrate and unify researchers’ understanding of the above issues, this article conducts an empirical study around the research question of “how technological embeddedness affects the discretion of street-level bureaucrats”,  and try to interpret the evolutionary forms of bureaucratic systems in the process of digitalization through the change of discretion of street-level bureaucrats.

This article constructs a comprehensive analytical framework based on the theory of technological embeddedness. The framework consists of three stages: “structural conditioning”, “social interaction”, and “structural elaboration/reproduction”. The “structural conditioning” stage describes the current state of technological embeddedness, focusing on the mainstream applications of technology in the government. The “social interaction” stage explains how technological embeddedness affects the discretion of street-level bureaucrats and their willingness to exercise those powers. The “structural elaboration/reproduction” stage discusses the impact of technological embeddedness on organizational institutions and the evolutionary trends of bureaucratic systems. Based on this framework, the article employs case studies, survey experiments, and normative research methods for the three stages.

The framework also involves two important variables, which are the independent variable and dependent variable of this study. Previous literature often operationalized “technological application” as a binary variable, thus limiting the comparison to the differences between “without technological application” and “with technological application”. In this article, however, by integrating Herbert Simon’s bounded rationality theory, this concept is replaced by “technological embeddedness” and innovatively operationalized as a variable that includes four technology embedding modes: technology-free embedding, data-based embedding, solution-based embedding, and preference embedding , allowing for the comparison of different types of technological embeddedness within a framework. Meanwhile, based on the technological embeddedness theory, we also deconstruct discretion into two dimensions of dependent variables: perceived discretion and willingness to exercise discretion.

Through literature analysis, the study identifies administrative enforcement as a suitable research scenario. Field investigations were conducted on administrative enforcement work in two districts of Beijing and one city in Henan Province. The study found that traffic police officers apply advanced big data and artificial intelligence technologies in their traffic management work. Then, based on the survey results, the technology embedding mode was adjusted to three types: technology-free embedding, data-based embedding and solution-based embedding. Based on the actual work content of traffic police officers, the study designed an intra-group survey experiment questionnaire, including three experimental scenarios representing three different types of technological embeddedness: without technological embeddedness, data-driven embeddedness, and program-driven embeddedness. The questionnaire underwent preliminary survey (N = 41), followed by modifications and a second preliminary survey (N = 62), and then a formal survey (N = 190) was conducted, generating valid data for subsequent analysis. The data analysis results confirmed several research hypotheses, including: (1) there is an inverted U-shaped relationship between technological embeddedness and the perceived discretiona of street-level bureaucrats; (2) there is an inverted U-shaped relationship between technological embeddedness and the willingness of street-level bureaucrats to exercise discretion, with felt administrative accountability partially mediating this relationship; (3) technological embeddedness has a negative effect on the perceived procedural complexity for street-level bureaucrats.

Based on the above research findings, this paper further explores the impact and mechanism of the “arc effect” (i.e., inverted U-shaped effect) of technological embeddedness on the discretion of street-level bureaucrats. This impact confirms the views of some researchers that data embedded technology can empower the perceived discretion of street-level bureaucrats, while solution-based embedded technology will weaken the perceived discretion of street-level bureaucrats, and also supplements researchers’ understanding of the willingness to exercise discretion of street-level bureaucrats. However, after further exploration, it was found that the characteristics of technology are not the only factors explaining the changes in perceived discretion. Therefore, in the process of analyzing the mechanism of the impact of technological embeddedness on the perceived discretion of street-level bureaucrats, we introduced a new factor, namely the type of technological embeddedness in organizational procedures, and proposed the “technology metaphor effect” as an explanatory model for the mechanism of the impact of technological embeddedness on the perceived discretion of street-level bureaucrats, and also proposed the “cognitive intermediary effect” model to explain how the willingness to exercise discretion of street-level bureaucrats is affected. Based on the changes in the discretion, we further analyzed the transformation of the work and role of street-level bureaucrats. We found that the work style of street-level bureaucrats is shifting towards “code-based law enforcement and management”, and the role of street-level bureaucrats is gradually transitioning to policy executors and decision auditors. The former indicates that the law enforcement and management process is realized through the coding of the real world by digital systems, that is, street-level bureaucrats govern the real world by influencing a “coded world”, which on the one hand improves governance capabilities, governance efficiency and professionalism, on the other hand, it also faces the contradiction that the “coded world” does not match the real world. The latter indicates that due to the compression of the decision-making power and information control capabilities of street-level bureaucrats, the vertical control toward street-level bureaucrats is strengthened, their role as contributors to policy innovation is weakening, and they are gradually transitioning towards policy executors and decision auditors.