基于参与式观察和田野访谈，文章考察了外卖平台中微观—宏观权力双向建构与循环增强的算法治理机制。通过对骑手个体进行微粒化的数据解析，算法奠定了微观权力运作的基础，为治理提供了对象和原料，进而依据宏观治理目标，将具有相似特征的微观数据聚合成适配不同配送情境的“微人口集群”。宏观调配的有效性依赖于算法指令对微观个体行为的精准引导，而面对规模化与差异化的骑手个体，算法借助空间层面的宏观规划，实现指令在特定区域内的精准投放。同时，平台通过数据拜物教与组织驯化等手段，促使骑手将宏观规则内化为自觉行动，实现宏观权力向微观主体的渗透。最终，在“算法黑箱”这一融合知识生产、真理确立与权力执行的“知识—权力”装置中，宏观与微观权力实现了闭环运作，使算法治理呈现出一种技术必然性。

Based on participatory observation and field interviews, this study examines algorithmic governance in food delivery platforms, focusing on the bidirectional construction and cyclical reinforcement of micro-and macro-level power. The algorithm establishes micro-level power by analyzing riders' data at a granular level, providing raw materials for governance. In line with macro-level objectives, it aggregates micro data into “micro-population clusters” suited to different delivery contexts. Macro-level allocation depends on algorithmic instructions that precisely guide individual behavior. In the context of a large-scale and heterogeneous rider pool, the algorithm employs spatial planning to achieve targeted deployment. Through data fetishism and organizational domestication, the platform induces riders to internalize macro-level rules as conscious actions, enabling macro power to permeate micro subjects. Ultimately, within the algorithmic “black box” —a knowledge-power apparatus integrating knowledge production, truth establishment, and power execution—micro and macro powers operate in a closed loop, rendering algorithmic governmentality an unquestionable technological inevitability.