模块化(生物学)
计算机科学
人机交互
人工智能
设计要素和原则
数据科学
复杂系统
理论计算机科学
认知科学
认识论
标识
DOI:10.1287/isre.2022.0444
摘要
Platforms from iOS to Salesforce to OpenAI rely on modular architectures to coordinate vast numbers of independent developers without micromanaging them, yet modularity’s relationship with coordination—direct human engagement between platform owners and developers—remains poorly understood, leaving platforms navigating by costly trial and error. This paper illustrates that whether modularity substitutes for or complements coordination depends on which aspect of modularity is in play: interface conformance—adherence to stable technical specifications—substitutes for coordination by absorbing integration into prespecified agreements, whereas decoupling and “design rules” complement it, as the independence they confer surfaces unforeseen dependencies and interpretation gaps that specifications cannot close. For platform owners, this means targeting coordination effort toward design rules and decoupling. Design rules, the guidelines governing how apps behave internally, need targeted human engagement to prevent misinterpretation and realize their evolutionary potential without homogenizing apps or managing with a heavy hand—a lens that applies to AI platforms, hardware-software hybrids, and platforms navigating national data sovereignty mandates.
科研通智能强力驱动
Strongly Powered by AbleSci AI