Paper 4Competition as Cost Discovery: Resource Competition as a Design Principle for Adaptive Intelligent Systems

How should an intelligent system decide where to invest its limited resources? This paper argues that the answer lies not in centralized scheduling but in decentralized competition among the system's own functional components. We develop a theoretical framework in which resource competition serves as a "cost discovery mechanism"—a process through which a system progressively uncovers the true costs (direct, opportunity, and interference) of pursuing different capabilities. Drawing on Hayek's analysis of distributed knowledge and Polanyi's concept of tacit knowledge, we show that cost information in adaptive systems is inherently local and dynamic, and therefore cannot be captured by any centralized controller. We introduce the concept of "cost truthfulness" as a criterion for evaluating competition rules: a rule is cost-truthful if it compels participants to reveal their true resource efficiency through actual performance rather than claimed need. 一个智能系统应如何决定将其有限的资源投入何处?本文认为,答案不在于集权式的调度,而在于系统功能组件之间的去中心化竞争。我们提出了“竞争即成本发现”的理论框架:竞争是一个过程,系统通过该过程逐步揭示追求不同能力时的真实成本(包括直接成本、机会成本和干扰成本)。 借鉴哈耶克对分布式知识的分析和波兰尼的默会知识概念,我们证明了适应性系统中的成本信息本质上是局部且动态的,无法被任何中心控制器完整捕获。我们提出了“成本真实性(cost truthfulness)”概念作为评价竞争规则的标准:如果一个规则能迫使参与者通过实际表现而非申报需求来揭示其真实的资源效率,则该规则是“成本真实”的。