敏捷软件开发
计算机科学
调试
灵活性(工程)
软件
调度(生产过程)
过程(计算)
控制(管理)
固定成本
软件工程
分布式计算
软件开发过程
水准点(测量)
软件开发
特征(语言学)
变量(数学)
项目管理
实时计算
软件发布生命周期
控制系统
功能(生物学)
降低成本
作者
Joohyun Kim,Monica S. Johar,Vijay S. Mookerjee
标识
DOI:10.1177/10591478261495695
摘要
Online games increasingly operate as continuously evolving live services enabled by agile and continuous integration/continuous deployment-based development. In such environments, release timing is no longer a downstream scheduling choice but an operational control variable that interacts directly with functionality growth, error accumulation, and system complexity. This article develops a multi-release optimal control framework that endogenizes release timing and user-facing value. Extending the functionality and error dynamics structure from prior work, we embed repeated release decisions directly into the development process and characterize how construction effort, debugging effort, and cadence jointly determine long-run value creation. Within this framework, we analyze two structured release regimes widely observed in practice: a time-based release policy, with fixed sprint duration, and a functionality-based release policy, with fixed feature scope. We show that incremental release value declines endogenously as integration complexity increases. Under time-based policies, the number of new features delivered per release decreases over time. Under functionality-based policies, maintaining fixed feature bundles requires progressively longer development intervals. In both regimes, effort shifts systematically from construction toward debugging as the system matures. We further introduce a hybrid release policy as a supplementary benchmark, allowing the firm to switch between time-based and functionality-based policies subject to a switching cost. This benchmark identifies conditions under which policy flexibility improves release performance while preserving the operational structure of the two release regimes. The results provide an operations-based theory of release cadence under endogenous complexity growth in software systems with repeated deployment.
科研通智能强力驱动
Strongly Powered by AbleSci AI