Reactive Programming Collaborative Agents for Self-Managing Software Teams

Abstract

In self-managing software teams, developers with complementary skills collaborate with each other to improve productivity and increase innovation. Due to nature of innovation, managers in these teams will acquire only summarized information from developers to avoid micromanagement. They will not plan fine-grained activities but leave responsibility and autonomy to self-managing developers. Developers discover collaboration opportunities from up-to-date assets and decide needed collaborative activities. To maintain effective and consistent collaboration, distributed knowledge of decision making from up-to-date assets must be captured and accumulated for these teams. This study proposes a method to distribute asset monitoring and decision making through reactive agents. Reactive programming technique is used to collaborate reactive agents. These agents present asset changes as event streams and provide time-varying reactions to each other as a network. The network can automatically react to decisions made by every agent. Through this method, self-managing software teams build their reactive knowledge network which responds to asset changes instantly and reveals collaboration opportunities intelligently.