Exploring Coordination Models for Ad Hoc Programming Teams

Software development is a complex task with inherently interdependent sub-components. Prior work on crowdsourcing software engineering has addressed this problem by performing an a priori decomposition of the task into well-defined microtasks that individual crowd workers can complete independently. Alternatively, ad hoc teams of experts recruited from online crowds can remotely collaborate, avoiding the up-front cost to end users of task decomposition. However, these temporary ad hoc teams can lead to high coordination costs during the session itself. In this paper, we explore the types and causes of these coordination costs for transient software teams in existing collaborative programming tools: a version control system and a real-time shared editor. Based on our findings, we suggest design elements of shared programming environments that help teams effectively self-coordinate on their task.