Shared mental models and coordination in large-scale, distributed software development

Coordination is important in large-scale software development because of the large number of people involved and because of the complex dependencies that exist among software tasks, in that even small improvements in productivity can lead to substantial cost-savings and competitive advantage. Despite great technological advances in software engineering and collaboration tools in recent years, coordination in large-scale software development continues to be problematic. Traditional theories suggest that collaborators coordinate by programming their tasks and by communicating with each other, but recent research also suggests that they coordinate through team cognition mechanisms like shared mental models. However, there is very little empirical evidence on how these models affect team coordination, particularly with asynchronous (i.e., non real-time) and geographically distributed tasks like large-scale software development. This dissertation research investigates how shared mental models and geographic dispersion affect coordination in large-scale, distributed software development. It is based on three field studies of software developers at a large telecommunications company—one study based on face-to-face interviews; one survey study; and an archival study using software production data. Results show that: shared mental models have a positive effect on team coordination; prior work familiarity with same software projects and parts reduces software development time; geographic dispersion increases software development time; and the effect of common prior work familiarity on software development time is stronger for geographically distributed teams than for co-located teams.