Open source software (OSS) development platform is one of the most popular software development environments that organizations use to source high quality information system products. For example, the most popular open source product, the Linux kernel powers world’s largest stock exchanges; and Apache, the most popular web server was developed in open source platform, powers 52% of all websites globally. Projects are increasingly being conducted in OSS platform due to increased interest from the organizations. For example, major IT companies including Google, Microsoft and Facebook have increased their reliance on software development initiatives on OSS platforms . These developments have important implications for service operations management, particularly in terms of how projects should be managed within these contexts. OSS provides an Internet based virtual community-like software development venue where software developers collaborate voluntarily toward the goal of developing unique information system products. Such voluntary participation is important for sustainability of the OSS projects as inadequate participation has resulted in 80% of the OSS projects being canceled in the development phase. OSS developers stay motivated to participate in OSS because they enjoy the coding process and value the sense of unity with fellow developers while interacting in the software development project.
Traditional software development in firms differs from OSS development in three important ways: first, for OSS development volunteers the source of motivation is different than pay or career incentives; second, the lack of formal governance structure in a OSS development team excludes managers from exerting project control mechanisms; and third, in contrast to traditional software development, the ability of volunteers to coordinate work among OSS developers is contingent on effective information exchange over the Internet. To that extent, OSS platforms have been created that facilitate information exchanges among project members by means of mailing list or thread communications relating to a software code. However, the effect of increased interaction among members in unregulated virtual environment on project performance in terms of its quality and service delivery is unclear. A lack of synchronous interactions in virtual software development environment may weaken flow of information which may, in turn, impact project performance adversely. Higher interactions present teams with increased overhead which lowers team productivity. However, interactions among team members help members to self-regulate to distribute project activities and enhance productivity.
It seems that some form of interaction in self-regulated OSS teams is necessary for efficient information flow within team and to keep members motivated. On the other hand, beyond a certain threshold it may bring in overload to weigh down team performance. Communication frequency in a cross-functional team in traditional projects may have a curvilinear relationship with project goal achievement. Given the increasingly virtual nature of project management and its implications for operations management, it is important to understand how team member interactions related to performance in projects that are not governed within the traditional project context. To the best of our knowledge, no previous empirical work has considered how team member interactions influence project performance in alternative project context such as OSS development that are characterized by their virtual setting and voluntary participation by team members. The first research objective of this study is to address this gap.
Proactive planning for problem resolution in the traditional software development context is challenging. Given the additional complexities of self-regulated OSS teams including, but not limited to, absence of governance structure in a team, independence of members to contribute to projects, geographical and cultural dispersion of volunteers and keeping volunteers motivated in project activities without traditional incentive mechanisms, proactive project management is even more challenging for a project manager responsible for an OSS development initiative. Traditional project management process, which allows physical interaction among members and offers associated visual cues for managing projects, will fall short in the case of open-source projects. Virtual project venues and online knowledge communities, such as OSS platforms, further accentuate these challenges since team members do not interact in person with each other.
To enforce project control, managers of OSS projects often engage actively in discussion threads to remove information asymmetry involving team activities. Such participation helps project team members to access critical knowledge resources that generally much experienced managers bring along. Project managers exploit their connection to other projects to bring critical knowledge resources that enhance project performance. In the OSS development context, as managers are unable to pick up visual and verbal cues, active participation in project discussions becomes all the more important. However, we do not know how self-guided OSS development teams respond to such project control mechanism. Additionally, participation increases information processing overload for team members. Understanding the impact of project managers’ active participation in OSS projects on project performance is relatively understudied and forms the forms the basis for the second research objective of our study.
Project manager’s active participation may not only directly affect project performance but may also affect performance by influencing team member interactions. Effective and efficient ways of enabling information exchanges within team have been a cornerstone of project management research. With decreasing performance the motivation of project members, the glue that binds them to OSS platform, starts to wane and OSS projects may see higher dropouts. Inadequate participation has resulted in 80% of the OSS projects being canceled in the development phase. Project manager’s active participation may help with establishing formal structural coordination mechanisms in OSS development projects, thereby enabling participants and projects to reach their goals. Enforcement of process structure by means of effective project management in virtual software development context can alleviate conflict and enhance performance. The importance of project managers’ active participation in managing team interaction in a self-organizing OSS team structure is notably understudied and warrants research attention. Our third research objective aims at contributing to this need for further research.
We draw on social network theory and task familiarity literature to guide our research. Social network theory posits that ties among entities in a network generate social capital that translates to performance. We use this theoretical lens to understand the impact of team member interactions in a project network on project performance. To examine this relationship, we consider project team member interaction networks for 1842 open-source software development projects on the GitHub platform. For each of these projects we collect panel data that include evolving network measures and other project characteristics for two years (104 weeks) from the time of project initiation. Specifically, within each week we consider the project team member interaction network formed by actual comments and replies by software developers in response to an open issues and pull requests for these projects. Open issues refer to bugs reported in the software bugs management system of OSS platforms such as GitHub. Project members work toward resolution of these bugs by making necessary code changes. The changed codes are then merged in the source code by issuing pull requests.
We draw on task familiarity literature to explain how project manager’s active participation impacts project performance. Task familiarity literature suggests that familiarity with team member activities enables the manager to be aware of the flow of information in the project, which enables effective project management. We build on this reasoning to study the impact of manager participation in the OSS context. To study the relationship, we collect data on project manager’s participation in the network formed by team member interactions in response to the open issues and pull requests. In the GitHub platform closing an open issue is an indicator of completion of a specific project task. We consider the number of closed issues in a given week as our dependent variable and examine how this performance measure is impacted by the direct and interaction effects of network characteristics of team interaction and project manager’s active participation.
The results obtained from analyzing 133,023 observations of unbalanced dynamic panel show that as the density of the network formed by communication instances among project team members increase, the number of weekly closed issues in an OSS project initially increase but then decrease. We find that project manager’s degree centrality in the team interaction network has a direct effect of increasing the number of weekly closed issues as well as a moderating effect of flattening the curvilinear relationship between team interaction and weekly issue closure rate. In essence, project manager’s active participation alleviates the reduction in weekly issue closure rate once the density of team interaction goes beyond the inflection point.
Source: Pal, S., Nair, A., Zhou, Z. XXXX. Managing Projects in Virtual Settings: Information Exchange Networks and Project Performance. Journal of Operations Management.