The Coordination Mechanisms of Organizational Routines: A Case Study on a Free/Open Source Software Project

How might distributed, peer-production organizations that rely on the Open Innovation paradigm sustain task specialization and achieve effective coordination in the absence of formal hierarchical control? This question is increasingly relevant because a number of productions processes are being shifted from the physical to the virtual domain. Traditionally, scholars identified in organizational routines the building blocks of coordinated action in organizations. In this paper, we draw on the evolutionary perspective of organizational change (Nelson and Winter, 1982) to investigate the emergence and change of organizational routines in the context of distributed, peer-production communities. We argue that change emerges endogenously from routines - which we define as ordered sequences of actions linking problem-solvers and problems within organizations. More specifically, we investigate how routines emerge dynamically from the dual association connecting individuals (or “problem-solvers”) and tasks (or “problems”) in organizations. In particular, we ask: How do organizational routines (structured sequences of actions) emerge, evolve and persist despite the demographic turnover of participants and the ever changing character of organizational problems? To answer this question we run newly developed class of Relational Event Models on the entire sequence of actions undertaken by software contributors on software bugs during the entire lifetime of the Apache HTTP server, a very large Free/Open Source Software (F/OSS) project. By examining directly the micro-level, socio-temporal interdependencies generated by individual actions performed by problem-solvers, we find that self-reinforcing processes underpin the endogenous coordination of a decentralized production community, by fostering task specialization learning mechanisms. Different types of activities can coexist and be coordinated over time in a decentralized decision environment. Our analytical goal is to show how sequences of interdependent problem-solving actions become embedded in temporal sequences of relational events, which then trigger specific self-reinforcing mechanisms that provide the social infrastructure sustaining the production of F/OSS.