The Allocation of Software Development Resources in Open Source Production Mode

Abstract

This paper aims to develop a stochastic simulation structure capable of describing the decentralized, micro-level decisions that allocate programming resources both within and among open source/free software (OS/FS) projects, and that thereby generate an array of OS/FS system products each of which possesses particular qualitative attributes. The core or behavioral kernel of simulation tool presented here represents the effects of the reputational reward structure of OS/FS communities (as characterized by Raymond 1998) to be the key mechanism governing the probabilistic allocation of agents’ individual contributions among the constituent components of an evolving software system. In this regard, our approach follows the institutional analysis approach associated with studies of academic researchers in “open science” communities. For the purposes of this first step, the focus of the analysis is confined to showing the ways in which the specific norms of the reward system and organizational rules can shape emergent properties of successive releases of code for a given project, such as its range of functions and reliability. The global performance of the OS/FS mode, in matching the functional and other characteristics of the variety of software systems that are produced with the needs of users in various sectors of the economy and polity, obviously, is a matter of considerable importance that will bear upon the long-term viability and growth of this mode of organizing production and distribution. Our larger objective, therefore, is to arrive at a parsimonious characterization of the workings of OS/FS communities engaged across a number of projects, and their collective productive performance in dimensions that are amenable to “social welfare” evaluation. Seeking that goal will pose further new and interesting problems for study, a number of which are identified in the essay’s conclusion. Yet, it is argued that that these too will be found to be tractable within the framework provided by refining and elaborating on the core (“proof of concept”) model that is presented in this paper.