Advantages of open source processes for reliability: clarifying the issues

Some authors maintain that open source software processes are particularly well-suited for delivering good reliability. We discuss this kind of statement, first clarifying the different measures of reliability and of a process’s ability to deliver it that can be of interest, and then proposing a way of addressing part of it via probabilistic modelling. We present a model of the reliability improvement process that results from the use of the software and the fixing of reported faults, which takes account of the effect on this process of the variety of software use patterns within the user community. We show preliminary, interesting, non intuitive results concerning the conjecture that a more diverse population of users engaged in reporting faults may give OSS processes an advantage over conventional industrial processes, in terms of fast reliability growth after release, and discuss further possible developments.