Forty Years of Lessons Learned in Software Reliability

The amount of software in a system is growing every year due to greater intelligence and the fact that software is now doing tasks that used to be done by hardware. Unless the level of rigor in software development and testing improves at the same rate, the number of failures due to software will also grow every year. More concerning than this is that software failures can be super-catastrophic. Even though software failures are regularly in the news, some engineering community members still deny that software fails and/or assume that software failures are random events that cannot be predicted. This article will illustrate more than 40 years of cost-effective lessons learned for predicting the functional and process-related causes of software failures and mitigating them before they cause a failure.