What is a system approach to prediction?

Abstract

It is proposed that the systems approach as formulated in the 1960s should be revived to resolve some of the difficulties in predicting reliability of systems with software. The basics of the approach are reviewed, and peculiarities of software are discussed. It is shown that the main difficulty in predicting reliability of systems incorporating software lies in improper understanding of the nature of software, in the lack of communication between those dealing with hardware and software, and in too much emphasis on 'how' while forgetting 'what' and 'why'.<>