Increasing Software Reliability through Use of Genericity

In our opinion, methods for construction of reliable software are of great importance. Reliability engineering must start in the earliest phases of a software project, and it has to consider not only the architectural level, but should be in the mind of humans even earlier when analyzing the problem space. In a previous paper, we argued that reduction of redundancy of software is a central factor for increasing economy and for increasing the scope of what can be handled by human groups. In this note, we argue that reduction of redundancy is also a central factor for increasing reliability. We review our subclasses of genericity, called universal genericity, compositorical genericity, and extensional genericity, with respect to their impact on reliability. We find that the superiority of universal genericity and compositorical genericity over extensional genericity with respect to their potential for decreasing redundancy will take over to the field of reliability, meaning that universal genericity has the highest potential for building more reliable systems. As a consequence, OO analysis and design methods (which mostly prefer extensional genericity) are not the highest potentials for making complex systems extremely reliable.