Software Metrics Reviewed
Keith J. Laidler · 1993
Abstract This chapter reviews developments in the field of software metrics from the early 1970s to the present time. Metrics can be categorized according to the stage of software development with which they are associated. Thus we have code metrics, design metrics, and specification metrics. Although our primary interest in this book is with software designs, code measures will also be examined. The chapter indicates briefly that the area of software metrics is afflicted by a number of serious problems, a theme expanded upon in the following chapter. This chapter is a brief review of work carried out in the metrics area over the last three decades. In a number of ways both this chapter and Chapter 3, which describes three metrics in detail, present a rather bleak view of what is a potentially promising and important field of software engineering. There are several recurring themes in both the chapters. First, the majority of metrics are presented in such a fashion that it is unclear exactly what is being measured. Terms such as ‘complexity’ and ‘quality’ proliferate. As a direct result, empirical evaluations have interpreted metrics in widely differing ways. Divergent results have ensued for every metric evaluated above. Almost the only unifying feature has been the strong statistical associations with size metrics such as lines of code.*