Appendix A. Good Enough Is Better Than Floundering

There is a lot of variability (or uncertainty or noise) in the core-metric data produced by software development projects. To narrow this range of variability, we stratified the data. By stratify, we mean to sort out the data into classes; in Figure 17-1, for example, we sorted the data into three main classes: information systems, engineering systems, and real-time systems. Sorted in this way, the data points fall in three strata. To gain still greater precision for some purposes, we have gone beyond these three categories and stratified data into nine subcategories: microcode/firmware, real time, avionic, system software, command and control, telecommunications and message switching, scientific, ...

Get Five Core Metrics: The Intelligence Behind Successful Software Management now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.