Root Causes of Poor Software Quality
Capers Jones · Auerbach Publications eBooks · 2021
In the 1970s, IBM developed two useful metrics for software quality: defect potentials and defect removal efficiency (DRE). The defect potential metric is the sum total of bugs likely to be found in a software application. (These include bugs in requirements, architecture, design, code, documents, and “bad fixes” or new bugs introduced when fixing old bugs.) The current U.S. average for defect potentials is about 4.25 bugs per function point. The DRE metric is the percentage of bugs found and removed before the software is delivered. The current U.S. average for DRE is about 92%. Poor-quality projects have defect potentials of over 5 bugs per function point; leading projects have defect potentials below 2 bugs per function point. Poor quality projects have DRE of below 85%. Leading projects have DRE topping 99%. The main reasons for the differences are failure to use effective methods such as structured development, design reviews, and formal inspections.