Most would agree that software productivity is low; lower than we'd like certainly, and probably lower than it needs to be. It is harder to agree what productivity is, at least in a quantifiable sense that would allow controlled experiments to be defined and measured scientifically. What is software productivity? Or much the same thing, How do you measure it? In measuring something, one hopes to understand the factors that influence it, and from these to discover ways to control it to advantage. This is the goal of Software Metrics, the study of factors influencing productivity by measuring real or artificial software development projects.