Recent papers have suggested that breakthroughs in software productivity may not be forthcoming. This essay reviews some of that speculation and offers a paradigm shift in which breakthroughs are possible. Rather than considering the process of program development, which may be approaching its productivity ceiling, this paper suggests that the goal of software engineering ought to be the development of applications. Viewed from this perspective, the software process becomes one of problem solving in the application domain. Two conceptual measures that offer guidelines for identifying productivity improvement are introduced, and some case-study data are presented. The conclusion is that major improvements can be expected if one considers software engineering to be the use of software to produce a software solution to an application problem.