Function point counting: one program’s experience

Abstract

A recent software development program implemented as a series of overlapping increments in a new development environment provided both the need and opportunity to evaluate several estimating methods for use in project planning. Several estimating methods including variants of function point counting were evaluated during this project. The project structure allowed evaluation of these methods in practice. The function point counting method chosen as the basic estimating method over most of the project was consistently applied in an essentially stable development environment for about two years. In spite of the stability in environment and development personnel, interesting variations in function point counts across time were observed. This paper documents the variations observed and speculates on how such variations may be measuring intrinsic differences within the development process rather than differences in counting ability. A framework based on this variation is proposed as a possible way to objectively measure development variations in design and implementation efficiency.