Economies and Diseconomies of Scale in Software Development

Abstract

Economies and diseconomies of scale—gains and losses in productivity arising from increases in project size—are of considerable, practical importance in software engineering. However, there is no consensus as to whether, in general, economies or diseconomies exist with respect to project size; neither is there any consensus on the relationship among project size, team size, and overall productivity. Using evidence from a large database of projects, this paper derives a simple parametric model for estimating development effort in terms of size and productivity factors. The model is compared with popular estimation models—the Putnam model and COCOMO II—and a possible explanation is provided for the lack of consensus. By considering also the value of the software produced, the model provides a basis for the calculation of the optimal team size for a given task, with respect to the anticipated net benefit from the development activity.