A statistical examination of the evolution and properties of libre software

Abstract

How and why does software evolve? This question has been under study since almost 40 years ago, and it is still a subject of controversy. In the seventies, Meir M. Lehman formulated the laws of software evolution, a first attempt to characterize the dynamics of the evolution of software. With the raise of the libre (free / open source) software development phenomenon, some cases that do not fulfill those laws have appeared. Are Lehman's laws valid in the case of libre software development? Is it possible to design an universal theory for software evolution? And if it is, how? This thesis is a large-scale empirical and statistical approach to analyze the properties and evolution of libre software, using publicly available data sources, hence enabling repeatability of the results and third parties verification, fundamental aspects of any empirical study. The main results are that a small subset of basic size metrics are enough to characterize a software system, software systems are self-similar, and software evolution is a short range correlated (short memory) process.