Software evolution, volatility and lifecycle maintenance patterns: a longitudinal analysis synopsis

Abstract

Despite the rapidity of technological change it is still true that many software systems remain productive for decades. To stay current these systems must evolve as they age. How can these lifecycle software changes, i.e. software volatility, be conceptualized and measured? What are the antecedents of software volatility? How do software volatility and lifecycle maintenance patterns affect lifecycle maintenance outcomes? This research defines and evaluates a system-level multi-dimensional measure of software volatility. Longitudinal analyses use a panel dataset built from a 20-year log of software modifications to 23 application systems. Contributions from this work include a multi-dimensional measure of software volatility, identification of antecedents of volatility and evidence that software volatility and lifecycle maintenance patterns can predict future maintenance outcomes.