Is it a bug or a feature? The concept of software obsolescence

Unlike physical objects, software sees no natural wear and tear such as oxidation or physical and thermal stress. Once a software is correctly and functionally implemented into a system, it could run forever, at least in theory. In practice, recent times have seen a growing number of empirical examples, where a malfunctioning software, high performance requirements or cancelled support services caused the end of lifetime of an otherwise functioning product. So far, software obsolescence (SO) was mostly studied as software rot or software blow in sectors like aviation or military, where reliability is most crucial. But with an increasing market penetration of embedded hard- and software systems such as SmartTVs, Bluetooth Speakers, Smart Home Appliances, and other Cyber-Physical Systems (CPS), there is growing attention to software induced obsolescence in the consumer electronics market. From an environmental perspective the problem is evident and is already targeted by recent strategies in national and European legislation such as the German Digital Policy Agenda for the Environment (BMU 2020) or the European Circular Economy Action Plan (EC 2020). However, applied research on SO is still young and lacks concepts and heuristics to systematically study the causes, appearances and impacts of the problem. The aim of this paper is to (i) propose a formal definition, illustrate the difference of direct and indirect software obsolescence (ii) to present a four-level scheme (legal-executable-usable-functional - LEUF) to analyze and classify mechanisms of software obsolescence; and (iii) to give a proper overview of the current state of regulations on SO. Based on our considerations we discuss mitigation strategies to prevent software obsolescence and identify further policy and research needs.