434.11
Heroes and Zeroes: The Anatomy of a Failed Software Project

Thursday, 19 July 2018: 08:45
Location: 401 (MTCC SOUTH BUILDING)
Oral Presentation
Arild BERGH, Norwegian Defence Research Establishment (FFI), Norway
Any software use, whether posting messages on Facebook or analysing cancer data, emerges from the work of programmers. Despite this, little sociological work examines programmers and their worlds, where they act as creators, mediators and translators of ideas into tangible applications. Existing work tends to focus on practical aspects such as education or management of programmers or they provide journalistic "hero” accounts that focus on clichés such as all night coding marathons to create compelling narratives with little in-depth analysis. Emerging research areas, e. g. software studies, seem uninterested in empirical work rooted in programmers’ social interactions. The programmer is either invisible, or at best, an interchangeable element in a broader discussion.

This paper attempts to fill this gap by examining a failed software project from the inside. The author is a former programmer and current sociologist who in this case provided specifications to external programmers and evaluated the resulting web application. Using a combination of auto-ethnography and document analysis I will apply a micro-sociological perspective to discuss an “ideal type” timeline from the “dream” stage to the “final breakdown”. The purpose is to highlight social interactions, power issues and conflicts between different actors’ goals and actions, how events were perceived and framed and the role of individual mediators in the larger “software development chain”. Using a grounded theory approach for a tentative, initial exploration of very rich data will hopefully lead to further research regarding programmers and their roles in science and technology.

Understanding software development failures also have a value in itself. Although failure is fetishized in start-up environments, the reality is that in most cases failure happen in custom software development; thus a real-world need goes unfulfilled. Evaluations tend to focus on management issues; this paper will thus also provide additional perspectives for assessments of such failures.