Information foraging in debugging

Graduation date: 2010Programmers spend a substantial fraction of their debugging time by navigating \ud through source code, yet little is known about how programmers navigate. With the \ud continuing growth in size and complexity of software, this fraction of time is likely to \ud increase, which presents challenges to those seeking both to understand and address \ud the needs of programmers during debugging. \ud Therefore, we investigated the applicability a theory from another domain, namely \ud information foraging theory, to the problem of programmers' navigation during \ud software maintenance. The goal was to determine the theory's ability to provide a \ud foundational understanding that could inform future tool builders aiming to support \ud programmer navigation. \ud To perform this investigation, we first defined constructs and propositions for a new \ud variant of information foraging theory for software maintenance. We then \ud operationalized the constructs in different ways and built three executable models to \ud allow for empirical investigation. We developed a simple information-scent-only \ud model of navigation, a more advanced model of programmer navigation, named \ud Programmer Flow by Information Scent (PFIS), which accounts for the topological \ud structure of source code, and PFIS 2, a refinement of PFIS that maintains an up-to- \ud date model of source code on the fly and models information scent even in the absence \ud of explicit information about stated goals. \ud We then used the models in three empirical studies to evaluate the applicability of \ud information foraging theory to this domain. First, we conducted a lab study of 12 IBM \ud programmers working on a bug report and feature request. Second, we conducted an \ud analysis of issues and revisions collected from Sourceforge.net. Finally, we collected \ud programmer navigation behavior, revisions and issues from a field study of \ud programmers working in various groups at IBM. \ud All three models predicted programmers' navigation behavior, including where \ud programmers allocated their time among patches, where programmers went, or where \ud programmers made changes to fix defects. These results indicate that information \ud foraging theory can predict and explain programmer navigation behavior, and imply \ud that tools based on the principles of information foraging theory will be able to predict \ud subsequent navigation behavior and potentially assist where programmers should go to \ud make changes to fix bugs