Patrick Wagstrom’s Research Projects

My current research interests address the ways that individuals collaborate, communicate, and learn while building complex software projects. This takes me to the intersection of software engineering, organizational behavior, and social psychology. These topics are a snapshot of my current research interests. For a more complete list of my past research output please see either my publications page or the publications section of my cv.

The following entries appear in no particular order. For example, although Technical Debt is at the top, it is not my primary area of research.

Technical Debt

At some point in time every developer has written a piece of code that they know will have to be replaced in the future. By knowingly writing code that will have to be replaced the developer causes a project to take on technical debt. Our research examines the different ways that projects incur technical debt and how it might be used as a positive tool rather than strictly a potential cost.

Software Ecosystems

Software projects are rarely developed in isolation. In most cases they interface with a number of other projects both through technical interfaces and also through the social interfaces provided by their hosting platform (e.g. GitHub, Google Code, SourceForge) or through their management framework (e.g. Apache Projects). I’ve conducted a lot of research that examines how both of these construct affect the development of a project. During my Ph.D. thesis work I crafted a large dataset from the GNOME project, which I’ve since made publicly available. More recently I’ve started to use graph and network analysis to collect a [large dataset from GitHub that is focused on the Ruby on Rails project][rails].

Software Organizations

On occassion I’ve created papers that study software engineering teams that are almost management style papers. In particular, I did a survey based work with Jonathan Krein, Chuck Knutson, Stan Sutton, and Clay Williams that examined how orgnaizations share information (or fail to) and attempted to understand the impacts of these issues. Later I did a single paper for the Workshop on Social Software Engineering at FSE 2011 that examined how information needs and brokerage affects the collaboration needs of an organization.

Collaborative Development

On occassion I’m involved with a project that involved direct study or implementation of a specific software development tool - most often these are somehow related to collaborative development. In particular, I created a tool with Anita Sarma and Larry Maccherone called Tesseract that visualizes communication requirements as defined by the Socio-Technical Congruence metric that Marcelo Cataldo, Jim Hebrlseb, Kathleen Carley, and I developed.

Within IBM I led a team that studied how all classes of users, including software engineers, designers, product managers, testers, and more use [Rational Team Concert][rational-team-concert] to create high quality software. This study examined the affordances of the tooling and included a study that evaluated the efficiency of interacting with the various components of Rational Team Concert. Subsequently this led to work with JazzHub team on instrumented and their tools to provide responsive analytics. As this work was internal IBM research, sadly there are no publications on these projects.

More recently I’ve become interested in the emerging possibilities unlocked through the use of cloud development environments. This was, which is ongoing with Yi Wang and Evelyn Duesterwald uses a novel method of recording user behavior that is possible in cloud based environments to better understand and predict the transitions in behavioral state as users develop software.

Personal Branding

Building a strong online brand requires more than just getting as many followers as possible. We argue that it is more important to understand the overlap between your interests, the interests of those you follow, and the interests of those that follow you. This research utilized Twitter data and publicly available APIs from organizations such as Klout and PeerIndex to calculate the overlap in interests and passions between you, your followers, and those that follow you. On a personal level, this proved to be insightful to me as it showed that I had a split personality on Twitter that wasn’t effective in creating a personal brand. Half of my personality was focused around running and the other half was focused on software engineering. This created a profile that was confusing at best and led to my forking of profiles into my software engineering profile (@pridkett) and my running profile (@RunPatrick).