Agile Maturity Assessment: Automating a Manual Process; Actor-Network Theory as a Method for Understanding Power Shifts That Lead to Successful Algorithmic Accountability Measures
Callen, Thomas, School of Engineering and Applied Science, University of Virginia
Neeley, Kathryn, EN-Engineering and Society, University of Virginia
Graham, Daniel, EN-Comp Science Dept, University of Virginia
Algorithms are widely used in decision-making contexts, from music playlists to advertising, credit, and public safety. Both my technical project and my STS research deal with the process of developing useful software projects. For my technical project, I developed an internal web application to administer agile maturity assessments to software development teams at Capital One. My STS research utilized Actor-Network Theory (ANT) to reveal that algorithmic accountability methods, such as transparency and participatory design, transfer power from algorithm designers to end users. This transfer leads to more positive outcomes in the implementation of algorithms due to increased outsider perspectives.
Both companies and developers want to design software that serves their customers needs and is efficient to make. The agile development methodology is a commonly accepted way of designing software that accomplishes these goals. As software development teams are transitioning into fully adopting this methodology, managers want to know their progress in order to better understand how the team is performing and where they can improve. An agile maturity assessment gauges how well a software development team has integrated agile principles into their development process. In my internship at Capital One, me and my team developed a web application for teams to take agile maturity assessments which were previously done using spreadsheets and online forms. This product serves as a central location for managers to analyze their teams' agile performance. It also provides a standardized assessment for teams within the same line of business which was not the case prior to our work. The largest part of the application I contributed to was the webpage’s main admin dashboard. This dashboard enables managers to quickly review data about the assessments of the teams they manage in a clear and visually appealing manner. Our tool can save managers hundreds of hours of time across Capital One as it replaces the current manual task of completing the assessments and compiling the data into a clear format. With this tool, developers can identify how they can work better as a team to get their work done more efficiently and incorporate user feedback into the design process effectively.
In my STS research, I use the ANT framework to understand how transparency and participatory design enable end users and outside stakeholders to hold algorithms accountable. Algorithmic accountability aims to ensure the biases that may exist in data or in the designers of software does not translate into biased outcomes which can further exacerbate discriminatory practices we already have. I found that these two algorithmic accountability methods work by translating power that software system designers have to the end users. This power shift enables the end users to have more input into the process of designing algorithms or determining whether it is acceptable to use in a community.
My technical and STS projects are related in that they both deal with the process of effectively creating software systems. The technical project enables software development teams to understand where they can improve in their development methodologies with the aim of creating software that meets customer needs. My STS topic focuses on the systems that allow for end users to more effectively hold algorithms accountable and to ensure an algorithm’s biases don’t negatively impact the end users. One of the methods described in my STS topic, participatory design, is an integral part of agile development. Agile development requires communication with stakeholders to understand their requirements for the system and to ensure the product they are receiving actually benefits them. My STS topic focuses on the methodologies that can be used to ensure professional developers are responsible with the influence their algorithms can have over society. While the technical topic encourages teams to follow a method of development that encourages interaction with all stakeholders to ensure algorithms serve their correct purpose.
BS (Bachelor of Science)
Agile Development, Actor-Network Theory, Agile Maturity Assessment, Algorithmic Accountability
School of Engineering and Applied Science
Bachelor of Science in Computer Science
Technical Advisor: Daniel Graham
STS Advisor: Kathryn Neeley
Technical Team Members: Thomas Callen
All rights reserved (no additional license for public reuse)