A growing body of work in algorithmic fairness has brought issues around 'data bias' to the forefront. Often, these discussions have focused on ways in which marginalized communities are under- and mis-represented in datasets, contributing to unintended consequences in algorithmic decision-making. In this talk, we argue that these challenges are only a part of a broader set of data inequalities. We present a framework encompassing data exclusion, inclusion, accuracy, usability, access, and ownership. Using examples in criminal justice, public service provisions, and health, we demonstrate ways in which processes in data generation, access, and use create and amplify harms in algorithmic decision-making. We conclude with a discussion on ways to combat these challenges using interventions drawing from a range of disciplines.

The list of governance challenges brought by digital technology is long and touches every aspect of modern life. Platforms such as Amazon, Uber, Google, Facebook, Linkedin, YouTube, and Twitter effectively control access to a wide variety of information, services, and products. Surveillance systems, algorithmic filtering of information, bias, discrimination, social and political rankings of citizens are a few examples of challenges of the digital world. I will first focus on the structural level to discuss three digital governance problems that are worth highlighting, namely the multistakeholder nature of the Internet, the climate of growing global polarization and the rampant online misinformation. I will also discuss how computing research can stimulate the development of future policies and regulation for the digital world, with novel conceptions of governance using new technologies of distributed systems. This talk will suggest new opportunities to construct the future of digital governance, in a legitimate, inclusive, and secure use of digital resources to produce sustainable services and public policies for the online world.