Bo Waggoner - Research Statement

Information permeates our world, and in many forms: data on disk, knowledge in an expert’s mind, statistical observations of the natural world. To make decisions or predictions – treatments in health care, climate and public policy, user behavior and suggestions in e-commerce – we seek to understand and design systems for acquiring, aggregating, and acting on that information. Such systems can involve computers (e.g. machine learning algorithms); humans (e.g. voting systems, prediction markets); or a mix (e.g. crowdsourcing platforms). However, there is a growing recognition of the consequences these systems have for the people they interact with, directly or indirectly. Statistics on health care data can help design lifesaving treatments – but can also have huge privacy impacts on that data’s owners. Platforms for prediction or decisionmaking can encounter strategic misreporting in order to manipulate the outcomes. E-commerce systems built on seemingly neutral principles can quickly become discriminatory if designed without fairness protections in mind. To overcome these challenges, we must understand how social constraints such as privacy, fairness, and strategic behavior impact these systems; and we must understand how these systems impact the people they touch. Then, we can hope to build better systems with societal impact as a primary goal.