This research will study human and machine co-curation of online news feeds and develop new software to compare the user's intent with respect to idealized feed curation, when algorithms filter what appears in online news feeds. This will be combined with research that investigates the user's perception of the algorithm's actions and the user's perceptions of feed settings in control panels. Not only are these algorithmic processes opaque to most users, but many users don't even know that algorithms are making decisions on their behalf. People who might be aware of algorithmic processes at work have no way to verify their existence; changes in a user's search results, for example, might have resulted from a change in the algorithm, or from a change in the user's activity. This work has the potential to increase public algorithm awareness that may lead to more in-depth exploration and possibly to learning about algorithms, with implications for the public understanding of science and technology.
The research stages in this project will generate evidence-based knowledge in five main areas: (a) the level of social media algorithm awareness across a general population, (b) how awareness levels and interaction behavior change when exposed to facets of social feeds via an interactive social feed visualization tool, (c) whether the use of feed settings result in better feed perception, (d) what people want to see in their feeds, and (e) how we can communicate algorithmic process through design of feed content and feed interfaces. The outcomes of this work will help researchers and practitioners in various fields (Human Computer Interaction, Social Science, Design, Engineering, Law, Ethics) critically rethink current computation and design practice and lead to interfaces that help people understand the algorithms that shape their lives.
Christian Sandvig, Center for Political Studies (PI)
American National Science Foundation (NSF)
July 1, 2016 - June 30, 2019