As algorithms increasingly mediate everyday life, they build the world. Algorithms shape the ways we understand and navigate our lived realities, and enact sociopolitical order as reflections of human values and assumptions. Recognition of this so-called "algorithmic power" has prompted various efforts to govern algorithms. As "algorithmic literacy" begins to enter into the vernacular of those involved in such efforts, the purpose of this dissertation is to interrogate what it means to know algorithms. In this, I advocate for an expanded horizon of "consequential" knowledge about algorithms, and treat algorithmic literacy as a bottom-up tool of governance that can be used to confront algorithmic power. I accomplished this task through an exploration of two case studies in which practices of learning about and making sense of algorithms are particularly salient. The first case focuses on Instagram influencers' pursuit of visibility; the second case focuses on "BreadTube," a leftist online community concerned with the visibility of far right ideology on YouTube.Through investigating these cases, I developed the conceptual framework of critical algorithmic literacy, which recognizes knowledge as situated (Haraway, 1988), constructed within and in relation to the discursive landscape of social worlds (Clarke & Star, 2008), and involving the cultivation of a critical consciousness through recognizing and responding to algorithms as expressions of broader systems of power (Freire, 2000). This framework recognizes that there are multiple ways of knowing algorithms-that technical knowledge is merely one of multiple ways. It also recognizes that what people know about algorithms depends in large part on who they are. Further, the framework of critical algorithmic literacy recognizes that what people know about algorithms is the result of negotiations of power, particularly those that grant epistemic authority to some over others.In focusing on practices of learning about and making sense of platform algorithms, I demonstrated platforms' centrality in these practices. I introduced the concept of platform epistemology in order to capture the ways that platforms set the conditions under which knowledge about algorithms may be constructed and legitimized. Through their design choices, in activating connections between users, and in connecting users to content based on inferred interests, platforms orchestrate the flow of information about algorithms online. Further, by shrouding their algorithms in secrecy and closing off access to user data from which insight could be drawn, platforms give rise to hierarchies of epistemic authority on algorithms. Thus, I show that platforms do not merely passively support learning about algorithms, they actively shape what is and can be known about algorithms.Ultimately, I argue that critical algorithmic literacy has the potential to be a means of increasing the involvement of private citizens in efforts to democratically govern algorithms in the public interest. However, in order to fully realize this potentiality, we need to ensure that users and other stakeholders can effectively advocate for their interests based on their unique insight about what algorithms mean to and for them. This means reimagining algorithmic literacy as contextual, heterogeneous, always partial, and inseparable from questions of power.
Read
