A new vernacular of algorithms

One of the challenges that I have faced through the course of my research into migration algorithms is finding a consistent definition of algorithms. Asking workshop participants or audience members at my talks what an algorithm is, they usually respond with the idea of a personalised program that learns their preferences and adapts online content and experiences to their unique tastes. This means that for a majority of people, algorithms are proprietary machine learning (ML) software connected to major technology companies. A YouTuber, for instance, might say ‘please like my video and subscribe to my channel for the YouTube algorithm to push my content’. The use of the word ‘algorithm’ in this context, is a personalised recommendation program that can be hacked through calculated engagement. This understanding of an algorithm bears paradoxes of user agency. The contradiction here is that although individuals are supposedly responsible for the recommendations they see, the commercial proprietors of the applications deploy a set of rules that prioritises specific content. In this blog, I want to think aloud about a critical definition of algorithms that incorporates both their computational, and social understandings into a new vernacular.