“Optimize user experience”: optimization techniques and the simulation of life, from the model to the algorithm

This article takes up the issue of optimization to consider the relationship between predictive algorithms and platform user experience. Corporate data analytic practices increasingly rely on machine learning algorithms that apply models to user behaviors, producing “knowledge” about users that can be bought and sold. This article considers the opacity of algorithms today in relation to optimization. Using a conceptual apparatus that draws from the study of cultural techniques, the following argues that optimization—the task of finding a sufficient solution to a well-defined problem—makes use of models to simulate possible answers to problems around the incomputablity of behavior. Tracing a set of examples that deal with the problem of predicting behavior—the “minimum point” problem, John von Neumann's automata theory, and the Facebook pixel—optimization is characterized by a shift from statistical model making towards predictive and algorithmic techniques. This shift is seen within the context of the decline of Cold War rationality towards the embeddedness of “intelligent” algorithms across technoculture.

Duke Scholars

Author Rebecca Uliasz