The Habitual Power of Prediction Machines

Abstract This article explores the concept of habit as a key to understanding the power of prediction machines in contemporary digital environments. Drawing in particular from pragmatist framework, I argue that habit is not merely a mechanism of repetition, but a temporally structured and future-oriented process of prediction, anticipation and expectation. Through this lens, I reconsider the logic underpinning predictive technologies, particularly recommender systems. By historically tracing the evolution of predictive endeavors, from statistical forecasting to today’s algorithmic anticipation of behavior, I show how the temporality of habit helps illuminate the conceptual assumptions embedded in these systems. I contend that contemporary algorithms operate through a reductive model of habit, privileging past repetition and computational correlation over the relational features of the human habitual environment. The paper concludes by identifying how predictive machines link to our habitual tendencies and dispositions, and where their power lies. In contrast to dichotomous views of human–machine relations, I propose a hypothesis that emphasizes shared anticipatory structures while critically examining the narrowing of behavioral possible pathways within algorithmically curated environments. This perspective offers a path toward rethinking our technological environment by reclaiming a richer, more flexible concept of habit in both platform design and public discourse.