Prediction Laundering: The Sociotechnical Production of Truth on Polymarket

Abstract

The growing reliance on decentralized prediction markets as epistemic infrastructures has positioned platforms like Polymarket as providers of objective, real-time probabilistic truth. However, the authoritative signals produced by these platforms often obscure the underlying uncertainty, strategic manipulation, and capital asymmetries inherent in their production, leading to an inappropriate epistemic trust by external observers. This paper presents a qualitative sociotechnical audit of Polymarket (N = 27), combining digital ethnography with interpretive walkthroughs and semi-structured interviews to examine how probabilistic authority is enacted and contested. We introduce the concept of “Prediction Laundering” to describe the sociotechnical process by which prediction markets sanitize subjective bets, strategic hedges, and capital-heavy moves into a single, authoritative probabilistic signal, obscuring the capital asymmetries and contested governance behind its production. We trace this through a four-stage laundering lifecycle: (1) Structural Sanitization, where a centralized ontology scripts the bet-able future; (2) Probabilistic Flattening, where the interface collapses heterogeneous motives into a singular percentage; (3) Architectural Masking, which utilizes selective opacity to hide capital-driven influence; and (4) Epistemic Hardening, where messy governance disputes are erased to produce an objective historical record. Our findings reveal that this process induces a state of Epistemic Stratification, where technical elites audit underlying data while the general public consumes a sanitized, capital-driven signal. We challenge the narrative of frictionless collective intelligence and conclude by arguing for a shift toward Friction-Positive Design, advocating for interventions that expose rather than erase the social and financial friction inherent in the production of synthetic truth.