Paper Number
ICIS2026-2871
Paper Type
Completed
Abstract
Machine learning is often understood as encoding expert judgment at scale, but this presumes an answer to a prior question: whose judgment grounds the model, and against what? Through thirty months of ethnographic fieldwork at a European betting firm, I trace how an ML sportsbook came to treat the wagers of "Wise Guys," external bettors who regularly beat it, not as a threat to curtail but as expertise to harvest, calibrating its output to the judgment of the actors it competes with. I develop the concept of adversarial expertise to name this configuration: domain expertise that is external to the firm, extracted from actors who would withhold it if they could, and rivalrous, existing only in their winning against the firm. The paper contributes a new configuration of ML-mediated expertise to IS research on algorithms, expertise, and ground truth, and specifies the conditions under which it holds beyond this case.
Recommended Citation
friedl, christoffer, "Wise Guys Into Informants: An Ethnography of Adversarial Expertise" (2026). ICIS 2026 Proceedings. 11.
https://aisel.aisnet.org/icis2026/implement/implement/11
Wise Guys Into Informants: An Ethnography of Adversarial Expertise
Machine learning is often understood as encoding expert judgment at scale, but this presumes an answer to a prior question: whose judgment grounds the model, and against what? Through thirty months of ethnographic fieldwork at a European betting firm, I trace how an ML sportsbook came to treat the wagers of "Wise Guys," external bettors who regularly beat it, not as a threat to curtail but as expertise to harvest, calibrating its output to the judgment of the actors it competes with. I develop the concept of adversarial expertise to name this configuration: domain expertise that is external to the firm, extracted from actors who would withhold it if they could, and rivalrous, existing only in their winning against the firm. The paper contributes a new configuration of ML-mediated expertise to IS research on algorithms, expertise, and ground truth, and specifies the conditions under which it holds beyond this case.
Comments
12_Implementation_and_organizing