Epistemic Risk and the Transcendental Case Against Determinism

A familiar intuition holds that determinism creates an epistemically adverse context. This paper gives that intuition a formal shape by developing a new epistemic transcendental argument (ETA) grounded in the notion of epistemic risk. First, we formalise epistemic risk through a metric space W equipped with two metrics, D and N, corresponding to distinct theories of risk. Drawing on the notions of modal closeness and normalcy, we argue that these metrics better capture our intuitions about risk than traditional similarity-based accounts. Building on these insights, we articulate an argument based on five axioms. The axioms are philosophically motivated using the two metrics, their independence is verified in Mace4, and the derivation of the denial of determinism is formally carried out in Lean 4.