A Question-Answer Architecture of Randomness

This paper develops a conceptual architecture of randomness centered on a question–answer relation. Its main problem is the retrospection problem: if every randomness-conferring articulation must be fixed in advance, genuinely event-induced articulations are excluded; if arbitrary retrospective articulations are allowed, randomness risks becoming trivializable. The proposed solution separates retrospective articulation from the prospective, process-relative openness that can ground a randomness attribution. A later question may identify the realized value of an event-aspect, but it does not retroactively create its randomness. To prevent post hoc constructions, retrospectively articulated variables must satisfy an admissibility condition involving process anchoring, counterfactual stability, and non-gerrymandering. The paper further distinguishes openness from randomness, non-uniformity from bias, event from event-aspect, and wider from symmetry-sensitive narrower randomness. It also develops a two-order model of articulation: pre-framed cases, in which the relevant distinction precedes the realization, and event-induced cases, in which an event first occurs and only subsequently makes a question salient. The framework is situated in dialogue with Aristotle, Cournot, Peirce, Husserl, Waldenfels, Romano, erotetic logic, contemporary work on chance and randomness, and information-theoretic approaches. Its aim is not to introduce a new probability calculus or to prove ontic indeterminism, but to provide an integrative philosophical account of the structures required for non-trivial randomness attribution.