Scientists should understand themselves as both learners and active creators of knowledge, according to a recent research paper on “Distributed Science: The Scientific Process as Multi-Scale Active Inference.” The study disputes the traditional inductive scientific method and suggests a Bayesian model, whereby multiple theories are weighed side by side for plausibility. The model also factors in the role of personal and community-wide interests affecting scientific advancement. Caution is called for when interpreting science-based policies, since scientific understanding is always evolving.