Can science correct its mistakes? Defenders of the self-corrective thesis answer  affirmatively, arguing that scientific method will refute false theories and find closer approximations to the truth in the long run. In his talk, Felipe Romero will discuss a plausible interpretation of this thesis in terms of frequentist statistics. Using computer simulations, he will argue that such an interpretation is true only under highly idealized social conditions that are hard to satisfy in scientific practice. In particular, Romero will show how some features of the social organization of contemporary science make the long run performance of frequentist statistics prone to a host of biases and distortions. Finally, Romero will argue that attention to individual inference methods alone is insufficient to guard against error in science, and scientific self-correction should be understood as an interaction effect between inference methods and social structures.