gwern comments on [Link] Small-game fallacies: a Problem for Prediction Markets