should be perfectly fine to hold extreme opinions. Bayesian distributions suggests that extreme opinions will be unpopular, and will not find much evidence from the real world, and will be less likely to gain mass following.