A (hopefully) comprehensive solution to the Newcomb's problem

Abstract The Newcomb's problem has been discussed in the philosophical literature for more than fifty years. So far quarrels mainly oppose causalism on the one hand, and evidentialism on the other hand. This paper wants to explain why causalists are right from a certain point of view, and why they should one-box --- instead of keeping two-boxing. The Newcomb's problem can be thought in terms of possible worlds. I assume here that this problem does not happen in our world, but it occurs in a world where causality and rationality obey different rules. But, in any cases, when the predictor is reliable, the player must one-box.