We analyze what's wrong with the ('the', meaning the most obvious prima facie) argument for A+B using epistemic logic and game theory.
As is often the case there is a hidden implicit premise in the argument that is false.
Alternatively we might say that the person who argues for A+B on the basis that it would be the "dominant strategy" isn't modelling the situation as the right game.
I'm not getting into the whole "free will", "what is rational?" yadda yadda that is usually associated with this problem. Let's just take the premises of the problem for granted. It is then irrelevant how the predictor can actually be correct with high probability. It is simply assumed to be the case. How the predictor can be correct with high probability is a different problem. (Though he could simply follow the reasoning in this video and conclude that any rational agent who wants to win more money will take choice B. Then he merely needs to evaluate the psychology of the person making the bet and decide whether she is a rational agent
who wants to win more money or not.)
See also
for a discussion of different aspects of Newcomb's Problem.
Music by AkU (Love Shine)
https://akumuzik.bandcamp.com/