Newcomb's Problem

The description of the problem in this transcript of the podcast is clearer than the one on this short video introduction.

This is interesting from a computational perspective, because it seems that you can map this problem onto the halting problem as formulated here, but with two predictors: one which "lies", saying "I don't halt" if it does halt, or going into an infinite loop otherwise, and  then a second predictor which says "It halts" if  the first predictor halts, or goes into an infinite loop otherwise. I am a bit tired for this sort of shit right now, but any interested readers (assuming, of course, such creatures exist!) might enjoy considering these two machines as oracles in Smullyan's problem of the road to heaven, where there are three oracles, one which lies, and one which tells the truth and one which either lies or tells the truth at random. In this problem if Smullyan's you can get a correct answer from two oracles' answers to two questions.