We'd need to get Alpharisc to state it for the scenario he foresees for XC. But the problem in general concerns two "rationally self-interested" prisoners who (a) get out of jail if one betrays the other prisoner, (b) get a longer sentence if both prisoners betray each other, or (c) get a shorter sentence if neither of them betray the other.

Edit: the idea is that it's rational to betray the other prisoner. However things get more complicated in the iterative version of the dilemma: prisoners get to repeatedly choose to betray or not betray one another. Alpharisc, to my mind this plays into the hands of a WOT model as long as it's combined with a means of punishment. What's your view?