One of the examples given in the article is the prisoner's dilemma.

They present it as a celebration of "rational" yet self-defeating selfishness.

There's a lot of crazy things said in and by economists, but I dont think I've ever heard anyone actually claim that.

What it is useful for, what it "predicts", is that even perfectly, better than humanly-possible, smart people can still find themselves in situations where they can't get the best outcome for themselves by doing whatever is best for them alone.

The general conclusion and recommendation that arises is to implement some kind of meta-regulatory framework to ensure that the more desired outcome arises.

And indeed, we see exactly that in situations like criminal gangs, that co-operate to punish those that "snitch" and so solve the dilemma. The evolutionary coop-cop things mentioned are other examples.

Because the ideal and reality of the "prisoner" part so often diverge, I generally prefer the example of trying to do a drug deal or spy swap. You have something they want, and they have something you want. But unless theres some guarantee you wont be ripped off, then the profitable to both trade won't happen.