I have to say when they went over this in game theory in a course in college I was like really, this was a Nobel on this? Well yeah behind the simple explanations is a lot of math so yep. But to me it seemed .. natural .. it would happen this way. Heck I remember a team building exercise built around this, the same model as this except the two sides can communicate. And the goal was to teach the people to work together to win win for everyone (as like in this one, they would agree to go up-right so they each get the best payout... the equilibrium can be shorted by communication and knowledge lol). Except in my case I figured this game would be one where later rounds it goes up, that sort of thing, and negatives get involved to reinforce. So I decided to play a different game, on top of the game, one where I win on my own personal goal (I get the most or lose the least if everyone loses .. I win!). And that meant agreeing to work togther on round 1 then backstabbing them to max my profit! Oh it was wonderful. The next round NO ONE on the various teams would trust the answers from anyone, they all tried to back stab each other, and the whole thing descended into chaos! The teacher took me aside at the end and said they had never seen that course devolve like that, and I answered that's because the course does not deal with the reality of human nature and by the way I won the game I was playing. LOL.
And you know that is the real answer. Smith and Nash are really both wrong. As they assume in the structures and rules that everyone is playing the same game with the same rules. But when someone plays a different game for other reasons, over the top of "their" game .. amazingly destructive results can follow. LOL