Satisficing equilibria

I presented the latest research on approximately-rational Nash equilibria. Pure Nash equilibria is, conceptually, a robust and strong notion that, if it exists, it is a good predictor of the behavior of rational agents. Sadly, not all games admit a pure Nash equilibria. In fact, only (1 - 1/e) portion of large games have a pure Nash equilibrium.

In contrast, if one of the players is allowed to respond with one of their top 2 actions (instead of their best response), then asymptotically almost all games have a “(2, 1)-satisficing” equilibrium.

In this class, I presented these results and the main technical tool in the proof.

Here are pictures of the board:

Here is the audio of the class: