This is pretty interesting post. I really like the relaxation of the rationality constraint for these models because it's so clearly not true. As someone who casually reads econ/game theory stuff as mostly a hobby I'm really looking forward to the field starting to harness modern data science techniques to get something closer to science