Models are emphatically not nature.

They are at best an attempt to use our tools of reason and observation to predict nature, and you can point to thousands of examples, from market crashes to election outcomes, to observe how they can be flawed and fail to predict.

> the future is made up of 8B+ independent actors and they're going to react to incentives

= “nature”, if you're an evolutionist

as for models, all models are wrong etc.

Game theory is no more a description of nature than Euclidian geometry is.

Sure, but go try to build a house while ignoring what Euclidean geometry tells you about it.

I agree that both geometry and game theory are very useful tools.