> Proof lacks that it actually works that way

I mean everyone know it doesn't really work that way.

The actual question is: does viewing the average actor as trying to perfectly optimise their utility function using all the information available constitute a good estimation of how actors work in aggregate and does it yield accurate and interesting predictions?

The real insight of Simon in Models of Man is not that actors are not in fact perfectly rational. It's that you can actually model the limits of actors while keeping a fairly rigorous and manageable formalization.