I think something that surprised me is the idea of incorporating "noise" into models. We can't really expect models to reflect real life if the conditions are pristine and the data is without uncertainty because nature itself has some essence of unpredictability. I also thought it was interesting how Palmer noted the brain as something he was particularly interested in because it had a lot of noise -- I agree with this statement. I think that it is important for making models more accurate for the future.
A question I want to ask him is what he thinks about the energy produced by generating models. I understand that there is a lot of energy used for modelling and internet use in general -- is there a way to offset this? Are the positive gained from the models offset by the energy released by the modelling itself?
Your point about needing real life models to not be pristine due to nature's imperfections made me think about how important it is to delineate the difference between seeing imperfection models and prediction though. In other words, while I think that it is true that models that model life themselves may not be perfect, it's important to say that models themselves should try to approach as perfect representations of imperfect things. lots of words there!