As an undergrad, I did some research work in a bio-computation lab trying to model the basal ganglia (a part of your brain that helps you move and learn). I left shaking my head at how much guess work, "fine-tuning", and hand waving there was in the field. The biologists didn't appreciate how crappy code can deceive and the engineers didn't appreciate the mind-boggling complexity and non-imperative world of biological systems.

I can't even imagine how much parameter tuning and hacks went into a model of such staggering complexity. Paraphrasing one of my old academic advisors, the curse of models is that you can always make them look good.