On "realistic" models

On the first day of my macro-eco class, my professor was discussing models (the economic kind, that is). He had me stupefied when he asked me what my response is if people ask me "What are we gonna do with your simplified models? They don't model real life, they are useless." At that time my mind was thinking about what kind of burger I will have after class because I was so darn hungry. After a few seconds of silence, hoping that he will get that I have no idea, I stammered and lamely replied "Ahm... er... wrong assumptions? B-b-because we have wrong assumptions?" Nope it wasn't, according to him, models are supposed to be simple and not necessarily reflect reality.

That got me to thinking. And I wonder if I agree. What are we supposed to do with non-realistic models? Economics has a great deal of theories. It seems that on paper, its able to solve everything. But looking around. Our economy is not perfect. I believe all the theories we study are logical and practical, but they will only be useful if we are able to apply them in the real life and eventually help us efficiently allocate resources. Fictitious models, however interesting and logical they may be, are of no use.

So what do we do with them? Take the theories, apply them, but make sure that it's applied on the right set of assumptions (backed by correct data of course). I remember an upperclassmen in my undergrad study saying to me the only reason why Economics fails to solve all problems in the world is because they have the wrong assumptions. I fully believe him.

That being said, I have two things on my mind:
(1) I love the book FREAKONOMICS - it simply presents unconventional findings and proves them correct by showing actual data. Regression analysis rules!
(2) I'm excited to take up statistics this semester. I've yet to make Dr Rufino (undergrade forecasting prof) proud of me again. The only bummer is that my class is 3 hours long in the evening, and my teacher just reads from his slides. Ugh.