This is Joseph.
Andrew Gelman has a very nice piece on the challenges of causal inference over on his blog, looking at the discussion between Noah Smith and Bret Devereaux. As a conflict avoider of heroic stature, I don't like people I follow fighting but, in this case, it is bringing up some good points.
But going back to just understanding patterns is also a very important step in epidemiology as well and there is a new movement towards trying to do more rigorous descriptive work. For example, this paper by Catherine R Lesko, Matthew P Fox, and Jessie K Edwards. I have some disagreements with the authors but it is a fantastic piece to get epidemiology thinking clearly about these issues.