Once again, I find myself strapped for time (the Madness of March has inflicted me) and so I can only offer limited comments on the very heady posts of late. My objective in this post is to highlight my views and concerns, particularly where they may differ from Henry's or Joe's. I hope this conversation keeps going so that I can add more later without the conversation having long past into the archives.
First, Henry, your post was very impressive and thought-provoking. My immediate concern is that I am not really sure what are the dividing lines between the physical sciences and the social sciences. My best guess is that you would say that a social science is that which is not susceptible to the scientific method -- if in fact that is your position, I would recommend the term "humanities" instead. Social sciences like psychology and sociology have made important advances through the scientific method, and I believe it is unfair to discard the entire field because not all academic inquiry within a field meets rigorous scientific inquiry.
Second, I believe that the softer sciences play an important role in the development of understanding, although I readily agree that many conclusions that social scientists reach extend too far. As Joe noted in a recent post, the same can be said for the frontier of natural science. In my view the criticism you make is better framed as one of overreaching. Yet that overreaching is what causes the next generation of the curious to challenge existing claims and add to knowledge.
Third, I disagree that the objects of study within social science are chaotic, at least any more so than the materials studied within the natural sciences. I recognize that strictly, a system cannot be "more" or "less" chaotic, but I believe what we see as an impenetrable morass of complexity may become coherent through advances in technology. For instance, neuroscience has allowed us to better understand human behavior and will continue to add to our genuine knowledge. Even if there are many variables that play into a complex system, there may be ways to isolate one or several events to demonstrate a causal relationship. I do not understand chaos theory that well, and maybe someone with better understanding of the subject could explain, but it seems to me that understanding a complex system must begin somewhere, and chaos theory allows a starting point.
I think Henry's post will give us fodder for discussion for a long time to come. I would find it helpful to take a point at a time, and Joe has offered some good counterpoints (most of which I stand behind) with respect to behavioral economics. Maybe we can focus on that point for starters and work from there.