Prediction in the human sciences

I wanted to discuss this topic in class (in fact I started it) but there was not enough time. We all know that a one criterion of a good scientific theory is making testable predictions – very good theories make numerous predictions. But when we come to the human sciences many think that prediction is impossible because the human sciences deal with human behavior which is unpredictable because people have free will. However, when we study people as social groups they tend to behave as a unity with shared values, codes of conduct and some kind of order and structure. (I think, despite our free will we all stop at a red light.) So, personality variable can be irrelevant in describing social fact. What we can say about prediction in the human sciences is that it cannot be precise as in the natural sciences and some times it can completely fail. For example, no one saw the credit crunch coming. The reason for this is that economic systems are so complex they suffer from two problems (like all complex systems) . One is that they cannot capture the full detail of the underlying system. We cannot measure everything that affects the system and sometimes it is difficult to measure them with precision. This is called the problem of initial conditions. The other problem is that complex systems (like the economy and the weather) are susceptible to minute changes. Slight mistake in the measurement of the initial conditions will have very big consequences. They call this the Butterfly Effect (A butterfly flapping its wing over Japan can cause a storm in Brazil). This is not literal; it is just a sort metaphor. But the reality is not far from it, especially when we talk about highly complex systems where everything affects everything else. I think that the more economic systems grow complex the less our predictive ability becomes. Or may be we will discover more powerful tools of understanding. Every crisis paves the way for a revolution.