Explained causation is too far away down the road IMO, first you need good models for describing the process. Then you can gather the data adjust the models accordingly and look for causes.
In any case
http://en.wikipedia.org/wiki/Chaos_theory
will have something to say about the accuracy of your predictions.
