On the other end of the spectrum, here is a short look at why using data to predict complex systems will never be satisfactory. The problem with this viewpoint, though, is it discounts any attempt at trying to predict. We won't know what truly lies within that realm until we fail at predictive models. Using scenarios and robustness / resilience measures is a means to dealing with the issues that are currently unpredictable. I wouldn't put biological and ecological systems in the unsolvable category yet.