Predictive Science vs Control Science

There are two types of sciences. Yesterday I wrote about Prediction vs. Control, which talked about how complexity affects the scientific process. There is Type I, Physics, Chemistry and some Biology, which are predictive sciences. They’re able to take things apart, put it back together, predict the outcomes, and come up with the basic rules by which the world operates. Then there is Type II, Economics, Medicine etc., which can only talk broad generalities, fix problems when they come up, and mould processes so that they result in broadly desirable outcomes.

This is a paradigm shift in how to think about sciences. It’s understood that the scientific process is all about testing and refining, but that only works if the counterfactual is measurable, and if the hypotheses are testable. In more complex fields, this is increasingly difficult, meaning that “control” might be a far more important goal. Complex Adaptive Systems (CAS) of all sorts therefore requires a fundamentally different approach. The lack of predictability is a feature of these systems, and therefore the outcomes have to be tested according to principles of “control science” and are closer to management than they’re to laboratory testing.