Seeds of Science has kindly published my essay “Prediction and control in natural and social systems”. This short essay : i) address and contextualize the modest predictive success of Social Science and ii) describes what kind of scientific intervention on the human society can be successful. 

Its main arguments are the following:

Exact Sciences and Artificial Systems: Physics and Chemistry are considered exact because they operate on artificial systems developed through scientific processes, unlike complex natural systems. The precision seen in Celestial Mechanics (the only natural system on which we can consider Science to be exact) is replicated in other physical systems like lenses, batteries, and microchips, but they are human-made. We mostly have exact predictive science on systems designed to be predicted. Even chemists' first task was to create pure substances, which are essential for applying the rules of stoichiometry and chemical thermodynamics.

Complex Natural Systems: Beyond exact sciences, other natural sciences study complex systems, both animate and inanimate. Complex systems can be categorized into modular systems, which allow controlled experiments (vg. Medicine random trials), and non-modular systems, where direct experimentation is not feasible (the system is too big, no easy control points are found). 

Non-Modular Complex Systems: These systems, such as the atmosphere, ecosystems, and populations, are studied by Climatology, Ecology, Population Genetics, and Evolutionary Biology, and are not easily amenable to controlled experiments on due to their scale and complexity (that imply that it is difficult to alter a single element alone in the system).

Social Sciences: When compared with natural sciences studying complex non-modular systems, social sciences' predictive success and techniques align with those of similar natural sciences, challenging the notion that social science is less rigorous.

Institutional Design and Progress: increasing the "hardness" of any science involves moving from observation to control, and that is why the Mechanism Design research agenda is superior to the policy design paradigm. Unlike the High Modernity project of vertical rule by a technocratic class, institutional design is (and has been since Solon and the American founding fathers) an inclusive project where the intellectual elite proposes the rules for the efficient integration of the preferences of the entire demos, instead of instituting itself as a ruling chaste.