In the so-called hard science (e.g., in physics) models that are at the same time simple but powerful in explaining and possibly forecasting real-world phenomena are considered a plus. In social sciences (that, by contrast, are considered as soft sciences) like economics, politics, sociology, the situation is quite different. On one hand, there are very skeptical positions (summarized in the well known statement ``The only function of economic forecasting is to make astrology look respectable”), on the other hand there are a number of extremely simplified models that aim at representing very complex situations by using a priori hypotheses based more on “feelings” (if not just the willingness of the proponent) rather than the evidence provided by the existing data. A third-way is possible, based on the realistic assumption that in many fields it is possible to provide models and tools j to specifically support decision makers. The idea is to describe situations in which available data (for instance in economics) are used as input to software systems that use them as much as possible as they are used in the real world. Other software modules generate data that are not available (e.g., future data about the behavior of macroeconomics variables) by using stochastic procedures that aim at producing a set of realistic (credible) scenarios rather than a single forecast. Then, high performance simulations allow determining the evolution of existing data in the different scenarios offering to decision makers a quantitative assessment of the possible effects of their choices.