Predicting the past: a philosophical critique of predictive analytics

If we address this topic from a conceptual and critical point of view, we need to address three issues: 1) why predictions are too often right, 2) why, at the same time, they are so often mistaken, and 3) what consequences arise from the fact that our instruments for prediction ignore at least four realities that must be true about future forecasts or at least be conscious of their limits: a) that individuals cannot be fully subsumed into categories, b) that their future behaviour tends to have unpredictable dimensions, c) that propensity is not the same as causality and d) that democratic societies must make the desire to anticipate the future compatible with respect for the open nature of the future.