Models will never be able to predict future real-world behaviour
More and better abstract analysis can’t lead to convergence on an absolutely true account of future real-world behaviour. Model-generated knowledge can only ever reach absolute certainty with reference to the model itself, rather than the territory that the model seeks to represent. For the messy, complex socio-technical situations to which RE feasibility assessment relates, there is no independent basis for assessing whether the models themselves are sufficiently fit for purpose. Models tell us what might happen, if their assumptions hold in future real-world conditions.