Widely used bottom-up energy system models are based on the minimisation of the discounted total system costs. This conference paper reviews the origins of this rationale and questions to what extent does cost optimisation approximate the real- world transition. Using the bottom-up D-EXPANSE model for the power system of the United Kingdom, the historic transition between 1990 and 2010 is modelled retrospectively. The cost-optimal transition pathway turns out to be significantly different from the transition that actually occurred. The deviation of 17% in terms of cumulative total system costs between 1990 and 2010 is observed. This is a novel and thought-provoking finding for the bottom-up modelling community because it shows that cost-optimisation does not necessarily approximate the real-world transition. Therefore, the implications of allowing such a high deviation from cost- optimality in modelling the UK future power system transition (2010-2050) are explored. Up to 17% deviation leads to significantly different transition pathways in terms of technology deployment. More caution is thus needed when interpreting the results of the existing bottom-up models that are based solely on cost optimisation. Further research is essential on systematically exploring the near-optimal transition pathways.