Our observations of dark matter and dark energy are, at least so far, based purely on their effect on their combined effect on the geometry of the universe and therefore ambiguous in a fundamental way. As we progress toward building ever more precise maps of the universe, it will become increasingly important to understand what we can measure definitely, in a model-independent manner and what we can only say on the assumption of a model. I will discuss how to construct such model-independent observables, and the fundamental limits to our knowledge that can only be broken by employing parameterisations resulting from particular models. I will discuss how to approach parametrising our ignorance in a minimal yet consistent and physical manner and ask to what extent such modelling can actually account for anomalies we are beginning to see in cosmological data.