Modern cosmology has, for several years now, been characterized by persistent empirical “tensions” between different measurement methods and cosmological epochs. These are usually interpreted as local inconsistencies that should be resolved within a unified theoretical framework through improved data or model adjustments. This paper argues that this interpretation obscures a deeper methodological problem. The persistence of cosmological tensions is less an expression of faulty data or inadequate equations than the consequence of an overextended claim to model-theoretic unity.The analysis shows that cosmological practice already operates, in fact, with different modeling regimes for different scales and physical contexts, without explicitly acknowledging this domain dependence. As a result, tensions appear as anomalies rather than as indications of limited domains of validity.The paper proposes a methodological reorientation in which a global cosmological world model is replaced by structured model sets consisting of domain-specific, empirically testable models. This also shifts the level of falsification from individual parameters toward the model architecture and the assignment of domains. Such domain-aware modeling increases epistemic transparency and better reflects the growing precision of cosmological data. The proposed reorientation can also be understood as a structural response to an efficiency threshold: persistent tensions then mark less local defects than signals of architectural overextension and the need for reorganization at the level of model design.
Stefan Rapp (Sun,) studied this question.