The standard LCDM model of cosmology has managed to explain a wide variety of observations with only six free parameters. Nevertheless a combination of surveys ranging from modern CMB experiments to mappings of the galaxy distribution reached a precision where extensions of the model are required. One 'known unknown' are neutrinos, and in this talk I will explain why they could be particularly important to take into account when looking for 'unknown unknowns' such as signatures of modified gravity. One of the most exciting frontiers to test cosmology now is the growth of structures in the nonlinear regime and I will mainly focus on the abundance of galaxy clusters. They yield tight constraints on deviations from general relativity and can constrain neutrino masses better than any laboratory experiment to date.