It is argued that realism and true randomness are fully compatible. Realistic true random events are acts of pure creation that obey strict laws, but do not necessarily satisfy Kolmogorov's axioms of probabilities. Realistic true randomness is some sort of nondeterministic force, or propensity of physical systems to manifest such and such properties under such and such conditions. Realistic random events reflect preexisting properties, as required by realism, simply the reflection is not deterministic; still, the preexisting properties determine the propensities of the different possible events. It is argued that deterministic extensions of quantum physics are necessarily incompatible with special relativity. Hence, from today's violations of Bell's inequalities one can conclude that all future physics theories will display true randomness as does quantum physics. It is argued that accepting true randomness and realism leads to new questions with interesting answers, allowing one 1) to study nonlocality in configurations with many independent sources and 2) to bound how much free will is needed for a proper violation of Bell's inequality.