The convergence of binary majority consensus algorithms is studied in networks with different types of disturbances. It is shown how randomization can foster convergence.