We propose a continuous version of the classical Gale-Berlekamp switching game. The main results of this paper concern growth estimates for the corresponding optimization problems.