This talk investigates the role of feedback in a network of amplifiers. We start by revisiting the theory of the individual feedback amplifier. Then, taking inspiration from neuronal behaviors, we present a simple architecture that makes the network amplification zoomable, allowing for a versatile modulation of the resolution of the network amplifier. The network property results from a feedback localization property of the individual amplifiers. We argue that zoomability is a fundamental property of controllable network behaviors and that positive feedback is essential to localization, This calls for a rehabilitation of the positive feedback amplifier in a field dominated by the magics of the negative feedback amplifier.