We argue that giant jumps of current at finite voltages observed in disordered samples of InO, TiN and YSi manifest a bistability caused by the overheating of electrons. One of the stable states is overheated and thus low-resistive, while the other,
high-resistive state is heated much less by the same voltage. The bistability occurs provided that cooling of electrons is inefficient and the temperature dependence of the equilibrium resistance, R(T), is steep enough. We use experimental R(T) and assume phonon mechanism of the cooling taking into account its strong suppression by disorder. Our description of details of the I-V characteristics does not involve adjustable parameters and turns out to be in a quantitative agreement with the experiments. We propose experiments for more direct checks of this physical picture.
We suggest to use `fluctuation spectroscopy as a method to detect granularity in a disordered metal close to a superconducting transition. We show that with lowering temperature $T$ the resistance $R(T)$ of a system of relatively large grains initial
ly grows due to the fluctuation suppression of the one-electron tunneling but decreases with further lowering $T$ due to the coherent charge transfer of the fluctuation Cooper pairs. Under certain conditions, such a maximum in $R(T)$ turns out to be sensitive to weak magnetic fields due to a novel Maki -- Thompson type mechanism.
