A crease is a self-contacting fold in the surface of a soft solid. While being closely associated with an unusual, infinite wave number linear instability that occurs at large strain, the real instability that precipitates the appearance of a crease is a zero-threshold non-linear instability occurring at lower strain. There is no way to predict this lower-strain instability from linear analysis -- indeed linear analysis wrongly predicts stability. Rather we can understand the lower critical strain through the spontaneous breaking of scale-symmetry, and a related topological change, in an auxiliary problem. The theory I will discuss gives excellent agreement with experiments on creasing, and is potentially useful in understanding fracture, cavitation, plasticity, and phase transitions (both in and out of equilibrium) -- especially nucleation phenomena.