When observations appear to violate the laws of thermodynamics (hot water freezes sooner than cold), there's something wrong with the data. The "proof" is other people making a similar mistake.