Aren't you overstating the fundamental nature of the (second) law of thermodynamics a bit? The simple way it's usually explained is via a statistical argument based on statistical dynamics. If you have many atoms moving more or less randomly then it is much more likely that they will form a disordered state then an ordered one. E.g. if you let two gasses swirl around in a room then they will mix, not separate. Not because there is some fundamental law that says they should, but because there are much much more configurations of the atoms that lead to a mixed state than a totally separated state. Seen in this way the second law isn't so fundamental at all, but an occurrence of something that goes against it is just extremely unlikely.
Not that any of this would rescue the idea of a perpetuum mobile.
