I still don't see it. I could see it if that plasma somehow limited the spatial positions in some predictable ways of the particles in it which would be less random than the position once the particles coalesced into atoms. Looks to me like just going from a plasma (fire, loose electrons, protons, etc) to atoms is a massive loss of entropy. If the space is small enough in the plasma phase and big enough in the atom phase, maybe.

Still would like Tim on this one.

I've had to calculate entropy (randomness) in many (security) systems and it is always hard to do and know you have it right. In fact at the scales of the universe I would not even know how to know I was probably right.

Thanks for the interchange!