Feynman's argument was about why you can't have systems that decrease entropy continuously. It doesn't disprove that there can (and will) be downward fluctuatio

Feynman's argument was about why you can't have systems that decrease entropy continuously. It doesn't disprove that there can (and will) be downward fluctuations in entropy, which is what I believe the OP was asking about. As someone else mentioned this is the idea behind a Boltzmann Brain (which is a pretty fascinating concept).