You seem to have it exactly backwards: we have 200 years of practical evidence in favour of modern fluctuation theory thanks to things like steam turbines. Structure does sponataneously appear in a gas in thermal equilibrium, one can show this in a classroom experiment.
Landau and Lifshitz vol 5 <https://en.wikipedia.org/wiki/Course_of_Theoretical_Physics#...> is the standard textbook. There's an older copy on the Internet Archive <https://archive.org/details/landau-and-lifshitz-physics-text...>. (Having the background of L&L vol 9 makes a classroom demonstration even easier: a small handful of electronic parts for a resistive-capactive low-voltage DC electric dipole, a decent oscilloscope or other apparatus to measure and record fluctuating voltage, and a thermometer).
The probability of an out-of-equilibrium structure spontaneously fluctuating (briefly!) into existence depends on the complexity of the structure, and Boltzmann brains are much much much less complex than the whole Earth, solar system, Milky Way, or the early universe in which these structures' precursors originated. So therefore any theory compatible with statistical physics in which the early low-entropy state of the universe is a fluctuation in a higher-entropy "gas" is imperiled.
For the philosopher or two who wrote comments above in this thread, https://plato.stanford.edu/entries/statphys-Boltzmann/ is probably of interest.
