Elementary explanations of the second law of thermodynamics refer to probabilities of system states and seem convincing. But not when considering time-reversals, because the same statistical arguments should also apply there but they produce contradictions regarding entropy increases with time. (I think the difficulty is in whether or not the assumption of statistical randomness is appropriate because it depends on what is given and maybe also on the direction of time but I'm not an expert and this doesn't answer my question anyway.) While reading some literature about the direction of time I learned that the direction of time and the second law of thermodynamics all come from a very low entropy immediately after the big bang, with increasing entropy produced by things that include gravitational clumping (e.g., the formation of black holes and the merging of black holes to produce larger black holes). I learned that this is responsible for the second law of thermodynamics but it seems to me that this is an incredibly large-scale thing. Given this explanation it seems amazing to me that we can randomly select a tiny piece of matter (large enough to be macroscopic but tiny from the point of view of human perception) and find that it obeys the laws of thermodynamics. Is there an explanation of how such large influences on entropy (e.g., objects produced by gravity clumping) can produce a second law that is so incredibly homogeneous that we find the law obeyed by all of the tiny specs of material?