Anyone who's truly interested in this issue, I suggest reading "Fabric of the Cosmos" by Brian Greene. He addresses this issue in much more detail.
Yes, in a world where entropy is more likely to increase than decrease, you could chose any given point in time and show that entropy increases in both directions. This implies that our past was much more disorderly, which leads to everything you've read in this topic already.
Brian Greene pointed out that all of our physics theories are only thought to be "correct" if they correspond to experiments. Thus, a theory that makes every experiment ever done nothing more than an illusion invalidates itself. Thus, although a theory in which entropy increases in both temporal directions are more likely from a statistical standpoint, they are inconsistent with themselves and can thus be disregarded for the most part due to its paradoxical nature.
XI Wiki

