We know that the harder a question is, the more points it is worth, and the easier a question is the less points it is. So, how do they determine difficulties of each question and assign point values to each question? It wouldn't make a lot of sense to just decide this completely in advance.
What I have wondered is if the difficulty and therefore the point value of each question is determined by how the overall testing population of that score release window does on that question. That would explain the passing rates and the score release timing.
So, if any of that is true, it is not technically curved in the traditional sense, but it really kind of is as all scores are based off of each other.