>Since you have all message histories, you should be able to simulate what charges members would have incurred under various pricing models.
Won't work. Pricing always changes behavior, as you can observe in the great sociological experiment called "mobile services pricing plans".
Presently, we had the same questions being asked over and over ("will VFP work on Windows nn", for one) because the returning members couldn't search but could ask a question for free. Now with new questions being under a cap, assuming the search was now free, they'd probably search for what they could find, and then continue some old thread.
People always do what they are paid for, and prefer what they don't have to pay for. So applying new rules to old data is an illusion - those data wouldn't look like that if these rules were in effect.