Well, of course. The question is "did you make more money with that than if you hadn't grandfathered?"

Who knows: there are too many variables at play, and it's not worth doing experiments with lest people raise a stink.

Yep, but that's why data would be valuable (although I don't think anyone will have any, due to the reasons you mentioned).

Due to the many variables, I think even rigorous data for one application might not apply to others. LiberWriter's customers are different from customers of other web businesses...