I'll actually talk econ in this one. I was a bit amused at what seemed to happen during one of the sessions. The presenter was talking about bans on smoking in "public" places, and how proponents argued that it was a myth that such prohibitions would hurt restaurants' earnings. The presenter said that statistics showing an increase in aggregate earnings weren't proof, since some restaurants could have benefited and some could have suffered.
In the Q&A I asked how this could be possible. Since we were assuming Coasian arrangements (which was the whole point of the session), if the restaurants all optimized before the ban, the institution of the ban could only hurt them. So wasn't it weird that aggregate profits went up?
The chair of the session jumped in to show me that my logic was wrong. He pointed out that even a pre-ban restaurant that had a no-smoking policy could have been hurt, since it would have lost its niche market. I answered by saying, "Right, that's exactly my point. Those restaurants are hurt, and clearly the restaurants who had catered to smokers are hurt too."
People were kind of stumped, and then one guy in the back said something like, "No, I've talked to restaurant and bar owners, and they don't like smoking. It ruins their clothes, employees need to be paid more, it ruins the walls, etc. But if one restaurant unilaterally bans it, it loses too much business. So they form a cartel and have government ban it everywhere. Our analysis is still right: The ban hurts consumers more than it helps the restaurants."
Everyone was satisfied with this and we moved on. But it struck me that this was a totally different theoretical justification from the one initially offered. I.e. we all knew the ban was stupid, and we just kept adjusting our argument to fit the facts.
Incidentally I'm not saying the people in the room did anything wrong or unscientific. I'm just pointing this out, because if something comparable happens in the leftist camp, I imagine many libertarians would flip out at the unethical statists.