Henry Farrell's interesting post on smoking bans reminds me of an ongoing question that I have never heard a libertarian answer satisfactorily. Smoking in bars and so forth is dangerous to bystanders who have pulmonary disease (the dangers of secondhand smoke to those who are not already breathing-impaired seem to be largely mythical). It's noxious to some other number of people who do not smoke. The libertarian rejoinder to the smoking bans is that bars could choose not to smoke if people wanted it. But in practice, despite the fact that smokers are a minority, and most people hate it, almost no establishment went non-smoking without government fiat.
This seems like a market failure. You can explain it through preference asymmetry and the profitability of various customer classes: heavy drinkers are more likely to also be heavy smokers, and they are the most profitable customers. Bar owners don't want big groups of people who are going to take up three tables for an hour and a half while nursing one white wine spritzer apiece. They want people who are there to drink. In a competitive equilibrium, they couldn't afford to go non-smoking because they'd lose their most profitable customers to all the other bars.
You can explain it, but this doesn't seem like a good market outcome by any measure. Let me be clear, I'm still against the smoking ban, even though I personally vastly prefer smoke-free environments; I think interfering with property rights like this has even heavier costs. But I also recognize that I'm in a minority. And I think that politically, if not intellectually, the success of smoking bans is a heavy blow to libertarian credibility.