The assumption is that the vast majority of conference-goers have good will and enough social skill to navigate these things adroitly. As is explained in the article, it's a scaling problem. If 99 people out of 100 are good eggs, then a conference of 2300 still has ~23 bad eggs.