There probably is a better solution - likely one involving some minimal human moderation paired with machine flagging, but the technical challenges don't make obscenity filters a bad idea. A young startup could face a make or break moment when its site starts attracting obscenity and thus repelling all its early users. This is clearly worse than having some odd misspellings. Consider the worst case scenario of implementing an obscenity filter and not doing so - the decision becomes quite clear, it would seem.
