Actually, not really a bug. It's some code that was purposely written to filter out certain naughty words. But it's toast as soon as the programmer gets to it.
I copied a novella-sized public post onto my clipboard and pasted it into a PM to Matt while he watched the CPU utilization of our backend server, and it spiked from the 15% it usually hums along at to about 70%. Really vicious overhead, as I suspected since it was parsing every word in a memo field (for non-geeks, what I mean is "working really hard")
The feature was doomed before our little test, though. Profanity's against the Terms of Use, we've got people who're conscientious about stopping it, and really don't have the problem with inarticulate buffoons that a couple of other sites have, so we don't need it. Besides, I don't like software meddling with user-provided content anyway. Seems just real wrong.
Some things have to be handled by humans. Software not only is incapable of handling the subjective calls, it's darned tough to even remotely approach infallibility on the objective calls. And somehow, I've got a real problem with a machine telling me I can't mention how nice the *****-willow trees are looking around here right now.