I'm not aware of this particular case, but I would say, obviously, that I'm sorry to hear it. I'm glad it sounds like we ultimately got to the right place on this.
I think that's why we should be doing better. It is hard. As I said, this will always require a combination of automation and human review. With automation—I think it sounds like you kind of missed this sort of thing—the context is important. That's why we can never really say that only the machines should decide. We need humans to do this because context is very important, and it's at the nub of what freedom of expression is.
I don't think we'll ever get to 100%. What we're trying to do is to do much better. I think we've come a long way in just the matter of a year, but obviously there is more work to be done.