--and our "terms forbidden in usernames" list is both a) regex-based instead of plain words and b) about 300 items long. A service the size of Twitter would have lists in the hundreds of thousands (at LEAST), probably have a two-level list (autosuspend and flag for human review),

and have a list of "accounts that are excepted from automod" that I'd conservatively expect to be in the hundreds of thousands. (There's also a TON of accounts of slightly less prominent terrible people that *would* be in that list if it were what it claims to be.)