Yeah, they really ought to work from the version of robots.txt archived along with a given snapshot. If the legal owner of the content wants it taken down, they can submit an explicit request for that.