How would one build such a service in a way that is resilient against legal takedowns?
Also, it has to get its data from somewhere, and this source can still be taken down, I suppose.
Perhaps you can hash the original url, use that as a key and upload the ascii text of the article as a torrent.