The article's argument is a bit like saying TLS protects plain-text passwords in transit, so there is no need to store them in hashed form in the database.
Sure, the article makes good arguments about the trust that is still implicit in E2EE, but it goes too far in its dismissal of it.