[Avamander]

Some mistakes are also worth recording. I like seeing bad predictions of the 2000s from the 1970s for example.

Not to mention that quite a lot what is archived today has been made by companies, there's no "right to be forgotten" that companies could ever deserve. For example I've uncovered quite a few mistakes in currently public datasets/websites based on archived sites, who knows how many mistakes are made now and never fixed because we lose the original sources. Point being that the lack of original source doesn't mean the information gets lost, it just becomes a big version of the kids game "telephone" where everyone recites what they heard and it gets distorted in the end.