If the internet is at its core is a system of record, then it is failing to complete that mission. Sometime in 2014, the internet surpassed a billion websites, while it has since fallen back a bit, it’s quite obviously an enormous repository. When websites disappear, all of the content is just gone as though it never existed, and that can have a much bigger impact than you imagine on researchers, scholars or any Joe or Josephine Schmo simply trying to follow a link.
Granted, some decent percentage of those pages probably aren’t worth preserving (and some are simply bad information), but that really shouldn’t be our call. It should all be automatically archived, a digital Library of Congress to preserve and protect all of the content on the internet.