But even before that, when there were no search engines, there were directories (https://dmoz-odp.org/) which tried to take a taxonomic approach to the whole web, webrings, and putting your URL into print to get people to type it in. The latter culminated in an early version of the QR code: regular barcodes and the "cuecat" barcode scanner.
I think a lot of the modern Web's problems can be traced to the desire to cut out human intermediaries in the early Web, mostly to focus on speed and monetization. (No curators in the models means no need to PAY curators). I also think the rise of the influencer and parasocial relationships are trying to fill the void. Most humans want human context in information searching. Those of us who don't need it are the weird ones, not the standard case.
No, they don't. Ask any librarian about how many patrons of their libraries only interact with them during the check-out process.
I'd argue that things like our cataloguing process counts as 'human context', as do things like our hold and check-out systems, library space arrangement, etc. Even if a patron doesn't interact with a staff member in person, there is a difference between a library and, say, Kindle Unlimited from a UX standpoint.
Human context and human curation doesn't necessary mean human contact.
Archives are similar: There's a ton of work giving things context and making things discoverable even if a researcher never talks to the archivist who's done these things.