For example (just as rough proof of concept)
Research articles: Pretty much any academic library website for published stuff, arxiv.org for white papers in computer science and math, pubmed for biosciences
audio equipment: gearspace.com
Encyclopedia: Wikipedia
Thesaurus: Wordhippo.com
The idea is to avoid needing to visit a search engine. Might be a high idea.
For a while DMOZ was the authoritative source for whether a website was real and relevant for a given subject. It was 100% human-curated (I administered a couple of topics for years) and was so trusted that putting a site into a category could get it to the first page of Google pretty much guaranteed. That power waned over time, but not because Google found something better, but because their motivations changed. Maybe it’s time for a dmoz comeback. Maybe something federated…
God I miss a well tended DMoz.
And then when Google copied it and sorted category links by PageRank? The good old days.
DMoz’s not well tended successor is Curlie.