You'd lose all the long-tail quality stuff (e.g. my own write-only blog :P ), but also all the garbage. It might be worth it.
I don't think you could kill Google with that, but it might make a viable lifestyle business.
Edit: and I meant to add - I was looking for info on a specific Synology device the other day. Google turned up Synology's own website (good) and a zillion crappy affiliate-link-spam sites. eBay and Reddit were the best sources of info when I sighed and started constraining the search with site: prefixes.
I know basically nothing of the mechanics of search, although I expect that a boutique search engine would have trouble accessing the guts of third party sites as they don't want to get hammered by the world.
I did just think of an interesting specialty engine. If you had fast access to the full library on libgen or sci-hub, a person could produce a pretty interesting set of abilities.
My impression is that if you're not dealing with people trying to game the algorithm then it's all a lot simpler.
As to data sources - with a constrained set of sites the creation of custom adapters should be do-able I'd have thought.
Not that I have any plans. Maybe I should though :)
Edit: come to think of it, it's practically an internet law that stuff you think of already exists. So maybe I should just post an Ask HN for the link to the search engine in question...
Exactly. I wonder how a person could construct a highly curated google search. Not just blocking of sites but also a bit of parsing on the site for disposal purposes. Could it all be done on a local machine?
On a related note, as eBay became more of an all-purpose webstore I'll tend to use it rather than Amazon search to find products. The actual purchase, if it happens, is wherever.