The issue is not SEO, it is google deciding that it knows better than I do what I want to search for and pushing useless irrelevant results to the top 10 pages.
huh update: another interesting semantic search engine, this one works like a language model but with a live index of urls and embeddings of them. seems to be a more direct style of prediction than some of the research retrieval engines I’ve tried: https://metaphor.systems/
oh I just use ddg to evade that problem. that and more playing with Google. I still like Google best for question resolution searches, but it’s true that you have to be more specific with Google’s current shitty English encoder.
kagi is way, way better than Google and Bing at finding the human web
Tried it. Gives the same actively useless results as google search. Thumbs down.
next levels of spicy above kagi, besides millionshort, are https://teclis.com/ and https://search.marginalia.nu/. there are some other interesting related engines on https://news.ycombinator.com/item?id=30783037 and even more on https://seirdy.one/posts/2021/03/10/search-engines-with-own-indexes/, notable indieweb mentions include https://wiby.me/ and https://searchmysite.net/ but I tend to prefer teclis and marginalia. kagi is apparently google+teclis, which I didn’t realize until just now, and means I probably should have just recommended teclis to begin with.
of course, it’s not perfect. there are still SEO groups trying to target these engines. but they’re more explicitly focused on anti-seo design.
The issue is not SEO, it is google deciding that it knows better than I do what I want to search for and pushing useless irrelevant results to the top 10 pages.
huh update: another interesting semantic search engine, this one works like a language model but with a live index of urls and embeddings of them. seems to be a more direct style of prediction than some of the research retrieval engines I’ve tried: https://metaphor.systems/
oh I just use ddg to evade that problem. that and more playing with Google. I still like Google best for question resolution searches, but it’s true that you have to be more specific with Google’s current shitty English encoder.