Liz Reid, Google’s head of search, said in a blog post that the company had made adjustments to its new AI search feature after screenshots of its errors went viral.
SearX/SearXNG allows you to aggregate results from a number of different search engines. You choose which ones, and they’re stored in your browser without setting up an account.
@makeasnek On a broader note, I think possibly the best approach for decentralised, open-sourced web search might be an evolution on the SearXNG model.
At the top of the funnel, you have meta search engines that query and aggregate results from a number of smaller niche search engines.
The metasearch engines are open source, anyone with a spare server or a web hosting account can spin one up.
For some larger sites that are trustworthy, such as Wikipedia, the site’s own search engine might be what’s queried.
For the Fediverse and other similar federated networks, the query is fed through a trusted node on the network.
And then there’s a host of smaller niche search engines, which only crawl and index pages on a small number of websites vetted and curated by a human.
(Perhaps on a particular topic? Or a local library or university might curate a list of notable local websites?)
(Alternatively, it might be that a crawler for a web index like Curlie.org only crawls websites chosen by its topic moderators.)
In this manner, you could build a decent web search engine without needing the scale of Google or Microsoft.
@makeasnek @schizoidman YaCy is still around.
And https://searx.space/ is an open source metasearch search engine with many instances. (Try https://searx.be/ if you want to test it out.)
SearX/SearXNG allows you to aggregate results from a number of different search engines. You choose which ones, and they’re stored in your browser without setting up an account.
Interesting thanks I’ll take a look!
@makeasnek On a broader note, I think possibly the best approach for decentralised, open-sourced web search might be an evolution on the SearXNG model.
At the top of the funnel, you have meta search engines that query and aggregate results from a number of smaller niche search engines.
The metasearch engines are open source, anyone with a spare server or a web hosting account can spin one up.
For some larger sites that are trustworthy, such as Wikipedia, the site’s own search engine might be what’s queried.
For the Fediverse and other similar federated networks, the query is fed through a trusted node on the network.
And then there’s a host of smaller niche search engines, which only crawl and index pages on a small number of websites vetted and curated by a human.
(Perhaps on a particular topic? Or a local library or university might curate a list of notable local websites?)
(Alternatively, it might be that a crawler for a web index like Curlie.org only crawls websites chosen by its topic moderators.)
In this manner, you could build a decent web search engine without needing the scale of Google or Microsoft.
@ajsadauskas sounds like what @pears is trying to do
@makeasnek