Google is bad, but people have no idea how bad search could actually get. Trying to find DPRK housing stats and DDG will only give me pages about recent canadian housing news.
Google is bad, but people have no idea how bad search could actually get. Trying to find DPRK housing stats and DDG will only give me pages about recent canadian housing news.
I will always suggest Searx. It’s a meta front-end, filters out all the AI garbage and ads. FOSS and self hostable but there’s plenty of public instances to try. You can customise which engines it uses, although what’s available to choose from is up to the instance maintainer.
Came here to say this but you beat me to it. So instead I will second the Searx suggestion.
Searx (technically now a newer development called Searxng) will grab results for you from just about any other search engine but it goes a long way towards anonymizing you to those search engines and helps break their ability to fingerprint and track you, removes ads, etc. If you use invidious for youtube (you should!) or nitter for “x” (xcancel is a nitter instance), then Searxng is kinda like that but can also aggregate all the search engines for any search you do.
You can host your own private instance which I’m told is easy, or you can use public instances, which I do. I do have to admit that a lot of the public instances get rate limited a lot, so you might get a few “0 results returned” before it works, if you have it set to use for example google and bing when both have rate limited that public instance. You can set it to use other engines at that point or try other public instances. Usually it works fine “out of the box” but sometimes it does take a little tweaking to get it the way you want. Mostly this is because search engines don’t like people using searx and try to thwart it because it ruins their raison d’etre: tracking you and advertising at you. But that’s what makes it all the more worth it. And if you set up your own private instance, then it greatly reduces or eliminates many of the issues that the high volume public instances have to deal with like rate limiting, and of course gives you full control over how you want your personal instance to work.
And yes, !bangs work on it, but you should use two exclamation marks instead of one, like “!!w beanis” to search wikipedia for beanis. Dorking more or less works to the extent that it works on whatever the target search engine is. If you’re using searx to search on an engine that doesn’t recognize boolean search operators, then it’s not going to return results as if the operators do work. But it will work when set to search on engines that do recognize them.
Setup is indeed very easy, I run a local-only instance on my PC at home. Been a while since I did it but from memory I think it’s just a docker one-liner. You are right that rate limiting can be an issue and that’s part of the reason I spun up my own.
+1
I thought about building a network of these to try and avoid rate limits. Not sure how viable that is, maybe its just a host your own type of service.
It’s easy enough to host your own, and if you’re just using it like a normal person you’re not going to hit a rate limit. Goog can’t really tell the difference between you hitting it from a browser vs. you hitting it from searx.
You could probably either proxy outbound requests through a bunch of different boxes if you really wanted to, or maybe there’s some frontend that queries public instances in like a round-robin kind of way (a meta-meta engine, I guess), but honestly I suspect that’s more work than just spinning up your own.
iirc the searx engine config yaml accepts a list of searexes too. Keeping the list updated with working instances is annoying tho. Maybe less annoying with friends so you don’t have to guess about reliability.
That is incredibly useful. I didn’t know it did that but it stands to reason, it’s just another search URL after all. In that case it’d probably be fairly trivial to just scrape the list at searx.space and jam it in the yaml file say once a week.
I’m sure it’s a config issue on my part, but I self host it and my internal instance gets like instantly rate limited on a bunch of services, so i often still use public instances since I can’t get a very good breadth of engines on my self host one.
Not sure how to use this all results were in Russian or Cyrillic
The list lets you find an instance by country, then after that results depend on which engines are enabled. Check the settings on whatever instance you select.