2mo ago

search engine megathread?

I'd like to start a thread for people to share what they're using for search now that quality & relevance of results are in steep decline on all major search engines.

I'm doing this for selfish as well as altruistic reasons, mostly because my "stack" has gotten a lot less useful over the past year, because the search engines I use repackage Bing, whose ability to return relevant results at all seems to be cratering.

My (shitty and getting worse) solution:

DuckDuckGo as default search engine mostly for bang commands, in my personal opinion DDG's relevance has been shit for its entire existence
StartPage as where I direct most of my general searches (by appending !sp on DDG)
currently attempting to tune the Ublacklist extension to remove most of the clickbait results; this works, but it's not easy to set up if you don't use Google which seems to be the main search engine the developer tests with

The problem with this approach is that increasingly, Bing simply doesn't return hits on topics I know should have plenty to choose from. Filters only solve the issue of too many hits, not too few.

BTW I tried Qwant, but their claim of having their own index seems to be bullshit, the results look like repackaged Bing to me. And, the UX is terrible in Firefox for Android.

I'd love to hear suggestions.

tl;dr pls share what you are doing for web search these days in order to work around the rapidly declining quality of major search engines

65 comments

I use Kagi. Paying for a search engine isn't something that I ever saw myself doing, but I do honestly believe it's the only way they can exist that means the users are the customers instead of the product.
- Second for Kagi
  Not only solving the customer vs product problem. It really does have a number of great customizable features that easily make it the best.
  
  Also can be purchased via crypto for maximum privacy. They also offer something I eould call "AI Studio", where you can easily switch models, and the chats are fully private and don't share context. You can customize which domains to promote, demote, block, pin, and they have some nice bangs functionality.
  I stopped paying for all streaming platforms, and Kagi is hands down the best bang for my buck as I already use it every day (in contrast of binging a show every two months).
- Yeah, I meant to say in the post but forgot... I'm currently housing insecure, food insecure, etc... so I'm looking for a non-Kagi solution. But it's definitely on my radar to try when I do have $ to spare.
  
  Sign up and you get 100 searches for free, plus unlimited !bangs. Worth making an account for when you can't find what you're looking for on the others.

Just try Kagi and you'll never look back.
- I forgot to include in the post that I'm too broke for Kagi, otherwise I'd have tried it already :(
- Last time I checked Kagi didn’t have an option for people who just do a few searches a month so it ended being too expensive for me.
  I might have to check again though😇

I also use Kagi. It has the option for AI but doesn't shove it down your throat like other providers.
One feature I especially appreciate from Kagi are URL redirects. I can have it automatically replace parts of the URL from any search result. For example, I have YouTube results redirect to Invidious, and Reddit redirect to redlib, without me having to replace the domain myself
- How are you doing this? I can't get it to work. ^https://www.youtube.com/|https://www.invidious.io/ doesnt work for me and I can't find what to put in for redlib
  
  invidious.io isn't actually an instance of invidious. If you replace your rule with "^https://www.youtube.com/|https://inv.nadeko.net/" (or any other instance: https://docs.invidious.io/instances/), then it should work. For Redlib, I replace www.reddit.com with reddit.nerdvpn.de (although reddit often blocks redlib instances from being able to access their content).
  An easy way to see if services can be redirected like this is to open a link (such as a video on youtube.com) in your browser and replace it with another domain. If it pulls up the correct content on the new domain, then such a rule can work.

currently attempting to tune the Ublacklist extension to remove most of the clickbait results
This is one of the killer features for me in Kagi. The ability to buff or debuff websites in the results. Whenever I encounter an AI slopsite, I immediately block it in Kagi and it'll never show up again. Other websites are annoying, but occasionally return good results, so I'll just debuff them in the search results.
Arch Linux forums? Guix mailing list? Max buff!
That being said, I do wish there was a more open source friendly alternative or a search engine that didn't just rebrand Bing or Google results. Maybe SearXNG or Mojeek?
- That feature of Kagi is so awesome and I wish other search engines would implement it.

For Lemmy, I guess 22 comments counts as a mega thread.
- You gotta see what's happening in Hexbear

I'm gonna comment on my personal experience with a few search engines.
Startpage I stopped using because, and it took me a while to notice this, the searches aren't included in the URL. This means if I want to bookmark a startpage search for later or find a startpage search in my browser history I can't. This is a dealbreaker for me.
Qwant didn't have advanced search tools of any kind, and it also didn't work in Singapore (where I occasionally set my VPN to).
Mojeek ... is not ready yet.
Kagi I tried for a bit but I'm not paying a subscription to an American company for search.
Ultimately I ended back up on ~~Duckduckgo~~ Startpage. I think perhaps we should go back to web directories instead of search engines.
- https://www.startpage.com/do/settings
  Privacy and Safety
  HTTP request method > GET
  still a ddg user myself but theres how you solve your startpage problem
  
  Oh thank you. Now I don't need to use an American search engine.
- We could use the fediverse the find answers today as well? If we could index and search across all instances and threads?
  
  I have no idea what you just said.

searxng :)

4get https://git.lolcat.ca/lolcat/4get https://4get.ca/instances
lots of public instances which tend to be very stable + it's extremely easy to self host. in my experience it's significantly better than every other proxy search engine

Use Brave search, It has a great UI and has good search indexing
- Yep, probably better than google at this stage.
- Aren't the just soin shitcoin propaganda scheme or something. People complain a lot about Brave.
  
  Well the browser is kinda bloated with those 'shitcoin' stuff. As far i can say Brave search doesn't have any of those stuff.

I use SearXNG self hosted, and I see 4get....
- You know, I've tried using instances of searx, and it was never satisfactory to me. Often it would show "no result" found. IDK what I've done wrong. I wanna know how you guys use it.

I use SearXNG, self hosted

I like DDG but it just uses bing under the hood.
I actually prefer Brave. (Though brave has been caught fiddling the results)
They all kind of suck in their own way. Brave or DDG gets the job done 99%nof the time.

I grew tired of all of them and I’m using mullvad leta. Too bad they don’t support images but it’s the only one without ai. It’s not perfect and it crashes if you use “site:” but so far it’s working for me better than the others.
- Interesting, I'd never even heard of that one!

Yandex is deptessingly good
- Oh hmm... their image search is second to none but it never really crossed my mind to try the regular search, I'm not even sure why.
  
  Among the western slop, Yandex seems to be the most conservative in adopting it
  I would use it with 4get or SearXNG

Just going to give Wiby a shoutout because no-one else has. More for curiosity / surfing than research, but you definitely get unique results.

I have my own SearXNG self hosted Instance but I'm probably going to switch it to 4get today. I've just noticed it getting worse and worse over time. slow, bad results, or just randomly starts 504ing when everything else (forgejo, vaultwarden, navidrome, jellyfin, Akkoma, etc) on my server continues to work fight.
Tried some 4get instances and it's just so damn snappy and quick.
- I noticed the same bad search issue a few weeks ago. I tried several other instances from searx.space and they all gave bad results. I reverted to DDG. I'll have to give 4get a shot.
  
  I don't know if your complain about serxng is something you'll eventually get with 4get. Whatever works at any time. See google and other search engines had been playing awful to searxng. At some point google will notice 4get instances and will take on them as well. On searx[ng] on the side one can see what has been suspended, denied, timed out, etc.
  Time wise, 4get allows for only 1 engine at a time. searx[ng] allows for several no wonders it can take time. BTW the user can choose any of the engines per search they want. That of course can take more time. The point is that right now all search engines have taken it on searx[ng], and options are welcome. Self hosting is working better maybe because the search engines don't feel threaten by it.
  The good thing is to have alternatives.
  BTW, I thought google found the way to finally get rid of invidio, but it seems there are a couple of instances still working. Good !
  
  I did end up switching from searxng to 4get on my server yesterday as I previously stated I would. It's much better. way faster.
  if you want you can try it out on my instance https://4get.andmc.ca/

I switched to using Ecosia a while back, and have had no problems with it. The results are generally relevant enough for everyday use, and it feels good knowing that my searches contribute (at least in some small way) to reforestation projects.
It’s not perfect, of course.. It still relies partly on bing’s index, but the experience has been stable and consistent for me. I also like that the interface is clean and privacy-focused without trying to upsell the search experience.
In the past I’ve tried alternatives like StartPage and DuckDuckGo, but Ecosia has quietly become my default. It just works well enough without much fuss, and that’s something I really appreciate right now.
- Rooting for ecosia with it plans to create their own index, but unhappy with the direction they're gping specially with AI
  
  I feel the same way. I’d love to see them move toward developing their own independent index. I also really hope they stay true to what makes them different and don’t get caught up in the whole “AI-everything” trend. Search doesn’t need to be artificially padded or reworded by a chatbot, it just needs to be genuinely useful, transparent, and connected to reality. If Ecosia focused entirely on building a clean, human-centered search experience powered by their own index, without the AI noise, I think that would be far more valuable than following the same path other major engines have taken....

I don't think Qwant and Ecosia's index has fully rolled out. They have been using it as a partial index in France and Germany starting in summer 2025 according to their FAQs on their website. So I guess it would depend where you live.

Here is a list of alternative (meta-)search engines:
Swisscow
Qwant
Ecosia
DuckDuckGo
SearXNG
MetaGer
Leta by Mullvad
Mojeek
Kagi (Paid I think)
Das all I can remember rn.
- SearxNG has been bad lately, so just put priority words first to get around it.
  
  Hwat exactly do you mean by SearXNG being bad? Cause in some sens that isn't even a thing. It's a meta search engine.

I selfhost searxng. It's good enough even if it's just a meta-search (but at least removes AI generated answers).
I've looked at yacy (https://yacy.net/) some time ago. Nice concept but I had technical limitations at my side to properly peer with others. Maybe I'll try again as I resolved them.
- How's the resource usage of SearX?
  
  Just checked, 138mb. It's a podman container.

I’ve tried all alternatives to google and I like Qwant the most so far

I just want to say I disagree about the decline, I think there was a time there was huge decline indeed, first there were too much shitty autogenerated and bait content getting high in the index, then they shoved more and more sponsored results and early filter bias were shitty, but I believe it all improved greatly in later years, specially with AI search, Google indeed manages to give you most relevant stuff on top now, and most of times it will cover all you are asking, I find the difference from other engines noticeable, the problem is that Google is fucking cancer...
I mostly use Starpage, sometimes results aren't too good and I go to DuckDuckGo. I sometimes use Leta and Brave Search as well, but there are times I need to Google, especially for shopping.

65 comments