It is becoming near impossible to find relevant information from search engines. Duckduckgo, SearXNG, Bing, Google, and so many more mainstream engines have a significantly high noise to signal ratio, and it is getting worse.
Here are a collection of the best search engines I know, please add more to the list.
- Forum Search Engine: https://crowdview.ai/
- Non-commercial Search: https://search.marginalia.nu/
- Libre Meta Search Engine: https://librey.devol.it/
- Golden Age Search Engine: https://www.wiby.org/
- Yandex: https://yandex.com/
If no more high quality search engines exist, would it be possible to host your own?
EDIT: Some new discoveries. The addon uBlacklist and filters can block super SEO sites from appearing in search.
You must log in or # to comment.
The best search I found is by asking questions to real people in forums such as this one. It’s way slower than getting an immediate answer for a question, but the signal to noise ratio is higher.
This is the reason why I think Google is prioritizing reddit so much in recent years, because reddit became one of the only places where you can get real people (well, relatively speaking more than most other parts of the Internet) answering all kind of questions.
I found that no general purpose search engine will ever serve my needs. Their goal is to index the entire internet (or a very large subset of it), and sadly, a very large part of the internet is garbage I have no desire to see. So I simply stopped using search engines. I have a carefully curated, topical list of links from where I can look up information from, RSS feeds, and those pretty much cover all what I used search for.
Lately, I have been experimenting with YaCy, and fed it my list of links to index. Effectively, I now have a personal search engine. If I come across anything interesting via my RSS feeds, or via the Fediverse, I plug it into YaCy, and now its part of my search library. There’s no junk, no ads, no AI, no spam, and the search result quality is stellar. The downside is, of course, that I have to self-host YaCy, and maintain a good quality index. It takes a lot of effort to start, but once there’s a good index, it works great. So far, I found the effort/benefit ratio to be very much worth it.
I still have a SearxNG instance (which also searches my YaCy instance too, with higher weight than other sources) to fall back to if I need to, but I didn’t need to do that in the past two months, and only two times in the past six.
Do yourself a favor and try kagi.com.
I switched to ChatGPT and find it superior to the mess Google and others have made of their search engines. I could never go back to a regular search after using AI.
You know half the shit chatgpt says isn’t true, right?
It’s okay for things that are pretty low-stakes. If you ask for cooking or cleaning advice and it hallucinates you’re still at square zero regardless.
I have not found that to be the case at all. While not perfect, it is miles above Google Search and has not more errors than the misinformation any search will yield. It is a significant business advantage as well and those who are not embracing are missing out.
I think businesses should be at a disadvantage of all things. Business caused millions of people to starve to death in Bengal.
How so? What size of business? I am a business of just 1 person.
I’ve found bing ai is quite good if you ask for the source after anything it spits out.
These models can invent a source. Their only incentive is to have a convincing conversation with you. They are unconcerned with the truth.
What I mean is I use it to get the links to those sources. Like when you use Wikipedia as a jumping off point. I don’t think we’re at the point yet where we have the problem Wikipedia sometimes has that the sources used sometimes themselves just cite Wikipedia.
The links to Wikipedia are actual citations to real sources. LLMs basically just generate something that looks like the link to a credible source which might support what it’s said. It doesn’t care if its “source” actually supports what it says.
The links to Wikipedia are actual citations to real sources
I read an interesting article a few years ago about the Wikipedia source problem. It did a dive into how sources that seem legitimate on Wikipedia can and up citing sources that are less so. They were able to trace back the citations to Wikipedia itself. So no, they’re not always real sources.
LLMs basically just generate something that looks like the link to a credible source which might support what it’s said. It doesn’t care if its “source” actually supports what it says.
Which is why you read the page it has linked for you as a source. Unless you’re trying to say it full on generates a page for you.
Isn’t Yandex Russian?
I agree with the person who identified that ChatGPT is better than a search-engine, but you have to check it, because, unlike a normal search-engine, the ai-engine, itself sometimes produces disinformaiton, instead of only linking-to disinformation.
Checking is now required in both cases.
https://en.m.wikipedia.org/wiki/List_of_search_engines#Metasearch_engines
for some alternatives, btw.
_ /\ _
no mention of kagi yet? Son, I am disappoint.
They are high thinking people will pay $5/mo for search AND being limited to 300 searches/mo. I avoid subscriptions at all cost, so if I were ever to consider paying for search it would need to be a completely forgettable number like .99/mo.
