Frustratingly, doing a search on that Changelog page for sources is mostly full of stuff not relevant to my search.
I did find a different post on Lemmy that talks about it, though. This post is incredibly thorough, and does an excellent job of undoing Kagi’s attempt to memory-hole the information about which sources they use.
This makes it all the more frustrating that Vlad refuses to re-add them, instead asking to know why we would care. Here’s a link to that conversation, which is on a platform controlled by Vlad, which appears to be resistant to archiving services that attempt to fetch those particular comments. Also for posterity:
slamor
Oct 27, 2024
https://help.kagi.com/kagi/search-details/search-sources.htmlThere is really no proper information about search sources. We need to know what resources are used and at what rate.
Please make a more detailed and clear edit.
Vlad
Oct 29, 2024
[@]slamor Is there any particular reason you are asking for this? More context will help us better understand the need.slamor
Nov 2, 2024
[@]Vlad why not?
Searching through kagi.com for “Yandex” yields a lot of dead links. The one living link is the Changelog, which says they added Yandex to their image search, back in December 2024… But that’s hardly a revelation. The changelog doesn’t go back very far either, AFAIK
As for the other links: Google says these links used to contain it the word, but I don’t know why. Maybe this one was for raised sites, maybe it was for lowered sites, which would at least give a little insight into whether users loved or hated the domain…
url: https://europe-west2.kagi.com/stats?sd=asc&st=percentage
text: yandex.com. zlibrary.to. androidcentral.com. answer-all.com. baijiahao.baidu.com. cbc.ca. developer.apple.com. eightify.app. github.getafreenode.com. gitmemory …
Another result seems to suggest Yandex Images served up a photo of Steve Jobs in a demo search, but that is no longer the case. Maybe it’s just a coincidence.
url: https://kagi.com/images?q=steve+jobs
text: 564 x 318 yandex.ru. 20 Steve Jobs Quotes: Wisdom from the Apple Co-Founder 20 Steve Jobs Quotes: Wisdom from the Apple Co-Founder. 696 x 418 cioviews.com. 75 …
What’s the issue with this? API calls to other search engines are anonymised, and naturally I want Kagi to search as many sources as possible.
So what’s the issue?
You don’t need Wayback machine to find their old page, it’s all open on Github: https://github.com/kagisearch/kagi-docs/commit/6baff1c066db9b3d804653ea19bc9d1c076a710b
I don’t see any “conspiracy” removing an itemised list - it’s just something which is a pain to keep up-to-date over time. Better to say “all major search engines worldwide”, which is what they’re now saying.
Again, I don’t really see any issue here…
I truly don’t understand the hype around Kagi. Maybe I’m just too wrapped-up in my own little FOSS and privacy world, but paying for a closed-source solution that ties you into an account just doesn’t make sense to me.
For basically as much money as an unlimited Kagi plan, you can rent a small VPS to host Searxng and a VPN service to run it behind and you’d have a much more privacy-respecting search solution, plus a VPS to play with. Kagi just seems like such poor value for your money. Granted, I’ve never used it, but I can’t imagine its results could realistically be good enough to justify paying 10$/month.
Granted, I’ve never used it
This is probably the reason you don’t get the value.
I do rent multiple VPS’s and I have run Searxng, and for me Kagi is worth paying for. It’s certainly not worth it for everybody though.
Regarding the account, I used an anonymous email and paid with an anonymous method. Perhaps they could assemble my identity based on my searches, but considering the new support for Privacy Pass it appears they are walking their talk.
Care to share what makes it worth it for you? I’m honestly curious.
- The results, first and foremost. When I’m getting better results it saves me time and it’s worth it to me.
- Site ranking up/down/block
- Removal of AI generated images from image searches
- Custom “lenses” mean that you can create subsets of the internet to search on
- The filtering options
- Modification of result URLS. For example I rewrite all Reddit links to old.reddit.com -
^https://(www.|)reddit.com|https://old.reddit.com - AI model integration (this one is definitely going to be polarising…). I’m already paying for Kagi, and this lets me query a stack of different models at no extra cost.
- Custom CSS (a minor thing, and sure I could use a browser extension, but this way it’s consistent on all my devices with no extra effort)
- Compared to Searxng, other than the results being better, it’s the ease of configuration. I can’t be bothered trying to tweak Searxng through YAML when I could just get exactly what I need OOTB.
There’s probably more that I’ve simply gotten used to over time, but that’s a few off the top of my head.
Thanks for the overview! I appreciate it, and can definitely see how these features could be worth the investment depending on your use case. I guess I’ve mostly seen Kagi pushed as a privacy-respecting option, which doesn’t appeal to me considering the alternatives such as Searxng.
Feature-wise, I can definitely see why some people would prefer Kagi.
I’m a Kagi user, and basically all of this.
The experience is far better than any free search engine for me, and the “summarize search” feature seems to be factual and complete, compared to Google’s Gemini feature that gives you 5 words that then turn out to be wrong or dangerous.
I thought this was a good presentation of concerns surrounding Kagi, a search engine I’ve heard championed as a great Google replacement, one that I’m definitely interested in using. But I’ve also seen many that critize it for a number of reasons. So I wanted to see if others wanted to share their perspective. Thanks to LWD for laying out what he did.
