I mentioned this in another thread, but I do worry that google is eventually going infect the APIs that metasearch engines like DDG, Kagi, searchxng, etc depend on.
In my experience, a lot of the sysadmins who run high traffic sites will treat all bots as scrapers that have to be blocked or slowed to a crawl. Then they make special allowances for googlebot, bing/msnbot, and a few others. That means there is a massive uphill climb (beyond the technical one) to making a new search engine from scratch. With Google and MS both betting the farm on LLMs I fear we’re going to lose access to two of the most valuable web reverse indexes out there.
I fear that as well. I use Searx-NG at home, so am expecting that to start dying a death of a thousand cuts soon.
Was thinking about standing up (or contributing to) either YaCY or Stract, but you made a good point about the bot allowances for the Googlebot et al crawler UAs. Wonder how frowned upon it would be to spoof the crawler UA in a self-hosted one?
I’ve just started using Searxng… you expect it to die soon? Is it because you expect other search engines to follow suit until there are no search engines anymore, only hallucination machines?
I hope there’s enough of a market for non-ai content that it doesn’t come to that. I think we already reached the pushback stage with image generation.
When I was running a site, I had special rules in my firewall to look for things that said they were googlebot but which didn’t come from one of googles published public IPs.
I had that thought after I replied when I realized that most of the reputable search crawlers will publish the IPs/ranges they use in addition to the UA. The disreputable ones (cough Bytedance cough Xiaomi cough) will just spoof Chrome on Windows 10 and flood you with requests from AWS datacenters in Shanghai or Singapore.
That said, I may still continue looking into working with one of the actual self-hosted search engines (vs meta search) and see how well that works.
Same. I don’t mind it as an option if that’s what some people want, but stop “enhancing” the default experience with it and shoving it down my throat. No lo quiero.
Oh, it will impact a lot more than one private search engine. Watch The AI Dilemma presentation given by Asa Razkin and Tristan Harris last year if you want an idea of what could be coming.
Is coming, and more.
Very good video, with good points. Slightly outdated already, with AutGPT being a thing.
What’s coming, is going to be orders of magnitude more than what they predicted in that video.
Slightly outdated already, with AutGPT being a thing.
That’s the really disturbing thing and what makes this challenge so different to all others humanity has faced to date. I think Asa even referenced in the presentation that some of his slides were going out of date on a daily basis, that’s how fast the technology was moving.
They also correctly identify the function of LLMs as a glue between siloed AIs. We have barely seen the beginning of that, but as the AI race continues, it seems likely that some LLM models will be created that will have less human language, and more “interop language”. Where nowadays LLMs can be somewhat probed for words and relationships, we’ll have zero chance to probe an LLM using tokens that are part of some made up (by the AIs) interop language. Black boxes inside black boxes.
A naive approach will be to “democratize AI”, and that will surely be better than centralized AIs responding to every query… but won’t solve the deepening of inscrutability.
One point made me chuckle: when they showed the graphs for cualitative jumps above certain network size. Recently someone commented about the “diminishing returns” and “asymptotic growth” of making a LLM larger and larger… but also so it was in these models: diminishing returns all the way up to a point… followed by a sudden exponential jump until the next asymptote. The truth is we don’t know where the asymptotes and exponential jumps lie, we don’t even have a remote hypothesis about it.
Sigh. AI has basically added a rocket booster to the enshittification train.
Hopefully this doesn’t impact DDG.
I mentioned this in another thread, but I do worry that google is eventually going infect the APIs that metasearch engines like DDG, Kagi, searchxng, etc depend on.
In my experience, a lot of the sysadmins who run high traffic sites will treat all bots as scrapers that have to be blocked or slowed to a crawl. Then they make special allowances for googlebot, bing/msnbot, and a few others. That means there is a massive uphill climb (beyond the technical one) to making a new search engine from scratch. With Google and MS both betting the farm on LLMs I fear we’re going to lose access to two of the most valuable web reverse indexes out there.
I fear that as well. I use Searx-NG at home, so am expecting that to start dying a death of a thousand cuts soon.
Was thinking about standing up (or contributing to) either YaCY or Stract, but you made a good point about the bot allowances for the Googlebot et al crawler UAs. Wonder how frowned upon it would be to spoof the crawler UA in a self-hosted one?
I’ve just started using Searxng… you expect it to die soon? Is it because you expect other search engines to follow suit until there are no search engines anymore, only hallucination machines?
Basically, yeah.
I hope there’s enough of a market for non-ai content that it doesn’t come to that. I think we already reached the pushback stage with image generation.
When I was running a site, I had special rules in my firewall to look for things that said they were googlebot but which didn’t come from one of googles published public IPs.
…yeah.
I had that thought after I replied when I realized that most of the reputable search crawlers will publish the IPs/ranges they use in addition to the UA. The disreputable ones (cough Bytedance cough Xiaomi cough) will just spoof Chrome on Windows 10 and flood you with requests from AWS datacenters in Shanghai or Singapore.
That said, I may still continue looking into working with one of the actual self-hosted search engines (vs meta search) and see how well that works.
Hopefully it doesn’t impact them too much. So tired of this AI train man.
Same. I don’t mind it as an option if that’s what some people want, but stop “enhancing” the default experience with it and shoving it down my throat. No lo quiero.
Oh, it will impact a lot more than one private search engine. Watch The AI Dilemma presentation given by Asa Razkin and Tristan Harris last year if you want an idea of what could be coming.
Is coming, and more. Very good video, with good points. Slightly outdated already, with AutGPT being a thing. What’s coming, is going to be orders of magnitude more than what they predicted in that video.
That’s the really disturbing thing and what makes this challenge so different to all others humanity has faced to date. I think Asa even referenced in the presentation that some of his slides were going out of date on a daily basis, that’s how fast the technology was moving.
Yup.
They also correctly identify the function of LLMs as a glue between siloed AIs. We have barely seen the beginning of that, but as the AI race continues, it seems likely that some LLM models will be created that will have less human language, and more “interop language”. Where nowadays LLMs can be somewhat probed for words and relationships, we’ll have zero chance to probe an LLM using tokens that are part of some made up (by the AIs) interop language. Black boxes inside black boxes.
A naive approach will be to “democratize AI”, and that will surely be better than centralized AIs responding to every query… but won’t solve the deepening of inscrutability.
One point made me chuckle: when they showed the graphs for cualitative jumps above certain network size. Recently someone commented about the “diminishing returns” and “asymptotic growth” of making a LLM larger and larger… but also so it was in these models: diminishing returns all the way up to a point… followed by a sudden exponential jump until the next asymptote. The truth is we don’t know where the asymptotes and exponential jumps lie, we don’t even have a remote hypothesis about it.