Most of this AI stuff is trash.
I think Google AI has maybe once given me a useful answer.
Amazon has this thing called Rufus that just slows down the process of searching customer reviews. Just like Google, it's maybe once or twice given me useful information and none of it worth the wait that it takes for the search results to come up.
But we are pouring billions into it and increasing our data center power usage by 10x because It's The Future ..
The problem is AI will always confidently give you an answer like its word is fact. Obviously since you're the one searching, you're most likely unable to verify its accuracy so it's impossible to trust.
I dont mind AI being wrong when i can immediately tell when its wrong but in searches I'll never trust it
I can't say I'm liking DDG. The first page gives me the same domain links. Like when I search for "eye washing gifs", the first three links is the same website. Where I'd expect it to give me three separate websites (and combine searches).
Granted I’m jaded as fuck from only a decade and a half as a software engineer, but from where I’m standing, it’s pretty much all just c-suite circlejerking. Competent, incisive, effective, and ethical c-suite leadership that doesn’t focus primarily on finance matters is vanishingly rare these days.
For the same reason it became more popular than other search engines in 00s. Those would give honest search results, Google would have various kinds of complex foolery approaching ML to give people what they wanted quicker.
One can say their corporate culture shows signs of overfitting for that situation. And not just theirs. In general those attempts to make products more competitive with even more complex foolery outside of the main functionality - are that.
Except when products are simply less usable for said main functionality, people use them less even if they don't consciously realize that.
Also - what is Google in essence? It's saying that some computing thing is too smart for you to run it at home or self-host it. It can only be done by the very smart and important people in companies with trillions in capitalization. And because you can't, you are by some cultural taboo forbidden, to run it at home or self-host it, they get to manipulate results to make you give money to the people partnering with them.
We all know there's nothing fundamentally or practically impossible in making a search engine. If we don't have to cache pages, it's actually easy.
The issue is in the service requirements. What the Internet needs is a technically transparent p2p market of services. Where storage and computing power can be transparently donated (or sold) just like in some countries you can sell power to the electric grid.
OK, I've described the magic wand. That's the strategy. Tactics is for someone actually capable of conceiving the thing. LOL
When you go to search something and you literally have to type site:reddit.com at the end of it to get a human response it's no wonder it's falling apart
And reddit is locking shit up as we speak. These corpos about stole 30 years of the internet from us. There is not much recourse either they control pretty much all of the IP and infrastructure. So their way or highway.
Well there is fediverse for social media at least but it will long time to scale. Other sub sectors will probably need to explore similar models if people want to get away from being gang by 2-3 oligopolies in every market/service they need.
The search engine magic isn't just about caching pages. It's also extremely expensive / complex to:
maintain an index of all the websites in the world. This is an extremely high cost
refresh that index in almost real time. How long will your self hosted crawler take to find new content for every website in the world?
there's also the algorithm for weighing results. The order of results and their relevance is not easy at all. How many times a word appears on a page is a terrible metric.
A hosted search is also a lot more environmentally friendly - that gigantic search index and all the energy poured into the work is something that can be shared by everyone. If everyone did that themselves at home, you spend that same amount of energy for every single household.
"Almost real time" is not what I'd call necessary.
Weighting results - I'd expect user feedback (good result, bad result, combined with keywords from the request) would be good enough. Similar to ed2k for files' reputation.
The index is going to be big, yes. But if we want a p2p system with split storage and computation, something between Freenet and Ceph, may be doable.
A hosted search is also a lot more environmentally friendly - that gigantic search index and all the energy poured into the work is something that can be shared by everyone. If everyone did that themselves at home, you spend that same amount of energy for every single household.
With some kind of such a p2p system I can imagine the overhead to be like 10 or maybe 100 times Google. But not what you said.
I've been experiencing this, but only when I'm not logged in to my Google account. It seems that they're forcing people to log in so that the search results are less shitty.
Can anyone confirm this? Sounds like something exactly they would do for better data farming. All corps over last few years are offensive with browser tracking bullshit.
At this rate will demand a camera in your toilet to authenticate you every time you shit to ensure proper ad sense "fingerprint"
I miss old mrwhosetheboss who was more technically knowledgeable. He was talking about android modifications like roms that he probably would never talk about today.
For the time being at least, there are options to still use the cleaner search results. For Firefox, this post walks through how to set it up: https://lemm.ee/post/42488635/14853159