It’s a story from April, I tried seeing if it was posted before, however it’s the story about how the advertising and finance team beat the search team into submission by ousting the core person protecting it to pursue “growth” and “revenue” at all costs.
Read this a week or so ago and it’s a fantastic summarization of the core problem. I almost never use Google search anymore. I go between DuckDuckGo and Perplexity.
I’ve been using Perplexity for 2-3 months now. On 2 occasions, it flat out lied to me. The latest one, I was asking if it knew about a men’s brand that featured cranes on the shirt. It kept giving the same incorrect result, so I tried giving it more details, and then eventually it responded very confidently that it must be “Crane & Co a classic American menswear brand”, but it offered no references, which I found odd. Then I asked if it just made that up, and it confessed to fabricating a lie…
That’s odd. Were you using Pro? I find pro to be less reliable, oddly enough. I always check the citations which is why I like their approach.
Would it listen if you told it to tell you when it didn’t know something specifically to try to keep it from lying?
I’ve tried to use DDG for a few weeks now, and I find it gets worse the more specific I search.
For general things, it’s fine. But if I’m looking for an installation tutorial for a certain kind of plumbing hardware, it struggles to show me anything but brand and retail pages.
I started using Kagi. Yeah, I have to pay for it. But we’re paying for Google too, kagi is just more honest about how.
Can I use kagi as a default search for my browser?
I’m trying to use Firefox on desktop and mobile more.
Yep!
I use it on FF mobile and desktop all the time, for the last 6 months or so. It is great.
You could try a SearXNG someone hosted, it’s FOSS and not worse than Google :)
I started finding DDG’s results just as bad as Google’s, so I switched to SearXNG and have been pretty happy with it so far.
Its open source so anyone can run an instance if they wish. I feel like this sort of model is much more resistant to enshitification.
I run a SearXNG instance myself and while it is a fine aggregator, it’s important to note what it is and isn’t. For instance, Sear does not have a dedicated search index and leans on third party API calls (to indexes such as the aforementioned Google and DDG listings.) This is my understanding, feel free to correct it.
For my money, I like the anonymity that Sear can afford and that it hides the AI bullshit pouring into the UIs. My son and I were talking over the weekend about how unreliable he is finding the move to AI search.
Edit: A list of public SearXNG instances for anyone that doesn’t want to spin up their own.
Yes you’re absolutely right. The problem of aggregators is that if all the aggregated searches go to shit, then so does it. Garbage in, garbage out.
SearXNG is just a meta search engine. It uses Google and DDG under the hood, among others. How is it possible that it’s better?
Because the aggregated weighted result ranking provides a more useful page rank than any individual search engine, and if any search engine tries to (accidentally or otherwise) stuff specific results into the top ranks, it doesn’t matter. It’ll be deranked because no other engine displays those results highly. In a similar manner, it deranks targeted SEO attempts unless multiple platforms are targeted.
Don’t get me wrong, it still has its problems. For example, if the individual search engines all get a bit too samey, then it will as well.
Check out You.com too, their search is far better than Google on a lot of the technical topics I’ve searched for.
Edit: ugh, it was good. Now they’re shovelling LLM interaction front and center and the original search functionality is completely buried. If you can find their original search interface it might still be worth a look.