@mauve Been using DuckDuckGo Lite: https://lite.duckduckgo.com/lite/

@mauve …Is ChatGPT an answer? It delivers pretty excellent results for 80% of the general questions I ask it that I’d otherwise search for?

@hank I don't really want answers TBH. I also worry that LLM based search is still not trustworthy enough of a source. I really just want what Google had like 10 years ago. 😅

@mauve My current strategy is to verify what it says by searching anything that seems BS through Google. I’ve caught it making up stuff a bunch but with a little manual verification it’s still a good tool?

I want 10 years ago Google too. :(

@hank Jeeze. :P The internet is already full of lies, so having potential lies that sound authoritative as your primary interface seems kinda exhausting.

Maybe I'll start looking stuff up in literal encyclopedias and reference manuals again or something.

@mauve In fairness, a lot of what I'm asking it is "find me a script to do x" and then once it gives me the name of a tool I want I go to the docs page? That or "suggest 5 names for xyz feature" as you might to a colleague to bounce ideas off of. Sometimes it comes up with good thoguhts, sometimes it lies, but generally more good thoughts than nonsense?

Meanwhile Google is ALSO just a bunch of AI generated garbage articles that offer 10 ways to do a task in [sponsorname] software.

@mauve It's less of a primary interface and more of a starting point to learn things I don't know anything about. Most of the time when learning anything you'll realize that things you initially thought to be true were more complex or wrong anyways. Same deal here, I can just get to that point faster?

@hank Interesting. I mostly use search to find others that have had the same stack traces as me and every now and then Linux CLI stuff (though that hasn't been _too_ bad if I know the command name already).

Lately I've been using duckduckgo to find the repo URL for some dependency and then I either clone and grep or use whatever keyword search is on the repo site.

@mauve That sounds like a potentially better method for what you do (no need to filter through the LLM as an abstraction model). A lot of the programming things I use it for are explaining how things work to me. It's been quite valuable from that perspective, probably more effort than reward for you though. :P

@mauve I'm happy with Brave Search mostly. There is also https://search.marginalia.nu/

@makeworld Oh wow. Search that focuses on non-commercial sites sounds great. I'll try giving it a go. 🙇

@makeworld Is there a straightforward way to get some of my sites indexed by this thing? Seems that stuff like the Agregore website isn't being loaded https://agregore.mauve.moe

@mauve see https://memex.marginalia.nu/projects/edge/for-webmasters.gmi

It doesn't look like there's an easy way to add your site, but see if you can't find it using a "site:" directive and worst case you can reach out to the dev.

@mauve okay using "site:" search I was able to request crawling of your agregore and ranger subdomain, turns out there is a button for that. It says it might take up to a month for the queue to get to it.

Strangely, it seems to recognize but also dislike your software subdomain. Not sure what's up. Maybe it triggers some filter.

https://search.marginalia.nu/search?query=site%3Asoftware.mauve.moe&profile=default&js=default