JESUS CHRIST I can't take it any more

WHY is EVERY fucking search process on the ENTIRE FUCKING WEB grossly and irreparably BROKEN?????

LinkedIn? broken
Google? broken
Glassdoor, Indeed? broken
Every fucking jobsearch site? broken

They all fail the simplest tests (if I restrict a search, does it return fewer results? NO) and lack the most obvious functions (can I use Boolean? can I subtract searches? can I exclude sources or locations? NO, NO, NO).

WHY ARE YOU PEOPLE LIKE THIS?

AAAAAAAAAAA

Sweet baby Jesus, if you run a search, change a filter, run it again -- the URL doesn't just show the filter change, it has a bunch of additional crap like &loc=X where X is their internal code for the location the browser detected.

It takes EFFORT to be this bad at something.

@sennoma nodds in agreement

I'm so furious about the #Enshittification of #JobSearch engines that I literally want to,fund one that lets one properly filter location, tech stack, industries and also allow to filter for direct positions or at least remove the 12+ copy & paste HR agencies that try to sell me the same job offer I already declined...

@sennoma my theory is that they don't really care about you finding what you need, but directing you to what they perceive as benefits their user metrics. It's a combination of AI slop, and a desperate need to bump up the metrics

more searching bc you can't find what you need? more time on the site!

@neurobashing @sennoma that's not just a theory, it's actually what it is!

@neurobashing Yeah, I was thinking along those lines too. Misalignment of incentives strikes again.

@sennoma Because it's not about service. It's about how you can service them.

@sennoma It's not just that. It's getting actively worse.

The history of GMail search and filtering shows that they're actively "dumbing it down", and I'm unsure whether that's a consequence of the "The only UX improvement I can think of is removing options, that's simplification, right?" brain damage or that it means lower hw requirements (indices etc).

@sennoma

I remember when you could restrict a search on google and only get 5 results returned that actually matched. But that wasn't bringing in the ad revenue.

@larsmb @sennoma which is why @tomscott seemed to have gotten almost an aneurism from #GMail only sorting stuff in the UI, not underlying folder structure...

@sennoma I hope monocles does it better...

Cc: @monocles

#monoclesSearch

@kkarhan @sennoma @tomscott I mean, that's the difference between labels and folders, I don't mind that much.

I do mind that GMail can neither search nor filter by specific headers, and just treats Sender/From/Envelope From as identical in searches. And numerous other failings.

I don't understand how a company can fail so hard at search and mail. Even O365 does it better.

@larsmb @sennoma @tomscott there's a reason I've been using @thunderbird for like 20+ years now...

• IT JUST WORKS!

@sennoma
All this AI garbage is just preparation for them not intending to keep the data centers warm.
Sure they'll keep few enough to update the models, but quite soon they'll be turning off the lights in most.

@dnavinci Seriously? I mean, sure, that's a slower fossil fuel burn, but I feel like it will mean even worse "search".

@dnavinci @sennoma Someone's over from Search Engine with a form to fill in, says it's because we think we know something they don't.

@sennoma
MP3s are worse than FLAC. FLAC is "correct" while MP3 loses something. Fake statistic, but 94% of search engine results are for low fidelity answers like, "what's kylee's baby named?!"

@buherator 😻 🙏

@sennoma You are laboring under a misconception. You think that search engines are built to find you stuff. That is not the case. Search engines are built to sell you stuff.

@glennsills Yep, that appears to be the consensus.

I hate capitalism.

@sennoma

FWIW, there's a fast-burgeoning alternative-search movement afoot ... people making, finding, promoting and/or using an ever-growing collection of non-corporate (oftentimes, literally hobby-level) #SearchEngine s.

https://searchengine.party is a decent collection.

Here is another ... https://seirdy.one/posts/2021/03/10/search-engines-with-own-indexes/

A guy on here, @amin is working on a new engine (still beta-ish) here ... https://clew.se

Just yesterday, I also found this one ... https://4get.ca

@sennoma ai reasons broke everything.

@f4grx @sennoma no no no, search was already broken before the AI craze. Fifteen years ago I could find most things on the internet with boolean searches. But not five years ago. Boolean searching has been systematically dismantled because accurate, direct search doesn't sell ads, and all search engines are fundamentally ad platforms.

@raffitz @sennoma now I want to build some web index in python, that would work like the original google.

@raffitz @f4grx
@sennoma

I wondered why I couldn't to Boolean searches anymore. Shows how naive I was.

Just one more data point that the market does not solve problems in general, just problems related to maximizing profit. (And it's simple but dangerous to optimize for only one parameter)

@PaulWermer @raffitz @f4grx @sennoma
> Just one more data point that the market does not solve problems in general, just problems related to maximizing profit. (And it's simple but dangerous to optimize for only one parameter)

Love the phrase, and fully agree on the parenthetical.

@raffitz @f4grx @sennoma

The AI craze had already started five years ago.

Just, the general public didn't believe warnings about what the big tech companies like Google were doing.

If this was about serving ads in search results, that's what the chatbots that replaced Boolean search would be selling. That's a peripheral problem.

The bigger problem with search now is that badly designed LLMs made to sound conversational replaced algorithms that were for actually searching the internet.

@f4grx @sennoma Issue is that the internet scaled a lot. Google still has the resources to index it all in a boolean searchable way, but I don't know how viable it is to make a garage search engine.

@shadowfals @f4grx @sennoma That is a problem now, I don't disagree.

But 5 years ago that was not the user-facing problem (despite the seeds of the AI craze being planted).

And 5 years ago the issue *for me* wasn't the interspersed ads. I could deal with those, at that point the ad blockers still caught those.

The issue for me was no boolean search, no advanced search. And boolean search was not directly replaced by AI, boolean search was just suppressed. Keywords such as insite or inurl or -insite or -inurl just stopped being parsed as keywords by Google. Quoted expressions became a weight modifier that favored the full expression, but didn't remove partial matches. These changes were all implemented deliberately, through deterministic, human-written code. Not opaque machine learning bullshit.
AI bullshit did come along and make things worse, I don't dispute that. But at that point, search was already ruined.

@raffitz @sennoma it probably would not be able to index the whole internet in one garage.

@shadowfals @raffitz @f4grx @sennoma I still blame monetization, not AI. "A""I" I suppose, is yet another symptom of monetization.

@raffitz @f4grx @sennoma This. Iirc it costs Google billions a year to index the web. That massive upfront cost is why only a few search engines exist and no new ones are coming that do the indexing themselves.

I suspect Kagi is trying but in the mean time they’re reliant on using other companies’ indexes.

@raffitz @f4grx @sennoma We can effing try!

@raffitz @shadowfals @f4grx @sennoma sorry for crashing in, but a person who can swear and say "boolean" in the same message makes me weak in the knees.

@raffitz @shadowfals @f4grx You don't know how validating it is to hear this! Every time I try an expression that *should* work and see it fail, it feels like being gaslit!

From this I conclude that, just as the internet itself should be regulated as a utility, like electricity, so search should be a public function.

@raffitz @f4grx @sennoma

Yeah, there was a transition while Google was setting up for the use of neural nets. And it sucked. It's been sucking for years.

@raffitz @f4grx @sennoma Honestly I feel like it would be entirely possible if you found a reliable way to just filter all the SEO mills (LLM-generated or otherwise) from your index entirely, because those seem to be responsible for a very large chunk of that scale

@Brendanjones @raffitz @f4grx @sennoma DIY implementation of search is probably unrealistic. Closest thing I know of is YaCy. About a year ago I managed to do the router tweaks to connect my rig to their network for a few hours. Mainly was shocked at the amount of alt right shit in my search results. Instead of indexing, perhaps DIY community can curate collections of links with webrings, maybe open source versions of "social tagging." Automate things like sorting known links by measures of bloatedness, maybe ratio of payload to length(textContent)?

@sennoma @raffitz @shadowfals net neutrality is the most important regulation out there. We must protect it.

@n8chz fair.

@f4grx @sennoma @raffitz @shadowfals Though net neutrality alone won't make *search* a public utility or an independent nonprofit. That would take a second, separate push.

@pteryx @sennoma @raffitz @shadowfals yes.

@sennoma @raffitz @shadowfals @f4grx

Somebody wants the internet seriously broken. They're doing it on purpose, only explanation I can think of.

@violetmadder Yep, the consensus in replies here is exactly that: a broken internet returns more quick profit in various ways, and that's the only thing the money-perverts care about.

@raffitz @shadowfals @f4grx