People Twitter

I feel like Google’s crap results predate the AI tech by at least five years. It has been garbage SEO stuff for a while.

This has the bonus side effect of being able to ignore any news that happened since 2023 to gaslight yourself into thinking that we're not all living in a hellscape of a world

And then SEO optimiser add the tag "before:2023"

nowadays, all tabs look like the shopping tab

This beats using "reddit" because that site has also become full of ai generated and bot spam/product astroturfing

Google's slow demise is entirely expected late-stage enshittification.

What is frustrating is that search is mostly a solved problem. Crawling and indexing are solved problems. Fighting adversarial SEO is a continuous task, that Google Search is essentially refusing to perform but is clearly cheap enough for an upstart like Kagi to do reasonably well (their only added-value is the aggregation and filtering of other indexers such as google and mojeek, and let's be honest it's probably 99% google's index powering Kagi).

This shows that the lack of meaningful competition in the space is actually merely a matter of capital. There are too many webpages to scrape, process, and save and nothing short of "indexing almost as much stuff as google" is going to cut it.

In the software world we're used to seeing FOSS alternatives to most things, because software's capital costs are typically almost equal to manpower costs. However for search this doesn't work, just like it historically hasn't worked too well for some really expensive software (such as audiovisual creation tools, with the notable exceptions of Blender and to a lesser extent Krita).

There should be a well-funded non-profit building and providing a high-quality, exhaustive, transparent and open-source indexing service for the world. It definitely sounds possible, and even rather easy in the grand scheme of things. Yet current economic incentives do not favor such models. However I do wonder if there are not options to be explored, such as distributed crawlers or even a distributed index (after looking it up, YaCy seems to be doing just that though at a glance it seems, uh, old and clunky). Or maybe the EU should finally put a real focus on meaningfully funding indigenous FOSS R&D so the enshittification process of American tech giants doesn't crush us as well.

Just done 4 searches and got the exact same results regardless of "before:2023"

I started switching everything to duckduckgo. So far it's been a much better experience.

Works great for old stuff, now how do I look up stuff on the bridge collapse near me last week?

The new episode of the Better Offline podcast is the first in a three part series on the death of the web. I'm halfway through and would recommend it. It's a good show just in general.

The year is 2235, and the warp drive has been invented. I go to search Google for the latest news on the tech but remember this old bit of advice from an old meme that was floating around almost 2 centuries ago I saw once while lurking in the Ancient Memes community. All I find are things taken from fiction. Only about 25% are factually accurate.

This only makes sense for a very limited set of things tho. The things where AI makes a difference are art stuff, news and creative writing stuff. What purpose do news from before 2023 have? Maybe for research purposes it filters out some bullshit but then you also miss out anything relevant that happened after 2022.

Just imagine - everyone who write content that's worth anything these days start dating their newest posts as if they were written in 2023, creating a community within still frame of a better time. It's like Old School RuneScape, but for the internet.

An alternative that actually uses its own web crawlers https://mojeek.com