Thank you for bringing up Kagi, I had never heard of it before. An intriguing idea for sure and I am not averse to paying for searches, but as a serial Google-fu practitioner $10/month for 1000 searches (1.5c per search after) seems quite steep to me. Some days I swear that would last me 24 hours at most. I need to start tracking that I think.
I do however applaud the seeming transparency on their website. It may or may not be for me, but if they really plan to operate how they lay out on their website, it truly is a breath of fresh air and I wish them luck.
Comparing the pricing on a "search value" basis is compelling, but I'll be damned if I can't find concrete details on what they actually do differently. Just looks like a lot of marketing-speak.
E.g., do they have humans reviewing the results? Or is it just LLM crap? Because if it's more AI bullshit, they can fuck right off to where the Titanic and Titan lay.
Comparing the pricing on a "search value" basis is compelling, but I'll be damned if I can't find concrete details on what they actually do differently. Just looks like a lot of marketing-speak.
Comparing search engines is not easy. I went down the rabbit hole and concluded Kagi is the best right now. A search engine has one primary function. Everything else is icing. That is to find the best result as fast as possible. As their incentive is to find you the best result, not the result which pays Google the most. I find relevant results more often, and much less spam. That's the value proposition.
They also have great features like the ability to block and rank domains, and a kind of filter feature which focuses on certain kinds of information you're looking for.
As for ranking, no search engine uses humans to rank results. There are billions more pages going online every day. It would take the entire world's population working on nothing but that to accomplish. So the magic is in the algorithm and how data sources are combined. As above, Kagi's motivation is to provide good results, and I think they're succeeding. I'm sure Google could provide amazing results to us, but that's not profitable. Of course, humans review the algorithms and test them and improve them, and Kagi appears very good at this.
FYI they have a $5 tier which includes 300 searches a month, and this is enough to cover most users. You might be surprised if you actually track your habits over time. You could start on that tier and upgrade if it's insufficient.
As their incentive is to find you the best result, not the result which pays Google the most.
Google is the reason recipe websites are notoriously bad. They try so hard to keep you on the website long enough for ads to work and for Google to think you like staying at that website and so on.
So that's why we have 20 paragraphs of life stories before each recipe with 10 add shoved in between them, 6 adds on the left and on the right of the page, ads at the top, and so on. You spend so long trying to find the recipe, it's a win for all the ads that you're staying on.
And it's a shame because obviously the smaller blogs don't WANT to do that but they have to in order to have any chance to compete with major conglomerate corporate recipe websites.
I don't know if I'm ready for kagi yet... But it's tempting and definitely something I'm considering.
It is SHOCKINGLY bad to search for recipes now. To the degree I don't even bother or I use ChatGPT now. Google is going to have their lunch eaten by others if this keeps up.
Just seeing this for the first time also, but one of the features they list is it looks like you can "block or boost" domains in your search results. I miss when Google let you block domains for results, so this would be a killer feature for me. I don't mind paying, but just the pricing seems a bit odd. I could see myself paying $5 per month for unlimited searches, not sure if 300 or 1000 is enough, $25 per month for unlimited is crazy.
After using a search engine with this functionality, I quickly realised how many domains I never wanted to see again, and how many I valued. Google has been promoting garbage domains for so many years now because they pay better than the quality sites. The algorithm goes both ways, as in, "show me more," and, "show me less." To make the bad ones disappear, you'd need a notepad next to your Google window with a list of dozens of domains you don't like to append to every query you make. Unfortunately Google doesn't offer you the ability to positively bias domains because they'd earn less. So you'll still see the first few pages of garbage, depending on your search.
I thought I make a ton of searches too (and yeah apparently I do vs normal internet user) but when I tracked my usage it turned out to be around 30-40 searches / day. So, it's not that bad. ATM I still have Ecosia as the default search engine and when it fails to turn up good results I rerun the search with Kagi (which is indeed quite a bit better). So far I seem to use Kagi like 1-3 times a day so it's not that bad.
Thank you for that. Since your comment, I realised I make around the same number of searches as you, so around a 1000 month should be fine. Also realised that not everything has to go through Kagi, I can use Google for super-obvious results, tech issues and such, and Kagi when you need higher-quality results.
remember, when the execs of an evilcorp announce that their users are unhappy with something, it usually means that the execs are unhappy with something (and it's usually their profit margin)
This would low-key be a great way to have your very own sweet sweet human contributed ML-data pipeline, and a pretty high quality community moderated one at that. Give access to 3rd party devs and lock all competing ML techs out. [r word] is valued at $10 billion (fucking lol) but I bet that gutless scumbag sp*z would fold for less than 10% of that if he has more than two braincells to rub together.
Big part of me hopes Google's myopic corporatised ass never sees the opportunity and we rebuild the mf free internet from here on out.
Eh, if you only have two braincells, you aren't capable of narcissism. Spez's problem isn't lack of braincells in general but lacking braincells in empathy, critical thinking and whichever brain part gives rise to theory of mind.
Now imagine I own the encyclopedia, and Walmart offers me money.
So i paste Walmart's Xmas catalogue pages in between the useful information in the encyclopedia. You ask about frog facts, you get frog pajamas. You try to look up cultural information and get travel ticket prices. You never planned on purchasing anything, and you are too poor too anyway. But somehow I and Walmart make money off of your displeasure.
This is ad revenue. This is the modern economy. Its a sham. Its an infinite money go brrrttt machine for billionaires.
I hope the articles and us talking about it so much can get some scientist to do a study about this!
What immediately comes to mind: how long does enshittification take to progress through the stages typically? Is there any company owned platforms who managed to avoid enshittification? What percentage? Is practicing enshittification more profitable and thus pushing out companies who don‘t do it? If yes, how much more profitable?
I need some cold and hard facts cause there is still an amount of denial around even these parts, which perhaps scientific research could help with.
I think Google is headed to breach the trust thermocline (warning: a twitter link). I think why these collapses seem sudden and so large in scale is because there's so much inertia. Services / products that have become the standard can go well below the line that would be accepted otherwise and that's why they don't see big changes in user base while the enshittification process goes on.. So, for them the point where a large portion of the user base is even willing to try alternatives is already way too far.. and no small corrections is going to cut it. They try to find out what they did in the last months to cause this exodus but the reality is that they've been worse than competitors for years.
That tracks so much. The two big I social media paltforms I was involved in were Facebook and Reddit. My distrust in Facebook/Meta is so large, I'm willing to block any fediverse instance that federates with them. And Reddit's only chance to get me back would be to become a trust-managed nonprofit within at most a year (but only if that's how long it would take to implement if they started to go that way within the next few weeks).
That was one of the first things that I thought about. People can’t affix “Reddit” to their Google searches in good faith anymore, so what is the next most reliable community?
Users weren't happy with the search results before the blackout either, and "quite" has no part in it. Google traded quality results for revenue over a decade ago... right about the time they changed their Don't Be Evil motto.
If Google makes changes that stop people from clicking through to reddit due to the protests then the protests will have likely done more lasting damage than anyone imagined.
Even though I think places like Lemmy and Reddit are awesome for hunting down other people's experiences, I wonder if guiding people to places like Codidact or Stack Overflow for actual questions and answers is a better solution for the world. If Codidact joined the Fediverse, that would be ideal. It would just be important to have redundancy in the datasets of each instance. Would suck to lose an entire subject's worth of answers if an instance shutdown.
Super interesting the trick we all thought was a secret, stopped working, and now executives from one of the worlds biggest companies are having trouble as a result lol
This trick used to be common before embedded searches were actually decent. Wikipedia in particular took a long time before their search was even close to "site:wikipedia.org [search term]". I think the saddest part is that people seemed to forget about it for so long until realizing they could use it for Reddit too. So far the trick seems to work for lemmy if you point it at an arbitrary instance like lemmy.world, the content just isn't quite there yet.
I think people forgot how things were before apps and the centralization of social media in so few domains.
People forgot web browsers are the app of all apps.
Search will probably come down to you having a personal AI agent that will broker with other AI agents for relevant data. It will likely become a hell scape.
Add comment