This is a most excellent place for technology news and articles.
Hosting shit has become too expensive, and the ads will generate less and less revenue, as it's mostly bots generating clicks and views. This is changing the internet at an incredible pace.
The worst is that any attempt to create a place free of bots will only become a target for LLM companies to gather clean training data.
Hosting has never been so inexpensive as it is today.
There seems to be two groups of people:
People that use services like AWS and pay hundreds (or more) per month: "Wow, hosting is so expensive now!"
People that host a bunch of stuff on a $40/year VPS: "Wow, hosting is so cheap now!"
I've got several VPSes with 40Gbps (shared) networking, 8 or 16GB RAM, decent amount of space on RAIDed enterprise NVMe SSDs, fast AMD EPYC CPUs, that cost less than $60/year. 20 years ago before VPS hosting became popular, you'd need to pay at least $100/month for a dedicated server with a consumer CPU, single IDE hard drive, tiny amount of RAM, on a 100Mb/s connection.
Not to mention being able to host a bunch of stuff at home for cheap with a low-power single-board or mini PC., and Hetzner making dedicated servers very cheap too.
Locally we have been setting up some mesh systems. Its been fun work. One of the best things about it is:
Its in the early phases but its been a sucess so far.
Its in the early phases but its been a sucess so far.
Mesh networks, you say? I've gotta get in on the ground floor of this new advertising opportunity so I can rake in that sweet early adopter cassssh!
Claude! Code me up an ad-injection framework for this meshtastic dongle I just bought, stat! Time to get me a big ol' chunk of advertising real estate!
(sad /s)
Well yeah. Of course it would, given the examples they are talking about.
It makes sense right? You have a computer, you can program it to do things. I have programmed mine to scrape data or examine websites, and even in some cases to pull entire directories down.
Now we have an agent do that. In other words a programmed computer tool that is acting on someones behalf using natural language as the question. Note the example given is an AI is looking across as many websites as it can for a review. Or maybe instructions to put something together. Or maybe a list of the highest peaks in the world with the least snowfall. Who knows, but that is endless.
So this "traffic" is still human driven. Its just using a computer to do the work. I think that changes the perspective a little bit. It is all human intention driven.
Also, all traffic by http plus every other request is still human, so that is a bit misleading.
Interestingly, its not like we havent been here before if you look at traffic. The syncing of usenet for a period of time was the largest user of all traffic. So all computer connections, much, much less humans.
Most traffic today is video, then gaming, than music, financial, screen sharing, p2p, and AI makes up a tiny portion. It just is making a huge number of requests for HTTP, but it is growing, and fast.
This is neither pro AI, nor anti AI. I am just trying to examine the landscape instead of the over simplistic headline that really is about Cloudflares business model.
That's not the point. Let's put aside that an LLM driven agent is not deterministic ie, not always doing what the user intended, just like regular text autocomplete messes up as well regularly. The point is that like with e-mails when spam became so easy to generate that it flooded the system, now more sophisticated actions on the internet get easier to do (that said, most of it was possible already before, just required more skill), at least as long as LLMs are not charged at real costs but highly subsidised.
At some point something has to give. Either tools will be developed to somewhat reliably classify the nature of the traffic or things will start to be priced out as everyone will have to pay the bill for this new kind of spam (ie low value high volume traffic, or ill intended traffic)
A lot of that was scraping. A huge amount. Just take everything possible. Hopefully that will slow down, or as I did, I allow requests and reponses, but scraping gets kicked.
But it is getting hard and harder to self host. Large websites and hosts (gitlab, self labs) get slammed and even DOS. Its been rough.
There are huge network upgrades in the works to try and deal with this 3.2T are being looked at already. But of course electricity is going to hit a hard ceiling both for the data center and the network center. Hence the costs you are talking about.
I am more concerned with the fact that if everyone is asking AI, who is going to put the answers somewhere to get them? Who is going to self host and publish like I do today? It already is rare to see people running their own websites anymore. Lemmy is pretty amazing for that, but it too has gotta be having a huge amount of traffic at each instance.
I guess I am saying I am aware of your concerns and I have mine too.
It feels a bit like in Cyberpunk, where the wild Internet has become unusable or even dangerous and only big corpo controlled parts are still accessible to the masses.
The other topic you mentioned, LLM's are going to hit a wall and they'll do so hard. Training will be increasingly relegated to information predating large scale use if LLMs to avoid rapid degradation of the models by training it with LLM output. That in itself creates a huge problem the more time passes. Nothing is safe btw, analog media are increasingly just as tainted.
You could say anything is human driven if you follow the thread far enough. I'd say if an automated process is cranking out hundreds of requests per minute then even though someone turned it on it still falls under a bot process. There is no human actually monitoring each request or interacting with it.
That's crazy! "The round error" was actually my nickname in high school!
I don't host things on the open internet anymore. I have a small set of services I host for a group of friends and family and I set them up with tailscale.
On the flip side, I've been constantly frustrated because every time I have an idea for some little quality of life tool that involves pulling data from a site I would normally have to browse to directly, I can't do it because everything is captcha'd to all hell.
I think http/https is dead at this point.
Why don't we just generate every video imaginable procedurally and host them in space? Get it over with. Every combination of anything. Because that's what this torrent of AI slop videos feels like.
I mean of course, how many pedestrians are on roads versus cars? Used to be lots of people walking on roads until cars were invented.
I had a realization recently that with driverless cars becoming more common that one day I'm going to be stuck in traffic surrounded by dozens of other cars that are completely empty. They'll just be going to pick up a human driver somewhere or going to park and no one will care because instead of eliminating traffic we'll just ensure humans are entertained while stuck in it.
Why bother to fix traffic congestion if you can just play PlayStation the whole time or whatever?
If most cars are autonomous, and capable of coordinating with each other (via mesh network, for example), then traffic congestion might become much easier to ~~prevent~~ reduce and ~~solve~~ mitigate. (Edit: Based on responses, I may have chosen my words poorly)
You're probably not wrong though. I imagine it'll get worse before it gets better.
Ha ha. Traffic will move to let billionaires through quickly.
Regular folks will be stuck in endless traffic trapped in driverless cars that charge them $5 a minute for the privilege all while playing the same loud and obnoxious audio video ads on repeat for hours on end.
Traffic congestion will always be a thing. The real world is too unpredictable. You can't pack a highway full of cars and expect them to drive highway speeds. One animal jumping on the road and you'll have a massive accident because nobody can avoid anyone
Traffic congestion doesn't happen because of lack of coordination. Coordination prevents collisions. Our success rate could be a lot better but the system does pretty well with coordination limited to blinking lights, a noise that says "fuck off", the mutual knowledge that we all got the same training and the private knowledge that everyone else on the road is a suicidal idiot who slept through the training.
Congestion is a physical property of the road network. There's a physical limit to the number of cars you can push through a road in a given time period. Drive the cars too fast and they fly off the road when it bends, or cars slam into each other when a deer tries to commit suicide.
Traffic is well modeled as an incompressible slightly non-newtownian fluid, like runny oobleck. If you apply more pressure/increase vehicle volume the fluid starts to bind up as it becomes harder for the cars to maneuver without collision, regardless of coordination.
You reduce traffic by taking cars off the road, or scheduling utilization. Self driving cars won't help, and we can do both of those without them. Long term they're likely to make collisions worse because humans don't know what they're going to do, so the primary mode we have of coordination, "the theory of mind" will break down. Right now they're so uncommon that you assume any car is a human who reacts and prioritizes like you.
Tear down the freeways and replace them with trains.
In other words - car usage expands to the limit because human nature is to use the resource until it gets inconvenient. The gas expands to fill the container.
The Metaverse could've been a platform free of AI, offering a protected version of the Internet for people to be themselves. It's kinda where we may be headed anyway. Instead we got this...
Facebook was always destined to be shit.
What we need is Tom to come back and just boot up the only MySpace servers and we can do a hard reset to the part 20 years or so.
Lol, lmao even
Sorry but that is a very misinformed idea of what Metaverse was, and was supposed to be
?
Don't '?' me, what you said is simply false, there is no correction or arguement to be had.
A better metaphor is that the signal to noise ratio is getting so small, that the internet is becoming useless
