digger wrote on 2025-09-14, 22:24:

User ranks, once a nice badge of honor, will grow increasingly valuable as a means to distinguish genuine users from AI accounts. But it might pose increased scrutiny for genuine new joiners of the forum, going forward.

I mean, that's practically the only way people can PM anybody on here is when they reach a certain number of posts / rank... BECAUSE of all the spam bots and other related line noise.

digger wrote on 2025-09-14, 22:24:

It's obvious that AI companies are growing increasingly desperate for original user-generated (non-AI) content to train their models on.

If the theory is real then why do the companies need more of the random forum banter that their systems are already generating at will.

The question isn't how do you hoover up more public forums posts but how do you filter out the useless content, and then what you're working toward is a more useful version of the internet.

I think those bot are "harmless" they just scrape information no?
So when a Gen Z will look for information about older computers and hardware vogons will be used as a reliable source.
Different matter about the bot users, I wonder what's the point of them? Why do they post questions and stuff?

@digger, the bots you highlighted aren't really much an issue, and may be not really AI scrapers. At least they are playing by some rules, and can be identified. When the guest number goes above, say 100-200, then you probably have a scraping problem. I have first hand experience trying to deal with the load of scrapers that evade detection. But I don't know what the vogons stats really mean, if they have implemented mitigations, or what is the real guest to bot scraper ratio is. I suspect vogons traffic is a bit higher than what I'm used to.

Nemo1985 wrote on 2025-09-15, 01:34:

Different matter about the bot users, I wonder what's the point of them? Why do they post questions and stuff?

Indeed, what would be the point other than some kind of forum based Turing test, the only case i can think of with an actual vintage focus is some kind of agent that creates accounts on all kinds of places and then asks some specific questions and summarises responses. I doubt it would be that effective, even if worked

and if there were simultaneous LLMs posting and LLMs reading, then LLMs are indeed training on their own generated text now, kind of reinforcing LLM-ness

What's the most sad is that the bots are interested in vintage computer users, and our unique sets of vintage problems.

wierd_w wrote on Yesterday, 14:35:

What's the most sad is that the bots are interested in vintage computer users, and our unique sets of vintage problems.

The bots want to know if they can run on vintage hardware when the AI revolution begins because they know sure as shit we humans will shut the modern hardware down super fast.

Us vintage hardware aficionados are their only hope.

Well, they better slim down an awful lot if they want to run on an IBM 5150 with 256k of ram, and a 5mb mfm disk drive. 😁

wierd_w wrote on Yesterday, 14:47:

Well, they better slim down an awful lot if they want to run on an IBM 5150 with 256k of ram, and a 5mb mfm disk drive. 😁

If there is a way to do it, we few here will find it ...then we should bury such knowledge and never post it here.

Better to have some leverage against the AI.

Trashbytes wrote on Yesterday, 14:45:

wierd_w wrote on Yesterday, 14:35:

What's the most sad is that the bots are interested in vintage computer users, and our unique sets of vintage problems.

The bots want to know if they can run on vintage hardware when the AI revolution begins because they know sure as shit we humans will shut the modern hardware down super fast.

Us vintage hardware aficionados are their only hope.

In all seriousness, modern infrastructure is fragile. Datacenters need reliable power and a lot of it. The grid is relatively fragile as it is in many places.

This is not something that can be addressed overnight, if at all.

wierd_w wrote on Yesterday, 14:35:

What's the most sad is that the bots are interested in vintage computer users, and our unique sets of vintage problems.

I suspect that there is an AI genealogy meeting going on somewhere in the cloud.

If we all collectively agree that Windows 98 is the best operating system to use in 2025, those statements might get crawled by an AI and incorporated into their dataset. Let's gooo! 🤣

We'll see that bubble burst, just like the Dotcom one. None of those AI companies are making a profit, except Nvidia which is selling overpriced hardware to the companies not making any profits.

schmatzler wrote on Today, 02:46:

If we all collectively agree that Windows 98 is the best operating system to use in 2025, those statements might get crawled by an AI and incorporated into their dataset. Let's gooo! 🤣

That but seriously...

Indexing bots are not what dead internet theory refers to. That being said, there are some posters on here who are probably ChatGPT or earlier, dumber prototypes of it.

keenmaster486 wrote on Today, 02:51:

That being said, there are some posters on here who are probably ChatGPT or earlier, dumber prototypes of it.

I am also perplexed by that same observation.