sp3ctr4l

They just want to know which accounts are permitted to scrape them for textual LLM training material (because Reddit Inc has a contract with them) vs which ones are not permitted.

The byproduct of this is intensifying a cult like environment where its easier to gaslight people, but that’s ancillary.

Even being able to better drown the user base in more, and more specifically targetted ads, even that ain’t the main thing. Everybody in the online advertising world knows that the actual metrics that ‘describe’ how effective ads are… are largely bullshit, and theyre only going to become more bullshit as more LLM bots are spun up and the average person just has less money.

Its enshittification 101, when user base growth phase stops/plateaus, lock in the user base, squeeze harder and harder untill there just isn’t any money left.

The main thing is maximize profit, ie wealth extraction.

How to read this:

The company Roblox decided to copy the entire database of a set of 3, mostly adult nsfw 3d model hosting websites.

You know, Roblox, that game all the kids play.

Yeah basically if you’ve ever seen 3D animated porn that wasn’t made by AI, and involved a reasonably well known video game character?

Yeah the … video maker nearly certainly got the model from either SFMLab or SmutBase.

The funnier part is this:

A huge majority of these NSFW models are actually variations of models made by … companies, that sell their models and character customization software, and have liscenses that generally permit use in something like a video clip, but almost never in an actual commercial video game, not without a subscription and agreement to liscense terms.

So basically DAZ and Realusion and such now very likely have a lawsuit against Roblox they could choose to pursue… unless Roblox wants to maybe very rapidly pay the liscensing fees, knock out a contract agreement real quick.

… clearly you have not been to SmutBase before, lol.

In other news,

SHOCKING! Economically critical demographic of CEOs FAIL reverse turing test! You’ll never guess who…

‘She is incredibly blunt with random questions’

Behold, autists attempting to mate, maybe.

I can run Bonsai, basically a distilled version of Qwen 3.5 (Qwen being the other major family of open weight Chinese developed LLMs) on my Steam Deck.

It ain’t fast, but its kind of nuts that it can run at all on hardware that is that underpowered, by typical LLM hardware standards.

You just run the thing in a sandbox of some kind, do not give it access to the outside internet or the rest of your computer, and the worst thing that is likely to happen is … you configure it slightly wrong, it uses too much RAM, and your system crashes. And your system can then reboot again, just fine, tone it down a bit to use less RAM.

For a large organization with networked security concerns and consistent uptime requirements, it is simply insane to just plug an LLM directly in to that, without architecting an entire system built around ‘assume the LLM will fuck up/go insane’; sanboxes within sandboxes with highly controlled connections to other sandboxes.

But that would be a significant feat of actual software engineering to construct that for/around an actually complex network, and would take time, so fuck that, we want line go up is what most companies decided to do instead, just fucking raw dog it.

But yeah the main ‘problem’ with LLMs in this sense is idiot CEOs just demanding the magic machine god be plugged right in to their very fragile, barely sensible networks and workflows, where the LLM is then going to break things in roughly the same way as would the nepo hire intern who was just directly promoted to sysadmin…

…not that an LLM has a secret activation phrase that will cause it to shut down or go berserk in a specific way.

Yeah, turns out the executive class, the would-be nobility, are incompetent idiots by their own standards, and are primarily driven by their own malignant sociopathic narcissism, not any of the stories they tell about themselves.

If you want competent executives, they’d need an actually effective selection pressure that punishes incompetence and rewards competence… but they already destroyed that.

So, either we, or somebody, builds some mechanism or system that does that… or they all preen themselves rigourously while steering the ship directly into an iceberg, with all of us onboard.

They will genuienly not understand why we hate them for killing us all.

Ah woops you beat me to it!

*Tim Apple

Let me know how many LLMs you can run locally without multiple entire suites of python libraries.

Muchless train an LLM.

Yeah, load bearing ‘wrappers’.

You know the majority of LLM stacks are in python, right?

We already live in that world, basically.

Python LLM training produces models that are then refined into things that 3D engines plug into to do AI upscaling and framegen, and then different python trained LLM models use more python to interface with MCPs to write more slop extensions and frameworks in/for game engines, and other programs.

Still different python derived LLMs do image/facial/3d object recognition, get plugged into moar python, and then produce models… often for game engines.

Its nearly literally an ouroboros.