This aims to be the home for certifying pure organic/human grade/human coded software, without AI vibe coded garbage included.
Apr 6, 2025 · 642 words · 4 minute read
This copy of blog.mauri870.com/posts/the-low-background-web will disappear as soon as this github PR is merged which will happen as soon as the missing https://blog.mauri870.com/css/style.css is fixed so chrome user using dark mode can actually read the content. Ps: I didn’t know and there was no way to contact the author to report that issue and ask if it was ok to copy, contacted since then and waiting to merge.
If you’ve ever read about nuclear physics or worked with sensitive instrumentation, you might’ve stumbled upon the term low-background steel. According to Wikipedia, it can be summarized as:
Any steel produced prior to the detonation of the first nuclear bombs in the 1940s and 1950s. Typically sourced from ships and other steel artifacts of this era, it is often used for modern particle detectors because more modern steel is contaminated with traces of nuclear fallout.
Now think of that, but on the context of the web and AI-generated content.
I’m calling it the low-background web: all the content that came before the rise of large-scale generative models. That’s blogs, wikis, forums, documentation, and general-purpose content written by humans for humans, even digitalized content that precedes the internet itself.
Back then, the web was slower, messier, and, let’s be honest, full of junk. But at least it was our original junk. Not content regurgitated by AI.
We’re living in a time where trusting the source of information is becoming increasingly difficult, and it’s only getting worse as the years go by. We’ve learned to spot things — counting fingers in pictures, noticing shading inconsistencies — but the truth is, eventually, it will be impossible to tell what’s real using human senses alone. What is real? How do you define real?
After the advent of large language models (LLMs) around 2018 and more specifically after ChatGPT 3, the web started to get flooded with AI-generated content. Articles, blog posts, images and video — you name it. This is not just a problem of quality, but also of authenticity. It’s becoming increasingly difficult to discern what’s real.
Besides the philosophical implications, there is also the feedback loop problem. LLMs are now trained on content from previous LLMs, which were trained on earlier LLMs… and so on. It’s like photocopying a drawing a dozen times in a row. It still packs information, but at some point, the signal gets lost in the noise.
Imagine searching for a product review for a popular gadget. You find lots of posts, half a dozen are AI-generated with more or less the same opinions, regurgitating the same points. No one comes up with new insights, no one tries to offer a unique perspective, we keep iterating on the same thing over and over. This is a recurring phenomenon across the web, from blog posts to news articles to even creative content. This is what I call the ice aging of information.
Humans are naturally creative and curious, creativity has been the force that drives innovation and improvement since the beginning of our species. When we lose that spark, I fear that innovation and progress tend to stagnate.
The low-background is in the signal, not the noise. Don’t follow the white rabbit.
The tricky part? The boundary keeps shifting. As LLM-generated content gets indexed, reposted, repackaged, and redistributed, it gets harder to know what’s clean and what’s not. There’s no metadata that says “hey, this paragraph was written 100% by a human.”. Even if there was, would you still trust it?
Today’s web is noisy, the low-background content is getting harder and harder to find.
So if you’re browsing a blog, watching a video, or just into the archaeology of internet culture, maybe start thinking in terms of low-background. You’ll know it when you see it.
That is it. The Low-background web. Let’s not lose track of where we came from.
See ya next time, unless I’ve been replaced by an AI. Hmmm, maybe I already have. Who knows?
Knock, knock, Neo.
© Copyright 2025 Mauri de Souza Meneguzzo