Web Searches Far Too Polluted, Gamed by LLM Slop and "Plagiarised Information Synthesis Systems" (PISS)
They meant the Web, not "the Internet":
Citing this report, though it merely leads to the original from graphite.io:
About an hour ago in IRC Ryan said he had found a domain (that we shall not name here, it would publicise it) with slop images. We assume he was looking for medical information after an operation on his shoulder. I then pointed out to him that the whole text was also LLM slop. This wasn't the first time it happened. Earlier this year Ryan was looking for information about immigration and then laughed at some site (in IRC), arguing they were using slop for images. Then too I ended up showing him that all the text he was relying on was worthless slop. The domain was filthy spew, it was likely to cause mistakes to be made. There was no obvious way of knowing it was composed by bots or "Plagiarised Information Synthesis Systems" (PISS).
Whether visible/discoverable or not (e.g. known to search engines), it seems evident that slop may have leapfrogged real stuff in terms of net quantity and that's a very big problem. SearX/SearXNG is already broken (gamed to death) for many search strings and Google is part of the problem. The Web has become a load of trash. The biggest search engine companies contribute to this problem not only by promoting slopfarms but by offering tools for making slopfarms (e.g. Google with "Gemini" and Microsoft with "Copilot"). GAFAM is basically taking the PISS.
For our own purposes we increasingly rely on our own site search, which includes Daily Links, curated for originality, quality etc. Those old articles are already getting difficult to find in mainstream search engines, even if they are still online. █
