Like many of us who value the human side of the web, I am looking for ways to keep slop at bay. It seems like LLM detection technology works reasonably well nowadays, with Pangram as the main provider. It is expensive, though!

Assuming pangram scans are trustworthy, and it seems they are, we could create a centralized index of pangram scans of articles. I have already set up a prototype where I publish results of my own pangram scans. I could create more users, each tied to its own API key, so others can contribute with scans as well. This is way more resource efficient than everyone running their own scans.

What do you think?

Update: since this is receiving quite some attention, let me clarify one thing. I'm not proposing to change Lobste.rs itself. I'm wondering if there are enough people who'd like to join forces to maintain an independent index of pangram scans, for our own use, not intending to impose it on anyone (e.g., maybe we could have a browser extension that automatically shows the scan results for articles when browsing Lobste.rs or the orange site).