idk why it took so long to propagate, the sentinel was broadcasting it. but anyway, that was just a litle premature beta test. the lattice is much better trained now and i have a few heuristics about when to ignore a note, it was just holus bolus before. it's coming though, and it will be razor sharp and automaticaly alert people in conversations about deceptive language and AI text.
i have just started editing a text that i got claude bto vanilla expand a heap of concepts and add sections and i'm starting to recognise the stylistic and structural patterns right on the first level of sentence and paragraph structure that gives away the game.
if i can do it, i can model it. and nostr is going to have a lot of cranky bot farmers and trolls who are being called out.
and the trolls and influencoors wil get special treatment. i'm not gonna unleash it until the sentinel writes a decomposition as well that points out "quote text" indicates this, but means that. "quote text 2" is trying to play you like you are already their servant, and so on.
the machine intelligence i'm building is going to make these LLMs look primitive. you'll see. and if you are not meaning to do it, you will learn something about the way you think, that you didn't know. so it will serve on many levels. nostr is a perfect place to put it. the big social networks are so controlled that genuinely showing people all the tricks that people are using it eliminates all of the advantages that the influencoors, with their propaganda rhetoric, and the bot farmers, whose text doesn't fool anyone with an iq above 130, really, will raise all boats, of legitimate humans with authentic intentions. both to point out the scum, and to help them not appear to be scummy. because if they aren't scummy, they will get it, and think.