thought email delivery instructions; proxy through language model or something to fix headers cryptography thought, amplified forensics daydreams, unsolvable puzzle fun high chance of wrongness due to category of things we don’t try regarding considering things we don’t have experience with i think this is likely known in cryptography? maybe not? i _think_ you could identify the sender of any encrypted text given sufficient text - by taking advantage of a problem in randomness collection where it is hard to differentiate between underlying information and random skews and biases - assuming you have models of the users in question it’s very hard to decrypt the text because that means identifying every character, but if you must want to identify the users you can sum the statistical information from all the characters together i’m having what’s described as inhibition now. the plan to excitedly make steps on it derails toward a closing off of my engagement eith it if it’s new, maybe we could use surveillance/forensics encouragement to make a paper out of it or someth8ng and from a crazy homeless person into a recognized independent researcher! note: this may be unlikely given qualifications on the daydream but it could help us work on it and we have so many ideas it would likely work with one of them