Something about the not-quite-rightness of Markov output has always really amused me, like I can’t help but giggle at it. My all-time favorite in this vein was

Something about the not-quite-rightness of Markov output has always really amused me, like I can’t help but giggle at it. My all-time favorite in this vein was @erowidrecruiter on Twitter, which generated posts with Markov chains from a corpus of tech recruiter emails and drug experience reports from erowid.org. Still up but no longer posting: https://x.com/erowidrecruiter?lang=en https://x.com/erowidrecruiter?lang=en