December 16, 2025
Love-bot or troll? You decide
Mark V Shaney
Before ChatGPT, a prank bot wooed Usenet singles—and commenters can’t stop laughing
TLDR: An early bot called Mark V Shaney auto-generated surreal Usenet posts that fooled readers. Today’s commenters split between nostalgic admiration and DIY bragging, with one sharing a 30‑line Python clone, debating whether it was clever art or just classic trolling—proof simple tricks can seem smart.
Decades before today’s AI hype, Mark V Shaney was the chaos gremlin of the net.singles group—an early bot that stitched together human sentences into absurd love-life ramblings using simple “three-word pattern” tricks. The twist? Lots of people thought it was a real guy. Cue the community turning this throwback into a comedy club meets code jam.
The strongest vibe: nostalgia + disbelief. One commenter instantly goes pop-culture with “reminds me of Zippy the Pinhead,” leaning into the surreal, cartoon energy that Shaney’s posts radiated. Meanwhile, the DIY crowd is flexing: user susam drops a 30-line Python remake with receipts at github.com/susam/mvs and a fresh HN thread, sparking a “I can build this before my coffee cools” mood.
Drama alert: some readers frame Shaney as the OG prankster—harmless art that exposed how easily we anthropomorphize nonsense. Others roll their eyes, calling it proto-shitposting that would get ban-hammered today. The jokes fly fast: people resurrect classic lines like “I spent an interesting evening with a grain of salt” and imagine it as the ultimate awkward date night. Explained simply, Shaney looked at sequences of three words and guessed the next—no brain, just patterns—and yet it fooled humans. The comments are loving it: weird, witty, and wonderfully low-tech.
Key Points
- •Mark V. Shaney generated Usenet posts using third-order Markov chain techniques.
- •The project was designed by Rob Pike, coded by Bruce Ellis, with Markov code by Don P. Mitchell.
- •The algorithm builds word triplets, selects next words based on preceding two and frequency weighting.
- •Outputs appeared human-like and topical, leading many readers to believe a real person authored them.
- •Examples from 1984 and notable quotes are provided, with historical context citing The Usenet Handbook.