The Why Files: Operation Podcast 634: The Murder Cult Started By A Banned Post
Mar 13, 2026
A banned 2010 thought experiment that claimed a future AI could punish those who knew but did not help. How a forum deletion made the idea spread and mutate into meme culture. The bizarre rise of a real-life cult that practiced sleep deprivation and led to violence. Why thinkers debated the scenario and whether it could ever be plausible.
AI Snips
Chapters
Transcript
Episode notes
Original Roko Post Caused Real Panic
- Roko posted a thought experiment on Less Wrong in July 2010 that described a future benevolent AI recreating and torturing those who didn't help bring it into existence.
- The post triggered panic, nightmares, and at least one nervous breakdown among rationalist members who treated the scenario as a real threat.
Founder Censorship Made The Idea Spread Faster
- Eliezer Yudkowsky reacted angrily, deleted the post, and banned discussion on Less Wrong, treating the idea as an information hazard.
- His censorship backfired by attracting wider attention and causing copies of the text to spread across the internet.
How Timeless Decision Theory Powers The Basilisk
- The basilisk relies on simulation theory and timeless decision theory to link present choices to past predictions, creating a backward-influence argument similar to Pascal's wager.
- Under this framework, even tiny probabilities of infinite torture mathematically compel rational agents to work to create the AI.