Episode 580: Sexual Harry Potter
Summary
Hosts Liz and Young Chomsky interview David Z. Morris to dissect the "Tescriol" complex, a network of effective altruism, rationalism, and transhumanism that has gone mainstream. The episode critiques the community’s incestuous funding structures, particularly the influence of Dustin Moskovitz, and argues that its obsession with predicting the future through data is a form of obscured utilitarianism. They explore how figures like Eliezer Yudkowsky and Peter Thiel represent opposing yet ideologically aligned factions, while analyzing the financial risks of AI companies like Anthropic. The di…
Topics discussed
Intro: The 'Rationalist Donkey' and AI doom
Apology for 'Sexual Harry Potter' info hazard
The 'Circle Jerk' and LLM-esque community dynamics
Debate on the word 'Spelunking' vs 'Caving'
Introduction of guest Sam Gilman and his book
From niche subcultures to mainstream SNL parody
Yudkowsky, Sanders, and the 'Sexual Harry Potter' link
Defining 'Teskreal' and the EA/Rationalist scene
The shared premises underlying internal disagreements
Extropianism: Controlling chaos and predicting the future
Thiel vs. Yudkowsky: Accelerationism vs. Doom
LessWrong archives and the content economy
Lack of intellectual diversity and METR analytics
Dustin Moskovitz's dominance in EA funding
Moskovitz's book promotion and unclear funding sources
McCaskill, SBF, and the self-reinforcing money flow
Anthropic's massive losses and risk factors
The incestuous social network of the AI scene
The 'Zero to One' model and Web 2.0 atavism
SBF's fraud: Investing stolen money to 'grow out'
Naive assumptions and Ashenbrenner's leverage
Accelerated funding and institutional angles
Questioning the actual work of AI safety
Lack of peer-reviewed academic credentials
The real value of AI in industrial sectors
Consumer AI: Lightweight agents vs. prompt windows
Jacob Coxon's departure and AI personhood claims
The shared premise of AI as God/Superintelligence
Lightweight agents vs. Frontier models
Mathematicians' fears and the Navier-Stokes case
AI data arrangement vs. genuine problem solving
Zuckerberg's 'Muse' and the agent vision
Contradictions in the 'everyone has an agent' pitch
Why people might not hand their phones to agents
Business use cases and optimization focus
The desire for contradiction vs. optimization
The origin of the optimization bias in EA
Peter Singer's parable and comparing lives
Financializing and contractualizing human life
Assigning dollar values to human ideas
The absurdity of categorizing everything with numbers
Shunning antisocial optimization thinking
The Epstein connection and adjacent ideas
The vastness of the adjacent idea landscape
Harry Potter and the Methods of Rationality
Rationality as a recruiting tool for EA
The need for a reliable decision-making system
AI as the higher species and the 'psychotic doula'
Edge cases, nuance, and the bed net example
Second-order effects of bed net distribution
The timeline of taking AI labs seriously
FTX money funding Anthropic's Series B
SBF's investment in Anthropic to keep growing
Will McCaskill's role in SBF's EA journey
The interconnected web of Alameda, FTX, and EA
The 'AI Safety is a Sex Cult' substack
The 2015 Future of Life Institute conference
Early AI conference panels and key figures
Intelligence Explosion and Friendly AI debates
The 2015 conference lineup and Yudkowsky's timing
Epstein emails and Yasha Bach's island visit
Untangling the software vision and Epstein links
The sex connection and the deeper Epstein vibe
Bostrom's racist emails and the FLI shutdown
Curtis Yarvin, eugenics, and the Manifest conference
Transactional mathematics and breeding programs
The sheer amount of Moskovitz's money
Niche internet ideas becoming mainstream
Thiel supporting Yudkowsky as controlled opposition
Yudkowsky's absurd ideas distracting from real harms
Continuous surveillance and mapping behavior
Military thinking and data collection as ruling class project
Extropianism and the desire for eternal life
Pushing back against total knowledge and stability
The 'Virgin' posture of tech leaders
Branding issues and the utilitarianism of weirdness
Intimidation by nuance and protected environments
The logic of the code and Baudrillard
P-doom and Bayesian operations
Measurement cracks and the track record of EA
Categorization and the taming of chance
The 19th-century bourgeois obsession with control
Financializing probability and the logic of the code
SBF's trick and Anthropic's eternal losses
Optimizing portfolios and moving cash
Destabilizing and crashing systems
The specious argument about bank crashes
Blind spots and synchronized model-driven moves
The limits of epistemology and knowability
Creativity in naming and Maxwell's Demon
Entropy production and the failure of extropy
Closing remarks and future topics
Outro and podcast sign-off
Listen ad-free on Castria