
N N Taleb's Probability Questions (UNOFFICIAL) Stephen Wolfram visits RWRI 20 (The Real World Risk Summer School, 2025)
14 snips
Nov 11, 2025 Stephen Wolfram, computational scientist and founder of Wolfram Research, explains computational irreducibility, the Ruliad, and how simple rules can produce universal computation. He connects hypergraph models to space-time, reframes quantum mechanics as branching histories, and compares evolution and machine learning as searches through rule space. The conversation also touches on AI governance, creativity, and limits to prediction.
AI Snips
Chapters
Books
Transcript
Episode notes
Cat-In-Party-Hat Example
- Wolfram illustrated a tiny corner of the Ruliad with generative-AI outputs for "cat in a party hat."
- He showed concept-space has vast "inter-concept" regions most humans haven't colonized.
Mathematics As A Network
- Mathematics can be viewed as networks of expressions connected by axioms; proofs are paths in that graph.
- Mathematicians operate at higher-level pockets of reducibility, not exhausting low-level axiom paths.
Evolution Uses Coarse Fitness
- Evolution succeeds because fitness functions are computationally weak compared to organism development.
- Adaptive evolution finds workable lumps of irreducible computation that satisfy coarse survival criteria.




