Fermi to Dyson: 'With 4 parameters I can fit an elephant'
mmbronstein · x · 2026-08-31
In 1953, physicist Freeman Dyson showed Enrico Fermi graphs demonstrating that his theory agreed with experimental results. Fermi was unimpressed, criticizing Dyson for lacking both physical intuition and rigorous mathematical methodology.
When Dyson defended his work based on its agreement with experiments, Fermi asked how many adjustable parameters he had used. 'Four,' Dyson replied.
Fermi smiled, quoting Johnny von Neumann: 'I remember my old friend Johnny von Neumann used to say that with four parameters I can fit an elephant, and with five I can make his trunk move.' This anecdote is often cited to criticize overfitting and model tuning lacking physical meaning.
More from Fun
- Tired of 'Claudish' writing; visually identifying AI slop via bio-prefill — beffjezos · 2026-08-31
- Dev fills heap with junk data; Claude locks everyone out, company goes bankrupt — tetsuoai · 2026-08-31
- Game Dev Vavra: Everyone Uses AI, Despite Public Boycotts — marian_nmt · 2026-08-31
- AI model fails recognition task, misidentifying subject in demo — ChrisGPT · 2026-08-31
- Reverse engineering binaries with GPT-5.6 Sol Ultra — teortaxesTex · 2026-08-31
- Google AI Overview Misspells 'crucifix' as 'cruifix' — Adventurous-Air-6503 · 2026-08-31