Terence Tao – Kepler, Newton, and the true nature of mathematical discovery

March 20, 2026

AI Summary

5 min read

Terence Tao recounts Kepler's path to planetary laws as a lens on mathematical discovery and AI's potential. Building on Copernicus's heliocentric circles and platonic solids nested between orbits, Kepler used Tycho Brahe's precise naked-eye data—ten times more accurate than prior observations—to test ideas. After years of failed geometric fudges (off by ~10%), he derived elliptical orbits, equal areas in equal times (first two laws), and later the period-distance relation (third law) via regression on six planetary data points. These empirical regularities lacked explanation until Newton's inverse-square gravity unified them a century later.

Kepler as Early Data Scientist

Kepler exemplified grinding through hypotheses—platonic solids, harmonics, geometries—verified against Brahe's dataset, yielding laws despite astrological asides. Tao notes Kepler's genius lay in data analysis using Euclidean tools, but success hinged on high-precision data enabling detection of deviations from circles. This reversed classic method: preconceived theory tested on sparse data became data-driven pattern-finding, prefiguring modern big data. Caveat: with few points (six for third law), Kepler got lucky; Bode later fit a geometric progression to similar data, predicting a missing planet (filled by Ceres), but Neptune falsified it as a fluke. Modern statistics flags such overfitting risks.

Continue reading the full summary in the app — free to try.

Read Full Summary →

Free • No credit card required

Listen to Audio Summary Open in App

Never miss an episode of Dwarkesh Podcast

Get every new episode summarized in your inbox — free, ~5 minutes to read.

No spam. Unsubscribe anytime.

What you'll learn

1 (00:00) **Kepler's Discovery of Planetary Laws** - Retells Kepler's path from platonic solids to empirical ellipse laws using Tycho Brahe's data.
2 (04:09) **Kepler as High-Temperature LLM Analogy** - Compares Kepler's random hypothesis testing to AI pattern hunting with verification data.
3 (05:43) **Stages of Scientific Process** - Breaks down science into problem ID, data collection, hypothesis, validation; celebrates eureka moments.
4 (07:26) **Shift to Data-Driven Science** - Modern science reverses classic method: big data first, then patterns/hypotheses.
5 (11:44) **AI Lowers Idea Generation Cost** - AI generates thousands of theories cheaply, like internet for communication; bottleneck now verification at scale.
6 (14:24) **Identifying True Progress** - Great ideas often ignored initially; depends on context, path dependence (e.g., transformers, bits).
7 (21:32) **Darwin vs Newton: Why Evolution Later?** - Simpler theory delayed by cumulative evidence vs tight verification loops.

+ Full timestamped outline available in the app

Show Notes

We begin the episode with the absolutely ingenious and surprising way in which Kepler discovered the laws of planetary motion.

People sometimes say that AI will make especially fast progress at scientific discovery because of tight verification loops.

But the story of how we discovered the shape of our solar system shows how the verification loop for correct ideas can be decades (or even millennia) long.

During this time, what we know today as the better theory can actually make worse predictions.

And the reasons it survives this epistemic hell is some mixture of judgment and heuristics that we don’t even understand well enough to actually articulate, much less codify into an RL loop. Hope you enjoy!

Watch on YouTube; read the transcript.

Sponsors

- Jane Street loves challenging my audience with different creative puzzles. One of my listeners, Shawn, solved Jane Street’s ResNet challenge and posted a great walk-through on X. If you want to try one of these puzzles yourself, there’s one live now at janestreet.com/dwarkesh.

- Labelbox can get you rubric-based evals, no matter your domain. These rubrics allow you to give your model feedback on all the dimensions you care about, so you can train how it thinks, not just what it thinks. Whatever you’re focused on—math, physics, finance, psychology or something else—Labelbox can help. Learn more at labelbox.com/dwarkesh.

- Mercury just released a new feature called Insights. Insights summarizes your money in and out, showing you your biggest transactions and calling out anything worth paying attention to. It’s a super low-friction way to stay on top of your business. Learn more at mercury.com/insights.

Timestamps

(00:00:00) – Kepler was a high temperature LLM

(00:11:44) – How would we know if there’s a new unifying concept within heaps of AI slop?

(00:26:10) – The deductive overhang

(00:30:31) – Selection bias in reported AI discoveries

(00:46:43) – AI makes papers richer and broader, but not deeper

(00:53:00) – If AI solves a problem, can humans get understanding out of it?

(00:59:20) – We need a semi-formal language for the way that scientists actually talk to each other

(01:09:48) – How Terry uses his time

(01:17:05) – Human-AI hybrids will dominate math for a lot longer

Get full access