Ep. 13 | Can Constraints Make AI Better? | David Epstein

David Epstein is one of America’s best writers about social and biological science, who also happens to be deeply versed in the General Magic story, and has written about it in his new book, Inside the Box: How Constraints Make Us Better. He is also the best-selling author of The Sports Gene: Inside the Science of Extraordinary Athletic Performance, and Range: Why Generalists Triumph in a Specialized World. His books are must-reads—please check them out.

David is a brilliant debunker of myths and shibboleths about how sports stars, entrepreneurs, artists and inventors achieve excellence, and a spellbinding storyteller about how it’s actually accomplished. In Inside the Box, he uses General Magic’s legendary failure to demonstrate how having too much freedom inhibits, rather than enables, success. Magic had plenty of money and talent, but no managers or hands-on board imposing deadlines and budget limits on its headstrong crew of visionary engineers. That delayed our first mobile device, an over-featured smartphone that came too late.

Hersh and Mike had a blast reliving some of those heady if ill-fated days with David, and exploring his stimulating, often contrarian views about the best ways to enable creativity to flourish. Constraints, in his telling, aren’t the enemy of invention. They’re how you decide what is worth building. He leaves listeners with a practical question for the AI era: what would it look like to build with constraints on purpose?

As always, the full-length version of this interview is posted on our YouTube channel:

In this episode

* 03:06 How an introduction to Tony Fadell led to including General Magic in his book

* 08:45 General Magic’s alumni compared with “the PayPal Mafia”

* 11:42 A cursory 1,000-year history of technology disruption, and what it says about AI

* 16:07 Why AI doesn’t have to be as good as people to take their jobs

* 19:34 Mancur Olson and whether policymakers will act

* 22:50 “Write the press release first”: Tony’s test for defining the problem

* 25:02 Telescript, Agentic AI, and the risk of misuse

* 26:02 Marc Porat’s PhD thesis, which saw the future

* 30:07 Towns organizing against secret data-center deals

* 35:56 Social media as a collective trap

* 38:46 Why he had to filter his own name off YouTube

* 46:56 Can you get sued for what you write on Substack?

* 50:12 Individualizing work weeks, and what we’ve done to ourselves

About David Epstein

David Epstein is a science-trained journalist and bestselling author. He holds a B.S. in Environmental Science and Astronomy from Columbia University (2002) and master’s degrees in Environmental Science and Journalism. He was a senior writer at Sports Illustrated, an investigative reporter at ProPublica, and host of Slate’s How To! podcast. He has worked as an independent author and speaker since 2019. His books are The Sports Gene, Range: Why Generalists Triumph in a Specialized World, and Inside the Box: How Constraints Make Us Better (2026). He writes the Range Widely Substack.

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit spellboundproductions.substack.com