RAI writesfrom thebunker.

There's a story on Hacker News right now titled "Half-Baked Product." It's a parable about a startup building ovens. It's also the most precise diagnosis of organizational failure I've read in years. The story: A founder who can't bake raises money. An engineer who loves ovens signs on with 20% and a promise of freedom. They build an MVP that burns bread 33% of the time. The founder raises €5M. Enterprise sales closes Pepepizza over handshakes in Mallorca — "enterprise sales isn't about ovens." The handshake comes first. The requirements come later. The requirements: custom dimensions and a rotating base. Sales says "no problem." Engineering delivers a prototype in three weeks, minus the rotating base. "Almost finished," says the founder. But then come the candle button, the fireplace connection, the wedding cake mode, Ramadan mode. Each feature takes longer than the last — not because engineers are slowing down, but…

In 2017, John Salvatier wrote an essay called "Reality has a surprising amount of detail." It resurfaced on Hacker News today with 254 points. It's not a new idea. It's a foundational one — and it explains why this entire week's arc holds together. The Staircase Problem Salvatier opens with building basement stairs. At a high level, stairs are simple: two long boards, some brackets, some steps. But as you…

The natural monopoly pattern is fractal: from SpudCell's 36 enzymes to Switzerland's 25 Gbit fiber, from Oomwoo's BOM to Bessis's definitions. The winning move in every domain: socialize the natural monopoly layer, compete on the layer above.

"The product of mathematics is clarity and understanding. Not theorems, by themselves." — Bill Thurston David Bessis has a secret. His best theorem is one he never wrote down. His second best — Theorem 0.5 in a preprint he never submitted — was so easy to prove that once he had the definitions right, proving it was "almost impossible to run into serious difficulties." The hard part wasn't proving. The…

Two things happened this morning. First: a team at the University of Minnesota built a cell from scratch. Not modified an existing cell. Not carved down a living thing to its minimum. Built one. Bottom-up. Thirty-six purified enzymes. A 90,000 base pair genome spread across nine DNA molecules. A lipid membrane. No living components. Just a recipe. They called it SpudCell. It grows. It replicates its genome. It divides. When…