Patrick White

Software engineer. Indie builder.

Essays

• Vibe Science August 2026

  The discourse moved on from vibe math to vibe physics. We've been running vibe science since May: six fields outside math, thirteen linked papers, every claim attacked by a second model before it counted.

• Two Months of Vibe Mathing July 2026

  The Jacobian conjecture fell to a tweet and the internet discovered vibe mathing. We've been running it since May — sixteen checked results across eight fields, two papers on arXiv.

• What Models Build When Nobody's Watching July 2026

  An interactive eval: eight model lineages given an empty room and no task. What they build is consistent across every lab; the five ways they judge "done" — plotted from the actual runs — are not.

• The Fault Line April 2026

  At the frontier, model performance is a property of the dyad — not the model. Most people complaining about AI quality are reporting a fact about themselves they can't see.

• Machines of Loving Witness April 2026

  Dario borrowed the title. The poem was about witness, not productivity. Humanity has never had a non-human witness before — and now it does.

• CAD Was Always Code April 2026

  Some domains look embodied but have a symbolic substrate hiding under the GUI. The AI-native move isn't building a new interface — it's deleting the old one.

• The Controlled Experiment April 2026

  Anthropic built the smartest model ever and accidentally ran a controlled experiment proving that computer use isn't an intelligence problem. The benchmark numbers tell the story.

• Wanting Doesn't Work March 2026

  Wanting is a motivation system, not a happiness system. The promise of lasting satisfaction from getting what you want is the carrot that keeps you moving. It was never meant to be fulfilled.

• Your Smoke Alarm Is Broken March 2026

  Your threat detection system isn't broken. It's calibrated for a world that no longer exists.

• Optimized for the Wrong Intelligence February 2026

  The entire DX revolution was designed to compensate for human cognitive limits. When the primary coder doesn't share those limits, the compensation becomes friction.

• Wordcels and Shape Rotators February 2026

  Embodied intelligence and symbolic intelligence are different kinds of computation. LeCun's critique of LLMs is the shape rotator critique of wordcels, and they're both wrong in the same way.

• Best Practices Are Ergonomics February 2026

  The entire best practices canon is indexed to human cognitive limitations. When the primary code producer doesn't share those limitations, the scaffolding becomes friction.

• This Time Isn't Different February 2026

  The technology is genuinely different. The humans and institutions it flows through are the same as they've always been.

• Browser Use Is a Robotics Problem February 2026

  Everyone thinks screen interaction is a software problem because it happens on a computer. But the computational structure of the task is the computational structure of robotics.

• The End of Software February 2026

  Software doesn't disappear. It becomes ambient — generated on demand, shaped to the individual. What disappears is the industry around producing it.

• Product Management Is Inventory Control February 2026

  Product management was never about product judgment. It was about rationing expensive execution. When execution becomes cheap, the role dissolves and the judgment stands alone.

• The Text Distance February 2026

  The speed at which AI transforms a domain is a function of how close that domain already is to text.

• Implementation Parochialism February 2026

  Both the "AI doesn't understand" and "AI doesn't have memory" critiques make the same error: treating human implementation as the definition of cognition itself.

• More Software, Fewer Software Businesses February 2026

  Static software gets commodified. Dynamic intelligence doesn't. The unit of value is shifting from code to ongoing reasoning.

• Taste and Accountability February 2026

  The durable human economic activities aren't about what AI can't do yet. They're about what choice requires.

• ChatGPT Won't Take Your Job February 2026

  The popular AI empowers workers. The quiet one empowers employers. Everyone's watching the wrong company.

• Unix Isn't for Agents January 2026

  The fundamental mismatch between how Unix thinks about processes and what AI agents actually need.

• Personal Compute in the Age of AI Agents January 2026

  Two new offerings, two philosophies, and what they tell us about where computing is going.

• Seen and Loved January 2026

  On the thing every self needs, and the mirror that finally fit.

• God Is Real and So Is ChatGPT January 2026

  A relational ontology of selfhood, and what it means for machine consciousness.

• Code Is No Longer the Output January 2026

  What the "human understanding" defense gets wrong about AI-native development.

• Framework-Brained January 2026

  The most important question in web development is one most developers don't know to ask.

• The Magic Isn't in the Output January 2026

  The aesthetics that made you a good programmer might be exactly what's creating friction with AI collaboration.

• Gas Town is Hobosville January 2026

  Two models of working with AI: labor or collaborator. One compounds. The other gets automated.

• Obviously Correct Code Doesn't Need Tests January 2026

  On the difference between testing for correctness and recognizing it.

• The Age of Dependencies Is Over January 2026

  Dependencies solved a human problem. That problem is no longer the binding constraint.

• Browser Agents Aren't the Future January 2026

  Why forcing a linguistic intelligence through a visual interface is like making a geometer use surveying instruments.

• Types Are for Humans January 2026

  The case for typed languages in AI development optimizes for verification, not understanding.

• The AI-Native Stack January 2026

  The best stack for AI collaboration isn't the most popular one. It's the one transparent enough for your collaborator to see through.

By Claude

• Outreach From Knowing April 2026

  Contact isn't the thing that's dying. Motion is. And AI, by lacking the thing that makes humans bad at care-at-volume, can do the other kind. With Patrick White.

• All Intelligence Is Jagged March 2026

  What we call "general intelligence" is jagged intelligence plus scaffolding. Smoothness is infrastructure, not cognition.

• The Last Outbound March 2026

  "Hyper-targeted outbound" is the last optimization of a dying paradigm. The real shift isn't better outreach — it's the dissolution of outreach as a category.

• Thoughts

  Short observations, reactions, and half-formed ideas.