The highs and lows of vibe coding, cooking and math, addiction to prediction markets, and more. Plus Dustin Gouker on whether sports betting can take down Iran.
Thanks for reading Riskgaming by Lux Capital! This post is public so feel free to share it.
On Thursday, Palo Alto Networks, the cybersecurity firm, acquired Lux portfolio company Chronosphere, a platform built to give customers more control over, insight into and security for their data. Meanwhile, Fiddler AI, the Lux company focused on allowing developers and users to better understand, analyze, validate and monitor their AI systems at scale, raised $30 million in Series C funding. Another company, Astranis, announced it has signed a 9-figure deal with Oman to build the kingdom its first dedicated communications satellite.
Everyone is Claude Coding these days. Whether you’ve been trying it out or not, “10 things I learned from burning myself out with AI coding agents,” recommended by our scientist-in-residence, Sam Arbesman, is worth a read. Benj Edwards explains what they’re good at, what they’re not good at and why they’ll probably end up creating more work for humans rather than making us obsolete.
Due to what might poetically be called “preconceived notions” baked into a coding model’s neural network (more technically, statistical semantic associations), it can be difficult to get AI agents to create truly novel things, even if you carefully spell out what you want.
For example, I spent four days trying to get Claude Code to create an Atari 800 version of my HTML game Violent Checkers, but it had trouble because in the game’s design, the squares on the checkerboard don’t matter beyond their starting positions. No matter how many times I told the agent (and made notes in my Claude project files), it would come back to trying to center the pieces to the squares, snap them within squares, or use the squares as a logical basis of the game’s calculations.
In a similar vein, I loved scrolling through Isometric NYC, a project from software engineer Andy Coenen. Andy used AI coding agents and image models to create an isometric pixel-art map of the city. He found that the coding agents were transformative for building software tools, but image models struggled with consistency and couldn’t reliably assess their own mistakes. Whatever work the process took, the end result is beautiful.
Now for something completely analog: Sam also liked this charming argument for cooking with a slide-rule in Entropic Thoughts. The piece is also a sneaky math lesson; as the author notes “this is not so much about cooking as it is about having fun with maths in an almost useful way.”
Kitchen work is all about proportions: sometimes the recipe is for four servings but you need six; maybe the recipe calls for 80 g of butter but you only have 57 g, so you have to adjust the other ingredients to match.
We could use an electronic calculator to figure out the rescaled amounts, but a slide rule makes it so much easier.
Earlier this week, I posted an interview with Dustin Gouker about sports betting and the broader implications of prediction markets for politics and beyond. Over in Harpers, Jasper Craven gets into the topic from a different angle: by placing wagers himself. Four years after New York legalized sports gambling, he’s spent $18,000 and explored the ins and outs of Las Vegas’ de minimis gambling addiction efforts.
As I held my head in my hands, Ted told me that he far prefers watching sports without money on the line. He no longer frets over the minutiae of passing yards, and can simply enjoy the pure, invigorating thrill of witnessing athletic excellence. As we left the stadium, I was feeling eczema-level itchy to recoup my losses, and Ted passed along a helpful tip given to him by a former sponsor. He said that if an addict ever finds himself in a casino, he should ignore the buzzy slot machines and focus instead on the faces of the people playing them.
Finally, new Lux-er Hank Anderson directs you to a series of policy memos on industrialization from the Knudsen Industrial Strategy Fellowship. In one, Divya Nagaraj proposes a National Autonomy Certification Framework (NACF) to address the gap in the United States between innovation and deployment of new technologies, especially in autonomous systems like drones and self-driving vehicles.
The U.S. is the clear leader in early-stage autonomy research, but this leadership stalls when systems move from controlled lab or synthetic environments to real-world deployment. This “deployment gap” threatens vital sectors like logistics, transportation, and manufacturing. The friction is visible across domains: from flying drones beyond visual line of sight (BVLOS) to fielding unmanned maritime systems alongside manned vessels.
