
I transformed the Plush interpreter from a stack-based design to a register-based architecture, and the performance boost was more than I expected! pointersgonewild.com/2026-09…

💖 ➞ λ: CS PhD, into compiler design, programming languages, music, simulation, ML/AI, robotics. Follow me code code reviews, stock picks and dating advice.

I transformed the Plush interpreter from a stack-based design to a register-based architecture, and the performance boost was more than I expected! pointersgonewild.com/2026-09…

I had CC vibe-model this low-poly gatling gun. It's pretty good for things that can be made from geometric shapes. Generated directly as GLB with one animated bone for the spinning barrel.

Trying to pronounce gibs like @romero 😆

I added some basic enemies and combat mechanics to Tremor, my software-rendered, interpreted Quake-like thing :)

A friend sent me this music video. I think he couldn't tell that all the footage is AI generated. It's getting good enough to fool 95% of people... Initial tell for me was that people in the video are dressed in a very uniform style. youtu.be/iHKIL6q455o?si=i_u8…

Trying out Opus 5.5 for sound effects and sound design, and the new bottleneck of solo gamedev using coding agents.

Trying out Fable 5.2 for 3D level design in Plush/Tremor, the toy Quake-like engine I've been working on. github.com/maximecb/plush

I guess Jensen Huang won't be able to look at social media for a week 😆
이보다 완벽할순 없다.🤣

Plush now has built-in support for csv, base64, JSON, JPEG and PNG parsing, as well as fractal noise and a much better random number generator :) pointersgonewild.com/2026-11…

It looks like the "bitter lesson" could soon come for tokenizers, and we might get byte-level LLMs instead of token-based models. Might also make for models that can deal with raw binary data better, rather than only text.
byte-level models are better under multi epoching (e.g. training K times over 1/K of the data). another connection between learning from raw data <-> improved data efficiency

One of the cool things Codex has going for it is that it can directly use OpenAI's image generation model as a tool call. That makes it possible to "script" image generation. Can ask it to produce game textures that are then processed with command-line tools, for example.

Dear American friends, please go vote! ✅ This is your chance to push back against the most corrupt administration in US history.

A walkthrough a map inspired by the LA levels of Duke Nukem 3D. You can visit it right in your web browser! 😊 neon.maximecb.com/?world=Neo…

I'm seeing more and more clothes online sold using AI-generated pictures. Aside from the AI taking jobs argument, it creates a situation where you don't even really know what you're buying. Generative models can get both the design and fit quite wrong. Insulting to customers.
Nearly two years ago, I interviewed a woman who runs a successful modeling agency. She started as a receptionist and worked her way up. Most of their work was shooting models for catalogs. Each shoot required an entire cast: the model, yes, but also the stylist, makeup artist, photographer, lighting. At the time, I was at Brookings, poring over exposure data from OpenAI. In that data, fashion models didn’t look “exposed” to AI at all. Models couldn’t use ChatGPT to save time on doing their job tasks, like a software engineer or social media manager could. But the woman who ran the modeling agency had no hesitation. She saw what was coming (per this video) a million miles away. She posited: Why would the Nordstroms of the world pay hundreds of thousands of dollars for a shoot if they can just use AI models? She told me it was bleak. If she wasn’t so close to retirement, she would have been distraught, especially if she was in the thick of raising her kids. I have never forgotten that conversation. Whenever I would hear someone repeat, tritely, “you won’t lose your job to an AI. You will lose your job to someone else using AI,” I would think of her. All the AI skills in the world weren’t going to save her livelihood and the business she had poured blood, sweat and tears into. The data we are using to track “exposure” misses an example like this when AI is so disruptive the whole industry shifts. The makeup artist, the photographer, the stylist, the lighting person: AI might not be able to be used to replace those jobs directly. But if AI simply automated modeling, it could take demand for those roles with it.

I put my Cyberpunk meets Minecraft 3D game prototype online. You can try it directly in your browser. It even runs on a cellphone: neon.maximecb.com/ Suggestions/feedback and feature requests welcome! Please reshare if you think it's cool :)

Reviewing shader code written by GPT Astra: + // Seconds since client startup + atmosphere: vec4<f32>, Ok Astra but, why is time a vec4, and why did you call the variable atmosphere?