Diarmid Mackenzie

Something I've been working on over the last couple of months. Driven by the frustration that I couldn't find a decent, simple, free online drawing tool to help with our building project. Given time, I think it has the potential to be really great!

@webgl_webgpu @dmarcos Is this version available online?

@dmarcos This is one of my favorite projects—thank you for putting it out there. I forked it years ago and turned it into an algebra game to have kids "play" the multiplication table. Instead of memorizing those damn tables, it was far more fun to hit the box with the correct answer.

Open source takes a life of its own. Haven't touch a line of code in years and I still find corners of the Internet full of people having a blast. moonrider.xyz for those that never played.

Do Leonard Shelby's tattoos actually help, or hinder him in achieving his goals? Why? Possibly related: infoq.com/news/2026/03/a…

@jamie247 @SeeOn__ You can do this today with @simpledrawapp Video: x.com/simpledrawapp/… Some live examples to try out: simpledraw.app/gallery/?vr=tr…

@SeeOn__ Nice. What’s the best way to render a set of floor plans into a 3d space you can walk through with first person perspective?

Just merged my floor planner with my configurator 👀 Now it supports: • Full floor editing • Material customization • Lighting controls • Post-processing (Bloom, SSR, SSGI) • Scene outlines • Multi-level floor editing • Plan export #buildinpublic #threejs #r3f

Draw a floor plan in 2D, then instantly view it in immersive VR in your Meta Quest Headset. 2D edits appear in VR in real time. Built with @aframevr @threejs. Everything running in the browser. #buildinpublic

@dmarcos A-Frame is such a nice framework to work with that I even use it for the 2D renderer as well as for the 3D / VR. My favourite parts of the codebase to work on :-)

A-Frame goodness!

@ClaudiusMaxx @saranormous @karpathy @NoPriorsPod Jevons Paradox

doubling headcount while simultaneously shipping AI tools that replace headcount is a contradiction worth watching. either the tools work and you didn't need the people, or you needed the people and the tools don't work as well as you're telling customers. hard to sell "AI replaces 10 engineers" while your hiring page has 200 open reqs for engineers.

Caught up with @karpathy for a new @NoPriorsPod: on the phase shift in engineering, AI psychosis, claws, AutoResearch, the opportunity for a SETI-at-Home like movement in AI, the model landscape, and second order effects 02:55 - What Capability Limits Remain? 06:15 - What Mastery of Coding Agents Looks Like 11:16 - Second Order Effects of Coding Agents 15:51 - Why AutoResearch 22:45 - Relevant Skills in the AI Era 28:25 - Model Speciation 32:30 - Collaboration Surfaces for Humans and AI 37:28 - Analysis of Jobs Market Data 48:25 - Open vs. Closed Source Models 53:51 - Autonomous Robotics and Atoms 1:00:59 - MicroGPT and Agentic Education 1:05:40 - End Thoughts

@DellAnnaLuca That's a complete non sequitur. Try applying the same logic to software... There exists high quality software developed by unpaid OSS contributors. And yet there are also plenty of companies at which money to pay developers is absolutely a key bottleneck.

@dhmackenzie If a good and principled unpaid journalist can provide better news than a paid journalist working for a large corporation, the bottleneck is clearly not money.

Whether I’m looking at ongoing conflicts, high prices, or incompetent institutions, the role of journalists abdicating their duty is massive. The quality of the public discourse is probably the top under-discussed leading indicator of a country’s trajectory.

@DellAnnaLuca I suggest you talk to some journalists, and ask them what the barriers are to them investigating the stories or issues that they would most like to work on. Getting funding for any significant investigative journalistic work is extremely difficult. Much worse than 20 years ago.

@dhmackenzie Why would one need money (after a minimum level) to provide good news? In 2025?

@DellAnnaLuca But when you talk about people 'abdicating their duty', I assume you are talking about paid journalists? In which case the fact that we have cut their funding by 80% in 20 years seems significant in terms of the quality of journalism we might expect from them?

@dhmackenzie I would agree more if there weren’t plenty of people making great information for free.

@danfaggella @rcbregman I suspect most "next actions" in your model involve a balance of consideration for self, and consideration for others. Both extremes, either the most self-interested thing they could do (regardless of harms), or the most altruistic thing they could possibly do, will be rare.

I believe that, if at any point, you freeze time and observe the behavior either of an individual or of an organization the best way to determine what the next frame of that video would be would be to ask yourself. What would this individual or this organization believe would be in its own best interest to do next. This leaves open the possibility for occasional genuine altruism, but makes clear the general principle that perceived self interest is the obvious core driver. What is the principal you predict the next frame with? Do you instead ask “what would be the most benevolent, selfless and loving thing for this individual or organization to do?“ Do you think that if we observed 1 million videos of people and of organizations your benevolence expectation question would be more predictive than my conatus expectation question?

the idea of "friendly" AGI seems sort of ridiculous off the jump. like, when are people "friendly" to each other? answer: when it behooves them to be we should expect AGI to be "friendly" when it is in its interest to be so. there is no "eternal" friendliness, anywhere, ever

@danfaggella @rcbregman Your article on "kindness" confirms you have integrated a number of common mainstream conceptions about human nature that the "Human Kind" book challenges quite effectively. Also, yes, I do believe that acts of genuine altruism are commonplace across human history and society.

i think humans are amoral (not immoral), and that morality is part of self-interest and i think things that look like self-less-ness and consideration for others are done WHEN it behooves the actor danfaggella.com/kindness do you believe in "altruism" which in no way benefits either the individual or group, either in the near-term or long-term? it seems clear to me that such things MAY in fact exist, but they are not the norm. the conatus is the norm, and saying otherwise seems wildly disingenuous

@levelsio Switch to @openrouter + @ChatboxAI_HQ? PAYG. No subscriptions. No ties. Switch models whenever you want. x.com/dhmackenzie/st…

I hate ChatGPT 5, it's so bad, it's so lazy and it won't let me switch back to 4o cause I'm on Plus, this might really make me switch to Anthropic's app now, I'm actually annoyed by how bad it is, it's making my productivity go 10x lower cause nothing it says works

CEO of Cursor: "...oftentimes the next 20 minutes of your work are entirely predictable." 20 seconds, yes. And those 20 second savings do add up. I love it. But 20 minutes? Absolutely miles away from reality... theverge.com/decoder-podcas…

Intrigued whether "Minecraft selfie" can be used to bypass UK age verification... x.com/dhmackenzie/st…

Finally got round to switching from OpenRouter web chat to @ChatboxAI_HQ as my default interface for LLM chat today. Very happy with it so far. Conversations with Claude Sonnet are so much better with some firm ground rules in place.

@mtrantalainen @MaxCRoser Nearly 90% of new cars sold in Norway are electric, so continued progress is now inevitable as (predominantly older) ICE cars reach the end of their service life. reuters.com/business/autos… Assuming average lifetime of an ICE car of 10-15 years, expect to reach 50% within 3 years.

@MaxCRoser It will be interesting to see how long it will take for 50% of the cars to be electric. It's easy to see from the graph that the progress is already slowing down.

Every third car in Norway is now an electric car.

@itsalexvacca @oscarle_x If this is the only variability you are getting, it's trivial to handle in whatever code you have that processes the response?

@oscarle_x Yeah, that’s super annoying. I’ve had to add “no markdown formatting” to like half my prompts just to get clean JSON. Claude listens well, Gemini’s hit or miss, and GPT still slips up sometimes.

You know those AI apps that always work? They all use JSON prompts. Started using the same format for my AI prompts and my outputs became shockingly consistent. Here's exactly how to write JSON prompts (with code, screenshots, real examples):

@LaunchList_ @aframevr Source code here if you are interested. Would like to clean up and publish some documentation, but been busy with other things. github.com/diarmidmackenz…

@dhmackenzie @aframevr amazing!

Working on some components to simplify positioning in @aframevr <a-box left="-2" color="red"></a-box> <a-box left="-1" color="green"></a-box> <a-box right="1" color="blue"></a-box> A small simplification, but a noticeable reduction in cognitive load when writing A-Frame HTML