Octamed

@dmwlff @ThePrimeagen I think people don't realise how hard it is to just know what you're after. I mean programming is just 'writing down what you want', so I see AI coding as the same. In fact, I think I write MORE when I use AI to code. Turns out english isn't a great programming language :)

@Octamed @ThePrimeagen Probably some day it will. We already have memory and dreaming. It's hard to get these features right without some human guidance, though. But the biggest things I have learned from doing this exercise are about _myself_.

Something that has greatly improved the reliability of the code I produce via LLM is a technique I took from cursor. Field guide. First, how I start. I nitpick everything and I always make it produce the exact code I want, I use voice dictation and talk for quite a bit. Second, every single back and forth is recorded to the markdown file. Along with each commit hash for the change. Then I have the agent review the back and forth and the changes and come up with durable and generalized feedback that it stores in a directory (field guide). The directory has init.md that is just a bunch of links and short descriptions for each file. This way I don't blow the context, but instead it can do some basic tool calls and operations to find exactly what it's looking for. Over the last 10 sessions, the in-depth feedback I've had to give has greatly reduced despite no change in model or harness that I'm aware of. Now the feedback is project specific, and is not universal. I am okay with that.

@kskrygan Yes, but mainly for knowing what code ISN'T AI.

Imagine a new feature: select any line of code and instantly see the prompt that generated it. Would you use this? In what situations would it be most valuable?

@MaxEllison2048 @ThePrimeagen @signulll Absolutely could. There's no animation, the background is just a picture, the trees and ground are very simple, the guy is very simple. If you have no talent at all then sure.

If they're actually building those assets from scratch then no, not an afternoon. You could get a bunch of primitives doing this but now you're just downplaying the complexity of the game dev pipelines. Getting a working prototype is quick. A solid casual game ready for app stores is still going to take most teams weeks or months. This Claude one shot also isn't ready for app stores. It's a pretty demo, but would take a lot of work to get ready for launch.

if you put this game on the app store in ~2012, it would’ve potentially grossed $5-10m at least over its life span, maybe even more. today it can be one shotted by roughly anyone with the correct physics, graphics, gameplay, etc. so $10m of economic value wiped out in ~15 years? that’s only one game. almost all software is going to be power law outcomes from this point forward. the temporary rise you see now is kind of a red herring for what’s to come.

@signulll But... it's not 2012? That's like saying some kid making a flappy bird clone can make 30 million dollars!!!*

@luckeyfaraday @signulll You can make games right now and have been able to for 50 years. This is really really simple.

@signulll 100% agree. Vibe coding games is going to be the next big thing. I always dreamed about making my own game when I was a kid, not it’s possible in less than an hour. Amazing times.

@AcePapers96261 @ChrisGPT It's really really simple though? Even the most basic of basic indie games are better than this in every way. Game Engines were the real thing that allowed indies to push quality.

@ChrisGPT This is indistinguishable from an early access steam game. This is crazy.

As I’ve been saying for the past 6 months the end of indie gaming studios is nigh. Soon it will be in the people’s hands. This is why I love Matt as well, he doesn’t produce slop. He pushes AI models to the max and makes all other demos look like absolute slop.

@ChrisGPT @teak0der It's 99.9% not COD. Coding is a tiny part of making a game.

@teak0der “Not even close” is such cope lol. Cmon

@mattshumer_ Wake us up WHEN it happens. Coming soon is meaningless. Also MAKING games is just 1000s of tiny decisions gradually as you play test it as a human for humans. The exact thing that AI is the most terrible at.

IMO it really sucks for those who have put their lives into game dev... but ppl burying their heads in the sand are doing themselves (and others) a disservice. The sooner people are aware of what's coming, the sooner they can position themselves well.

@dmwlff @ThePrimeagen Serious question. If all the things people talk and talk about worked, why doesn't Claude just do it anyway?

@ThePrimeagen Now do the meta version of this. Have the AI analyze all your sessions and start to generalize your priorities and techniques.

@ThePrimeagen The thing about all the endless talk about the best x/y/agent/harness workflow etc etc, is IF IT WORKED why doesn't Claude/Codex etc just do it anyway? I only ever see people talk about optimising their LLM workflow, then find out they're making something piss easy to do manually

@nulltron_ @ThePrimeagen People use that analogy all the time, but actually, it most dense cities, horses would be better

@ThePrimeagen i think its just virtue signaling. i picture how life was when we transitioned from horses to cars. im sure people clung to their horses for a long time after

what do you think of codeberg banning ai code? why? what is the benefit?

@willgoldstone @unity Great. Yeah I'm sure you've tried the hot reload plugins. I use the MIT opensource FastScriptReload, which works amazing for the contents of functions. Helping them out and intergrating that would be high value as I basically write half the game in play mode now. It's SO good.

@Octamed @unity At the moment it’s purely watching code changes on disk and sending them to the dev build, but we are working on various other ways to tackle what you’re talking about (eg tweak params in inspector for device)

Introducing live code iteration! Change gameplay code and see it live at runtime in development builds. No more lengthy rebuilds. Modify values or add logic, hit save, and see updates instantly on your device.

@muslum9ursex @unity It's a real build

@unity Unity remote mi kullanıyorsun? nasıl bu kadar iyi görüntü aktarılıyor?

@Artho340 @unity Changing code inside methods is huge

@unity What’s the difference between this and changing the value in the inspector? If we need to tag the function to be editable live, we still have to rebuild the game for this tag to be effective?

@unity I actually don't know what this means. Why... wouldn't you want to use the editor? How is typing a better interface than just... doing it? Is the multiuser stuff going to be in the editor too? THAT would be great.

We’re opening the Editor to EVERYONE 🤝 For 20 years, the Unity Editor is where a whole generation of games has taken shape. But a game isn’t built in the Editor alone. That’s why we’re building an API and set of services that open a Unity production to the tools, roles, and agents around it - no full Editor access required. Preview the real game, automate from the terminal, and use whatever tool fits the job. Now available in closed beta.

@David97717063 @mrdoob yeah, but you couldn't though. Code is like 10% of an actual finished successful game.

@mrdoob I don't think he hires many people. He got famous off an indie game 18 years ago, one which you could probably vibe code in a week now.

Can someone fire Jonathan please?

@SteveFr18760057 @OverfitForTruth @robertvaradan @ThePrimeagen *gah* clearly the argument is about games that compete with more complex fully human made succesfull games that are traditially made by teams of people, that people want to play. We know a 'game' is loose concept that could be made in 15 minutes. That's not the argument.

@Octamed @OverfitForTruth @robertvaradan @ThePrimeagen Video games do not need sound. They don't even need art, and it can be entirely programmed if needed. They don't necessarily need menus, levels, writing. They just have to have some interactivity and a basic game loop. That is a video game. It's front end can be a terminal.

its been a long time since i did one of these tweets, but i was reminded of this gem... where are all these video games? I WAS PROMISED BESPOKE, UNIQUE, VERY FUN VIDEO GAMES

@SteveFr18760057 @OverfitForTruth @robertvaradan @ThePrimeagen As in designing the game, playtesting it, changing it, play testing it, loop that a million times. Doing art, sound, music, menus, levels, writing. Let alone the whole thing of getting it playable for random people. It's A LOT.

@Octamed @OverfitForTruth @robertvaradan @ThePrimeagen Actually, a video game can be made with ONLY code. There is also no distinction that separates games made as "hobby" projects from games made as "whole" projects. They are both video games. The prediction was accurate.

@SteveFr18760057 @OverfitForTruth @robertvaradan @ThePrimeagen yeah, it's the exact same situation as any hobby project. Making a 'game' is really easy these days, even before AI. Coding though is only like 10% of a whole game project that actually exists in the real world for players.

@Octamed @OverfitForTruth @robertvaradan @ThePrimeagen The point was that people would be able to vibe code video games by the end of 2025. That's true. They can. They do. There are probably hundreds of thousands if not millions of playable vibe coded game demos just sitting on peoples personal hard drives.