on the vibecoded adobe suite i think both the effort and the expectations are completely misplaced. illustrator is 39 years old, photoshop 36. people really did expect to catch up in two weeks. attacking the full suite makes no sense. if you’ve been using photoshop, indesign, or illustrator for 20 years, you’d have opinions. you’d notice even the smallest inconveniences and have ideas for how to improve your workflows. with ai, you’re finally in a position to make bespoke software that serves you perfectly and brings you joy. look at aseprite, which specializes in pixel art. losslesscut, which does one thing premiere makes unnecessarily complicated. photopea, which started as a browser-based psd viewer and grew into a serious editor over 14 years. figma, which rethought design around collaboration. each of these projects has a point of view. each understands something about how people actually work. instead, we get seven adobe-shaped applications, recreating the surface without understanding what led to the underlying decisions. not an alternative vision for creative software, just a checklist of features to imitate. it’s slopware, and it trivializes decades of genuine attempts to free people from adobe. the tragedy is that we finally have the tools to make the software we actually want, and instead we’re making the software we already have. and if you needed any more evidence that this was never about making better software, having finished 40 years of adobe development in two weeks, he’s now moving on to microsoft office.

"You are not allowed to use this interface". More and more "decentralized finance" sites greet us with exactly this. So we built a public dashboard: 64 official DeFi interfaces, rated by what they block and for whom. Go see who locks you out 👇

Vibecoding with no understanding and learning just in time is the default for all major codebases School brainwashed you into thinking you need to learn something first completely before doing it. No, you just do it and learn along the way. If you get stuck learn more. Everything is figureoutable. I like the term code archaeology but I prefer exploring the code labyrinth. You never know what you will find

crimson 0.14.5 is out watching a replay? hit enter to take over from that moment and play it yourself. retry from the same spot as often as you like. the desktop build has joined ranked survival and quests on leaderboard

💾 Two years ago I started working on these crazy nostalgia projects, kind of randomly, because I was bored and wanted to do something different than just build another SaaS or AI app/website I wanted to try to run Windows 3.1 inside a web browser, and to do that properly already took me months When that worked well enough, I wanted to build a dial up internet connection so it could browse the web, with the basic levels of AI back then, that was VERY hard, and I'd just go around in circles forever. Finally I had to find the only person in the world who got that working @bai0 and beg him to help me make it work, and we did it! Then I started hosting classic games like Quake with full multiplayer, porting them to WASM, or if not already like Quake III by @lukathedev and hosting those with a full multiplayer server ready-to-go in your browser And then I spent 6 months trying to reverse engineer Windows XP to run on v86 by @copy_sh, very difficult (because it's so big) and only possible because modern AI like Opus 5.5 is good enough to figure that out For the last two years pretty much nobody cared about any of these projects, I'd tweet them into the void, which was fine because I thought it was very fun just for myself But now with this wave of AI-assisted reverse engineering WASM ports of games and apps to web they're finally taking off, pretty much anything is portable to WASM on web now, and if you add server to it via WebSockets, people can instantly play with each other, or interact inside apps you host, it's super cool Every game and app I have on @pieter now is full or almost full to the limit, yesterday 80,000 people played around in Windows 3.1 or Windows XP inside their browsers chatting to MSN and AOL Messenger AI bots and playing 3D Pinball or playing Quake etc. And that was my idea for @pieter: a non-commercial preservation project, kinda like an online computer museum but interactive! The most cool thing to pay respects to these old games, apps and OSes I think is to keep them alive by making them easily and freely accessible, playable on mobile so old and new generations of people can jump back in the past interactively This is one of my first projects to take off that does not, can not and will not make money because it's non-profit, and that's fine Every time I tweeted about these projects would always ask me "but why?" Because it's really fun! 🤩

not a comment on this particular project, but i keep noticing llms recommending heavily seo-optimized projects whose promises are far ahead of what they actually deliver. they sound great until you look closer. exhausting having to constantly separate substance from hype.

i was avoiding codex cloud because i thought it would be cumbersome to set up the environment. little did i know that it could figure it all out on its own just by looking at the repo. for some reason it has never crossed my mind.

Glovebox is live! play Half-Life: Alyx standalone on Apple Vision Pro 👉 all you need: 🥽Vision Pro (M2 or M5) 🎮PSVR2 controllers 🖥️any Mac (for installation only) 🕹️own the game on Steam 🍎free Apple account

btw the site now has instant replay seeking. i optimized it really well. it works by pre-simulating the run, which just takes a second since the sim is 900x realtime, then it also notes when the decals appear. in the game decals keep piling up during the run (shells, corpses, blood, explosions), so you can’t easily take them back. it makes state checkpoints, or keyframes if you wish, as well as more sparse ground checkpoints. then when you seek it reads the closest previous ones, does a little bit a simulation and re-applies the missing decals since the last snapshot. it all works really fast, scrubbing anywhere takes 2-5ms, and you can get the full replay compressed to just tens or a hundred kilobytes since it’s just inputs which is enough for deterministic simulation.

I have a games industry background, where everything needs to fit within 16.67ms to hit 60fps That mindset has carried over into other work So now @Nethermind processes Ethereum mainnet blocks in 16ms at p50, on standard EIP-7870 hardware