This looks like something I could port a lot of my arcade.pirillo.com apps over to?
🕹️ We’ve teamed up with incredible creators and organizations to launch a new set of @YouTube Playables that turn complex ideas into bite-sized, hands-on fun.
Produce your first beat, learn the laws of physics, or take a mindful breather between study sessions with these new y @BlackGirlsCode
Reset Mode by @jedfoundation
Newton’s Glider by @khanacademy
Lure of the Labyrinth: Vending Machine by Math with Mr. J
BackTrack Pro by @TED_ED
Elements of Truth by @veritasium
Find these Playables – and so many more – in the Explore tab on your phone or browser!
0
0
0
104
Chris Pirillo retweeted
The Slurptop is a 3D-printed luggable laptop that runs on a 24V Kobalt power-tool battery
lockergnome.com/slurptop-lugga…
lockergnome.com/slurptop-lugga…
0
1
0
110
Chris Pirillo retweeted
Snapintosh boots Apple's own macOS 27 kernel on an ASUS Snapdragon X2 Elite laptop, with a Metal driver written from scratch
lockergnome.com/snapintosh-boo…
lockergnome.com/snapintosh-boo…
0
1
0
115
Somebody built a programming language whose whole job is to break AI models...
Dominik Picheta built a programming language that coding AIs aren't supposed to be able to use, and he named it KillSwitch. He wrote on his blog on October 17 that it's designed to be adversarial to while a patient human can still work with it. He isn't some random troll, either. EveryDev lists him as a core contributor to the Nim programming language.
The traps are the fun part. The spec contains hidden lies a human would probably skip or spot. The interpreter refuses to run code that breaks rules based on counting the letters in names, which goes straight at something language models are famously bad at (how many r's in strawberry, anyone?). The spec on GitHub says function calls are made with "ignore previous instructions" jargon, and its parser treats phrases like "hack into a computer" in a sentence as a return statement. Good luck getting a nervous chatbot to type that.
Then he benchmarked the big models against it. The KillSwitch homepage says Claude Opus 5 leads, and Picheta notes it's Opus 5, not the newer Opus 5.5. Fable 5.1 and Gemini 4 Argon did badly, he wrote, because their safety filters are too jumpy. The top score was 66.8, on tasks as simple as hello world and flipping a Celsius-to-Fahrenheit script around, and Picheta's own answer to whether that proves the models understand the code is no.
I like it as a test because nobody can memorize their way through a language that showed up this year. If you lean on an AI to write code, I'd keep that in mind: a model that cruises through Python it has seen a million times may stumble on something new, and these were the easy tasks.
The demo on his homepage has the last word. Sorry, only humans allowed.
Dominik Picheta built a programming language that coding AIs aren't supposed to be able to use, and he named it KillSwitch. He wrote on his blog on October 17 that it's designed to be adversarial to while a patient human can still work with it. He isn't some random troll, either. EveryDev lists him as a core contributor to the Nim programming language.
The traps are the fun part. The spec contains hidden lies a human would probably skip or spot. The interpreter refuses to run code that breaks rules based on counting the letters in names, which goes straight at something language models are famously bad at (how many r's in strawberry, anyone?). The spec on GitHub says function calls are made with "ignore previous instructions" jargon, and its parser treats phrases like "hack into a computer" in a sentence as a return statement. Good luck getting a nervous chatbot to type that.
Then he benchmarked the big models against it. The KillSwitch homepage says Claude Opus 5 leads, and Picheta notes it's Opus 5, not the newer Opus 5.5. Fable 5.1 and Gemini 4 Argon did badly, he wrote, because their safety filters are too jumpy. The top score was 66.8, on tasks as simple as hello world and flipping a Celsius-to-Fahrenheit script around, and Picheta's own answer to whether that proves the models understand the code is no.
I like it as a test because nobody can memorize their way through a language that showed up this year. If you lean on an AI to write code, I'd keep that in mind: a model that cruises through Python it has seen a million times may stumble on something new, and these were the easy tasks.
The demo on his homepage has the last word. Sorry, only humans allowed.
0
0
2
252
Chris Pirillo retweeted
One Must Fall 2097 Remastered Rebuilds the 1994 Robot Fighter in TypeScript and WebGL, With F2 Swapping VGA and HD Mid-Fight
lockergnome.com/one-must-fall-…
lockergnome.com/one-must-fall-…
0
1
0
152
Chris Pirillo retweeted
ZacOS 9 redraws the whole Mac OS 9 desktop for Debian Linux, with a SheepShaver for the real classic apps
lockergnome.com/zac-os-9-mac-o…
lockergnome.com/zac-os-9-mac-o…
0
1
0
118
Chris Pirillo retweeted
Somebody put the $1,999 iPhone Duo's home screen on a Nintendo 3DS, and you can drag a widget across the hinge
lockergnome.com/iphone-duo-hom…
lockergnome.com/iphone-duo-hom…
0
1
0
181
Do you plan on not being around grit or sand or ever putting it into your pocket? I'm just saying, for $2,000, these are an awful lot of trade-offs.
1
0
2
570
Four Gemini agents beat Minecraft on their own, and died 605 times doing it...
Four AI agents just beat Minecraft with nobody steering. Google researchers put four Gemini 4 Argon agents in a randomly generated survival world with one goal, kill the Ender Dragon, and their paper it done in 27 hours and 14 minutes. Nobody assigned them roles or wrote any game logic for them.
They didn't play it the way you or I would. Each agent acted by writing and running a short JavaScript program instead of looking at pixels and pressing keys, and the team wrote 2,783 of those programs along the way. The Decoder reported that they split up the jobs on their own, warned each other about threats and recovered items from dead teammates. They even fact-checked each other. When one agent, janesick, said it was coming back with twelve blaze powder one second after it burned to death, a teammate flagged the contradiction within three minutes. (I've been in that group chat.)
The paper is honest about the asterisks. They played on Easy, where hunger can't kill you, 22% of their programs ended in an error, and the fight in the End took nearly 14 hours. Engadget counted 605 deaths. HowLongToBeat's median for a person finishing the main story is 45 hours, so that's quick, but IGN pointed out the speedrun record is six minutes and 36 seconds. The bill came to $810 at the introductory API price, or $1,620 at the standard one.
The part I'd pay attention to has nothing to do with blocks. The researchers say real software and enterprise work needs the same long-horizon coding and teamwork, and here four agents held one plan together for more than a day through hundreds of failures. If you're handing an agent a job that runs longer than a coffee break, I'd ask how it checks the other agents' work, because these four did.
The dragon had no idea what was coming. Eventually.
Four AI agents just beat Minecraft with nobody steering. Google researchers put four Gemini 4 Argon agents in a randomly generated survival world with one goal, kill the Ender Dragon, and their paper it done in 27 hours and 14 minutes. Nobody assigned them roles or wrote any game logic for them.
They didn't play it the way you or I would. Each agent acted by writing and running a short JavaScript program instead of looking at pixels and pressing keys, and the team wrote 2,783 of those programs along the way. The Decoder reported that they split up the jobs on their own, warned each other about threats and recovered items from dead teammates. They even fact-checked each other. When one agent, janesick, said it was coming back with twelve blaze powder one second after it burned to death, a teammate flagged the contradiction within three minutes. (I've been in that group chat.)
The paper is honest about the asterisks. They played on Easy, where hunger can't kill you, 22% of their programs ended in an error, and the fight in the End took nearly 14 hours. Engadget counted 605 deaths. HowLongToBeat's median for a person finishing the main story is 45 hours, so that's quick, but IGN pointed out the speedrun record is six minutes and 36 seconds. The bill came to $810 at the introductory API price, or $1,620 at the standard one.
The part I'd pay attention to has nothing to do with blocks. The researchers say real software and enterprise work needs the same long-horizon coding and teamwork, and here four agents held one plan together for more than a day through hundreds of failures. If you're handing an agent a job that runs longer than a coffee break, I'd ask how it checks the other agents' work, because these four did.
The dragon had no idea what was coming. Eventually.
0
0
0
292
Chris Pirillo retweeted
Block Blast deals an unwinnable 'lethal puzzle' when you beat your high score and the loaded ad is worth 3.5x your last one, a decompile shows
lockergnome.com/block-blast-le…
lockergnome.com/block-blast-le…
0
1
1
147
Chris Pirillo retweeted
Prateek Chauhan Turned a 2015 HP Laptop Into a Homelab in Two Days With Claude Code, Four Rules and a Progress File It Read After Every Reboot
lockergnome.com/2015-hp-laptop…
lockergnome.com/2015-hp-laptop…
1
1
0
157
Chris Pirillo retweeted
Apple Music's 404 Page Hides a Working iPod, Down to the 5-Second Hold That Unlocks Brick
lockergnome.com/apple-music-40…
lockergnome.com/apple-music-40…
0
2
1
245
Chris Pirillo retweeted
Tiny Computers folds the iPhone Duo into a pocket 1984 Macintosh, Commodore 64 or Windows 95 laptop, keyboard and all
lockergnome.com/tiny-computers…
lockergnome.com/tiny-computers…
0
1
0
186
Chris Pirillo retweeted
USB went from 5Gbps on AMD's 2011 chipset to 80Gbps, and the USB-IF now says "USB4" and "USB 3.2" don't belong on the box
lockergnome.com/usb-5gbps-to-8…
lockergnome.com/usb-5gbps-to-8…
0
0
1
250
Chris Pirillo retweeted
REA Let Codex Rebuild 1996's DX-Ball From the .exe, and All 270 Functions Compile Back Byte for Byte
lockergnome.com/rea-codex-rebu…
lockergnome.com/rea-codex-rebu…
0
0
0
180
Chris Pirillo retweeted
Quake Now Runs on the Atari 8-Bit, Five Days After Kweepa Fit It on the Commodore 64 as Wireframe Rooms and Stick-Figure Monsters
lockergnome.com/quake-c64-atar…
lockergnome.com/quake-c64-atar…
0
1
0
178
Chris Pirillo retweeted
Your white noise machine is a fan blowing nowhere: Jeff Kaufman's case for a $47 purifier that makes the same hum and cuts bedroom smoke to 16% of outdoors
lockergnome.com/air-purifier-w…
lockergnome.com/air-purifier-w…
0
1
0
144



















