Jay Weeldreyer

every organization makes decisions based on reality assumptions that largely go untested.

@nicolekcha @d4m1n My brain was reading it as "YO CLASSIFIED INTERNAL." Seemed more fitting for some reason

@d4m1n YC CLASSIFIED INTERNAL: DO NOT SAHRE

people will lose their jobs because of this 💀 today is one of worst day in software, the gstack internal architecture was leaked via a test suite that was accidentally bundled in the root stylesheet. make sure to load in 4k: - 3.23 trillion LoC (incl lockfiles) - 1.3 million agent skills - 6.9k+ timezone implementations 👉 it lays down the blueprint for AGI, but runs at 120fps on a raspberry pi 2 talked to some experts. many confirmed it's even beyond that. possibly even further.

@FactoryAI Languages that retain control of critical structures and systems of power, stubbornly embedded. Totally out of sync with the modern world. Unchanging, unchangeable. Utter disregard for the generations that follow. "Legacy" is too kind. These are Boomer languages.

No major benchmark is designed for COBOL, Fortran, or Assembly - the languages powering trillions in transactions and infrastructure that must be modernized or risk catastrophic failure. We built Legacy-Bench to measure frontier agents on the code the world actually runs on.

@0xSero @alexinexxx 😆

@alexinexxx For cuda Claude For coding GPT For frontend Kimi For unhealthy relationships grok

still haven’t figured out what my AI stack should be

@Gentleman_Ways These standards are worth *extending*. The work is never finished.

There was a time when how a man spoke, dressed, and carried himself mattered. This is The Ways of a Gentleman. I write about bringing those standards back through culture, etiquette, and timeless values. Give me a follow if you believe these standards are worth preserving.

@bitsko_xt @Fried_rice Interesting. why not?

@jayweeldreyer @Fried_rice Ah yes, thanks. I wasn't impressed with terminal bench.

Claude code source code has been leaked via a map file in their npm registry! Code: …a8527898604c1bbb12468b1581d95e.r2.dev/src.zip

@echo_fox_bravo @Fried_rice You're probably right.

@jayweeldreyer @Fried_rice No, you muppet, obviously not.

@truth_hur1s @Fried_rice You're probably right.

@jayweeldreyer @Fried_rice No, it’s just you stuck in a troll cave.

@chriskapilla @Fried_rice You're probably right.

@jayweeldreyer @Fried_rice I see you are a member of the "if it's popular I can (try to) flex by dumping on it school"

@AnandButani @Fried_rice If you're using it for technical purposes (which I assume you are), this is from Terminal Bench:

@jayweeldreyer @Fried_rice What harness are better?

@bitsko_xt @Fried_rice Check out terminal bench tbench.ai/leaderboard/te…

@jayweeldreyer @Fried_rice please elaborate

@stevemur This fucking state . . .

No Kings, but we are taking away your historic right to elect the chief law enforcement officer for your county

@iruletheworldmo For high uncertainty/ambiguity situations, human orchestration with the old school cut/paste plus some choice commentary to nudge/clarify around nuance, along with sparse initial prompts . . . man, what you can actually get from these models is incredible.

having a ‘council’ of llms is strangely effective in improving the output. different models have different strengths, using this well let’s you harness them all. this article perfectly sums that up and gives you a neat skill bookmark it. and yes. thank me later.

@Adrian173075707 @morganlinton @EnoReyes @0xSero It's nuts how much of a difference a harness makes. I keep sharing this reality, and people keep thinking only the model matters.

@morganlinton @EnoReyes @0xSero Good choice. The models perform way better than inside codex (gpt 5.4) or claude code Opus 4.6). And their compaction is dark magic.

So it begins. Thanks to @0xSero for putting this on my radar.

@iruletheworldmo If true, we're nearly at that moment in the timeline when the AI-enabled breakaway civilization/evolutionary branch begins.

🚨BREAKING FRONTIER MODEL NEWS claude mythos set for release april 16th dario has more leaks than the titanic, here’s some info from anthropic staff. >95 or higher on every single benchmark. except arc agi 3, yet to be tested on. >dramatically outperforms opus 4.6 on coding, reasoning, and cyber >anthropic privately warning government officials about its capabilities >so powerful they’re calling it “unprecedented cybersecurity risk” >already being tested with early access customers >priced at $120/$600 per million tokens >10 million token context window >enterprise use only capybara is here.​​​​​​​​​​​​​​​​ capygpt is agi.

@Elijah_Meeks I look forward to digging in/checking it out! Thanks for sharing.

@jayweeldreyer The latest version of the dataviz library I've been working on for the last 10 years. Streaming-first, AI-first, networks, geo, connected scatterplots... semiotic.nteract.io

These are literally are the prompts I used to make it. I did, to my credit, make Semiotic really good at working with AI to make data visualization... with AI.

@Elijah_Meeks @claudeai I for one wholeheartedly embrace the sort of play that AI now affords people. There's nothing wrong with things done for pleasure/curiousity/fun alone. The world of data, engineering, etc now has an entirely new route to do so, and it's really enjoyable to see things like this.

I don't know what it mean anymore that I can just spin up something like this in 15 minutes by asking @claudeai to integrate Semiotic and Pretext using the wikipedia streaming API. This would have been such an extraordinary thing but now it's just... Tuesday.

@Babygravy9 I recently discovered that the Chinese have both a massive research capacity with interests well beyond technology, engineering, manufacturing, and that it is wholly unconstrainted/uncontaminated by that which ails the thinking in the West. It's a rich resource for the curious

My friend K9 posts some of the best content about the ongoing decline of South Africa, right from the frontlines. If you care not just about South Africa but the future of the West, you should follow him, promote his content and, just as importantly, pay attention.

@shine_gomiseifu Japan is fucking BASED

左翼は弱者じゃありません。 差別を盾に悪さをしたい犯罪者たちです。 奴らの声に耳を貸す必要はない