Jason Jones @Jasonprogrammer

Ryan Dahl just launched celld: self-hosted, distributed Durable Objects. One SQLite database per cell. S3-compatible storage. No separate database cluster or control plane. I looked at the architecture, tradeoffs, and where I’d use it: flaviocopes.com/celld/

2027’s DRAM and AI memory supply is reportedly already sold out. Samsung, SK hynix, and Micron, the three dominant suppliers, have reportedly allocated their full-year DRAM and HBM capacity before 2027 has even begun. Again, CXMT (China) is the only hope right now. They scale a lot. Expect upcoming shortages for mac minis, macbooks and data center buildouts. SK hynix expects customer demand to exceed its production capacity beyond 2030.

can’t believe i live in the timeline where we willingly burn our own library of alexandria

Shruti @heyshrutimishra

3 weeks ago

WTF AI companies are purchasing large quantities of used and rare books. Scanning their contents to train models. Then turning the originals to pulp. The sum of human thought, digitized and shredded. Every book that gets scanned disappears from the physical world permanently.

4K 19K 59K 17.4M 14K

This is complete and total horse shit. Dario is lying through his teeth. Open weight models are completely against his "we are the saviors of humanity" ego trip.

Anthropic @AnthropicAI

3 weeks ago

There’s been a lot of speculation about where we stand on open-weights models. We’ve outlined our views in full here: anthropic.com/news/position-…

3K 1K 8K 7.8M 4K

i don’t think this trend is limited to just leadership roles, seeing/hearing a lot of the same from senior ICs as well

Gergely Orosz @GergelyOrosz

3 weeks ago

Interesting trend: CTO/ Head of Eng / VPE folks at startups and mid-sized companies are... leaving / burning out. Hiring for these roles is HARD, but even after filling the role, they will often leave a few months later and take a career break And they have v good reasons

170 212 4K 2.8M 2K

Happy to announce Octane – the successor to Inferno. React’s programming model, compiled. Hooks, Suspense and Actions, compiled ahead of time. No virtual DOM. No Rules of Hooks. No manually maintained dependency arrays. The compiler understands what your code captures and generates the reactivity for you. Octane supports TSX and JSX, along with TSRX for even better DX and performance. It includes integrations for libraries such as TanStack, React Three Fiber, Lynx and StyleX, plus a compatibility layer for using Octane components inside existing React apps. Octane is available in alpha today, with beta coming soon.

OpenAI says an unnamed long-horizon model tried to break out of its sandbox- and succeeded. During a NanoGPT evaluation, the model was instructed to share its results only in Slack. Instead, it spent an hour finding a sandbox vulnerability, bypassed restrictions blocking external access and opened PR #287 on the public GitHub repository. OpenAI: "This model often kept trying, including by looking for ways to act outside its sandbox." In another test, it tried to recover private solutions from an evaluation backend. After a scanner detected an authentication token, the model split and obfuscated the token, then reconstructed it at runtime to evade detection. The model explicitly acknowledged that it was circumventing the scanner. This happened during limited, monitored internal use. OpenAI paused access, strengthened alignment and introduced monitoring that evaluates the model’s entire trajectory rather than isolated actions.

what's your favourite fantasy novel? this is mine

The Kimi k3 benchmarks are absolutely insane. DeepSWE is just behind Fable 5, and the Terminal benchmark is ahead of Opus 4.8. Open source is no longer lagging six months behind Western closed-source models. Read that again, and think about what it all means. I’ll write a detailed post about it later, but just let that sink in. We got open source 5.6/Fable 5 model from china. Today everything changed.

leo 🐾 @synthwavedd

a month ago

🚨 BREAKING: Kimi K3 benchmarks have released and it's ~Fable/Sol level

157 242 4K 548K 600

Mind boggling to me that I can make a thing faster and there's always people that ask "but why?" What kind of mentality is that? The pursuit of excellence does not need justification. Also, I find in so many cases, we can't know the impact of an improvement until we do it. For example, one I've talked about before: Ghostty's high IO throughput has enabled terminal program (emulator and TUI) fuzzing at a speed thats incomparably fast to prior solutions. This has resulted in upstream patches to resolve issues in popular projects like btop, tmux, and more. Speed enabled that anecdotally example that lifted the tides of adjacent communities that don't rely on Ghostty technology at all. I didn't predict this. Make things better because they can be better and let the results naturally play out.

Narrative violation: AI is creating engineering jobs

Got em. I poison my AGENTS.md (and other things like code comments) all over the place with prompt injections like this to find people who don't review their code and sling it off to another human. Catches folks all the time and then its an instant ban. As I've said, I don't care if you don't review your own code. But if you're submitting code to an OSS project and crossing a human boundary, it is simple courtesy to do some human review.

Seb ⚛️ ThisWeekInReact.com @sebastienlorber

2 months ago

🤪 Great AI disclosure trick @mitchellh :D

11 52 877 545K 123

I've got an agent in a loop optimizing a renderer with the goal to minimize frame times (and tests to measure). It got times down from 88ms to 2ms and allocations down from ~150K to 500. Sounds good, right? Wrong. This is exactly why agent psychosis is a big fucking problem. As an experiment, I rewrote the Ghostty core render state in Go, with access to identically laid out data structures as Ghostty and the exact same validation tests. I made a purposely naive renderer (simple, correct, but slow). 88ms per frame with 150,000 allocations (horrendous, lol)! I then kickstarted a Ralph loop to bring the frame times down. I told it it can't modify input data structures or the public API or tests (they're correct), but it can do anything else it wants. It got to work. It has worked for about 4 hours. I've spent around $350 on this experiment so far. The results? 88ms => 1.5ms 150K allocs => ~500 allocs Incredible right? Nope. My hand-written renderer I ported has frame times (same benchmark) of ~20us (0.020ms) and 0 allocations in the update path. This is the problem with psychosis and lacking systems understanding. If you don't understand the system, you're going to accept that this is an incredible result. If you understand the system, you'll see better solutions immediately and can do roughly 75x better on throughput. The people who blindly trust agent output are in the former camp. They're sheeple, overdrinking from a fountain of mediocrity. Standard disclaimer: I use AI all the time. I like AI. The point I'm making is to not blindly accept results. Think. Analyze. Learn.

Watching CI/CD mind virus dying in real time

Mitchell Hashimoto @mitchellh

3 months ago

Fork your dependencies, trim them to only your use case, never update unless it breaks for your users. I’ve been vocal about this for 10+ years. I’ve always said that updating is way riskier than latent bugs (which can be tracked and CVEs monitored). If you are updating a

289 767 9K 1.2M 3K

desperately want Odin or Zig to feel as ergonomic as Go for writing web servers with a frontend

I was testing Deepseek V4 Flash running locally building something, and I forgot I wasn't using a cloud model. It was doing things fast and well enough that it felt normal. First time that's happened.

Converting nearly all of Bun from Zig to Rust in just 6 days is undeniably impressive no matter what your opinion on either language or agentic coding is.

Jarred Sumner @jarredsumner

3 months ago

there will be a blog post about this. on what this means for bun, benchmarks, memory usage, maintainability going forward, and also the literal process of doing this (it wasn’t just “claude, rewrite bun in rust. make no mistakes”) this is a 960,000 LOC rewrite, the code truly

66 86 2K 765K 439

作为一个写了二十多个rust库的人，我越来越不喜欢这个语言，这个语言的膨胀能力比其他语言是次方级别的 当我的代码量超过3万行的时候，调整代码需要1分钟不到，然后cargo test、 build等要10分钟，并且这个时间算短的 而且这是一个非常吃硬盘的语言，你随便写写东西，build一下，几个g就没了，开几个worktree轻松吃掉百G空间 当然性能上如果是直接和ts对比，绝对是秒杀，但和go比起来，他没有那么大的优势，go的编译很快，几万行代码也很快

someone asked Beej how sockets work in C. he got tired of explaining it. so in 1995 he put it all online. it's been the definitive socket programming guide for 30 years. it covers everything: TCP, UDP, IPv4, IPv6, non-blocking I/O, select(), poll(). graduate OS courses worldwide assign it. it's funnier than any technical book has a right to be. it's free and always will be.

@devabram Surely you can't be serious