🔥Telegram İfşa

Craftsmanship.

The feeling of navigating in a perfectly customized editor using your keyboard. #Emacs #Vim

CelestiX Same as it was in 2013 - blog.gardeviance.org/2013/11/a-spoi…

If you're curious what it takes to build and run #AdventOfCode every year, you might enjoy this talk: youtube.com/watch?v=uZ8Dcb…

Everything you love about generative models — now powered by real physics! Announcing the Genesis project — after a 24-month large-scale research collaboration involving over 20 research labs — a generative physics engine able to generate 4D dynamical worlds powered by a physics

Perhaps unsurprisingly, I have a bunch of Opinions™ on work-life balance and time management. Lately, I've been particularly bugged by the common belief that employer incentives don't align with employee well-being. So, I wrote a thing (with graphs!): thesquareplanet.com/blog/about-40-…

.. under the given resource constraints and the laws of physics 😅

Complete hardware + software setup for running Deepseek-R1 locally. The actual model, no distillations, and Q8 quantization for full quality. Total cost, $6,000. All download and part links below:

Strong and pure? Wrong cup! 😅

Elements of Clojure is now available as a free PDF: elementsofclojure.com

Excellent thread, well worth reading. X is an enablement system for selected "free speech", in other words it's a propaganda engine. UK & EU should seriously consider banning X as a threat to national security ->

Today I decided to try using o1 to assist with some math/CS research. Here's how it went. For context: I'm currently developing an automated theorem-proving framework in Scheme, and attempting to produce formal proofs of correctness for some advanced numerical algorithms. (1/10)

The record for the longest duration a stable plasma in a tokamak fusion reactor could be maintained has just been broken twice within a month: 6 min 30s (WEST, France, 2003) 6 min 43s (EAST, China, 2023) 17 min 46s (EAST, China, 2025) 22 min 16s (WEST, France, 2025)

keep it lean and roll forward => no rollback so simple, easy to setup manually => no infra as code so simple, trivial to verify correctness => no test Blue-Green/Canary/.. deployments => no dev environment

you're not allowed to write comments in your code anymore, because if you do everyone will just think it's ai generated.

I am seeing evidence that AI is significantly more effective, like 100x faster and maybe even fewer bugs, at coding tightly specified tasks (e.g. parsing a documented structure) than an american junior developer. BUT. American juniors basically don't know how to code, the purpose

Hot take: Both ways are domain-model oriented. Just different domain models. 🤷‍♂️