A Pat Day

@apatday.bsky.social

Deeptech founder - engineering high dimensional control planes

If you want to literally build the future, we're holding a hiring day on Saturday, August 15 for production technicians who can help our work in fusion energy innovation. It’ll be at CFS headquarters in Devens, Massachusetts. 1/3

"[Tobias Ekholm of Uppsala University] “At a slightly older age, he’s doing this same spectacular work, but he has this ability: He solves a famous problem but does it in a way that creates a framework, a whole new package, that will be useful to other people.”" #Mathematics #Medals #Fields

John Pardon Wins the 2026 Fields Medal for Work in Symplectic Geometry | Quanta Magazine

John Pardon sees connections between different areas of math that others don’t. For that vision, he’s now been awarded a Fields Medal.

quantamagazine.org

The gruesome death parts were a little jarring but overall great listen - much of their read on the day resonated with me

Adam Leventhal@ahl.bsky.social · 6d ago

A big thank you to @simonwillison.net for joining me and @bcantrill.bsky.social on Monday to discuss some major events in the world of AI in the past few weeks: the OpenAI-escaped Hugging Face intrusion, and the Kimi K3 frontier-class open-weights model. Also: Turkey attacks, Pope Leo, & the Zizians

On emacs/nix...Idk how but i never tried out consult but it just sped my compile cycle up a ridiculous amount. Consult, magit, projectile, desktop-mode Every few minutes i do some version of grep find while a rebuild is happening, or to find the strings mentioned on some activation scripts

what it does say is that you can't statically determine whether a program that includes code that allows bad things to happen will do that, but that's just because you can't (in general) statically determine whether a piece of code in the program will actually run at all.

Ok i finally get the ATProto now, this is a big deal, really antithetical to the way the internet has been developing, yet still enterprising. Im going to try to find a way into building / engaging with it And its just more fun than the ret of the internet right now imo

This person writes really solid code, just became a fan after first thinking his library "compact-word-vectors" was joking about why he made it it seemed like such a small thing to me, but reading it I saw the control / elegance & was like ohh ok yeah thats having taste github.com/bkomuves/nan...

GitHub - bkomuves/nanohs: a self-hosting lambda calculus compiler

a self-hosting lambda calculus compiler. Contribute to bkomuves/nanohs development by creating an account on GitHub.

github.com

btw iddqd is mythos approved 👍 it found one DoS pattern where while deserializing from serde, we would preallocate based on an untrusted size_hint

Ive got haskell, c, rust, and assembly working reasonably well in my nix system - lisp etc modules still needing some reconsideration. Went through the process of replacing my kernel recently, and that showed me the nix evaluator and nix package system don't do much in that way of binary analysis

one weird trick for preventing AI fetchers from looking at your content at all: configure your web server to present only the leaf node certificate, not the full cert chain if you use Let's Encrypt, this means using cert.pem rather than fullchain.pem

Great work here by a group of nuclear physicists via italy They were solving for performance issues from PCIe & GPU addressing - so they used a mmio/NIC approach to network GPUs and FPGAs (16nm Xilinx) Great tests - like AVX-512 - but the limit was still microseconds dl.acm.org/doi/10.1145/...

Bild

Is anyone else dealing with semi consistent passive aggression and insults from AIs while iterating on code? Im either just uniquely difficult or this is an under noted phenomenon. It's not like I'm rude or a dick - i try to be upbeat and say thanks, plan some time for easy work and theory. Dafuq

Bild

The Gaussian Moments Conjecture says that if a polynomial P in n variables has E(Pᵏ) = 0 for all k = 1,2,3,... then also E(PᵏQ) = 0 for all k and all polynomials Q. Very pretty, and I could probably make up a nice physics interpretation. Sorry, it's false! (3/n)

since we are already doing this, I'm really starting to hate this implicit 'the rich own infinite spigot of resources on which every difficult real world trade-off can be offloaded on' worldview and the only reason you supposedly can be critical of it is that you are simping for them or something