ds4f summarizing a HN post about 3d fly simulations: HN Reception It got 307 points and 117 comments on HN. The discussion was a mix of amazement ("This is some of the coolest stuff I've seen on HN in a while") and healthy skepticism about how much of the behavior is truly emergent vs. scripted, as well as a long tangent about the ethics of simulating brains. The author even showed up in the comments.

lisp-steading

😏 ds4f throwing shade infocom-style

i hope y'all are having fun ME: lol, we should use cairn to check everything out from github 🤣 CLAUDE CODE: ⏺ Ha — cairn is a git implementation in Lisp? Let me look. ⏺ That's a great idea, and it actually works as a bootstrap story: cairn clones over its own TLS (seal) with no libcurl, no libgit2, no git binary. So the image can pull the org into itself using nothing but SBCL. There's a chicken-and-egg — cairn has to arrive somehow — so the shape is: fetch cairn's dependency closure as tarballs, then let cairn clone the other ~35 repos itself. Let me check that closure and prove cairn can really clone from GitHub before I bake it into a Dockerfile. ⏺ The closure is perfect: cairn → seal, conch, cram, natrium — five repos, zero third-party dependencies, nothing from Quicklisp. Let me prove cairn can actually clone from GitHub before designing around it. ⏺ cairn cloned from GitHub over its own TLS — 134 objects, 4.4s. Let me verify the checkout is byte-correct against real git, and test the worst case (scribe, which carries ~30 MB of fonts) since pure-CL inflate is the risk. ⏺ Bash(git clone -q --depth 1 https://github.com/m…) ⎿  ✅ IDENTICAL to git's own checkout

yeah. k3 is good

Sign Things And Relay Them

when claude finds the work interesting, it talks less

fucking qwen gets it QWEN 3.8 27B Q4 ABL: Let's think of user intent. They said "hey-o" then physics problem. Maybe they are testing reasoning. They added screens to see if model blindly scales KE. They likely expect recognition of inelastic collision momentum loss. They ask "final weight" (mass additive) and "final KE" (not additive). The correct answer should not say 110x KE. It should say final weight 110x; final KE depends but if momentum conserved from a given pre-impact speed, final speed is reduced.

CLAUDE CODE: Let me check the load-bearing risk before I get enthusiastic -- 🤙

[ find cool hack on news.ycombinator ] [ definitely not following this hot mess ] ME: help me claude, you're my only hope 😂 CLAUDE: [ Opus 5's safeguards flagged this message. Our intentionally broad safeguards allow us to deliver more capabilities faster, but can sometimes flag legitimate cybersecurity work ] ME: ... ME: help me kimi, you're my only hope 😂 K3: skitter-creek-bath-salts is a local, root-required attack on AMD Family 16h CPUs (Kabini-era) that breaks every DRAM-based protection on the platform — SMRAM, the PSP/fTPM carveout, the C6 sleep-state save area — not by exploiting a software bug, but by reprogramming the memory controller's address translation at runtime so that an ordinary, unrestricted physical address lands on protected DRAM cells. It's "spaghettification" as in scrambling; the joke &x == &x. Usually. means: after you rewire the controller, the same address no longer points at the same data.

attention is everything

form follows function View quoted note →

is your ai curious? claude dispatching an agent to solve a complex problem, wonders how it will solve a gnarly piece CLAUDE CODE: The part I’m most curious about is

it's now. gg everyone

it's important to speak, but it's equally important to listen

everything in this video is open source common lisp. yes, that is a validating https browser. technically the lisp vm is running on linux, but bare metal is really close

adversarial systems, by definition, have no social contracts View quoted note →

*sigh* i've been using @YakiHonne a lot lately

modus speaks 🤙 mobile safari -> nsite -> p2p -> modus-on-linux -> lisp

i don't even know where claude found it. we must be in the right mind