Ben Carson

Power Word: "put on your i-have-good-theory-of-mind hat"

Inanna drops the best threads, just banger LLM whisperer stuff.

Spent too much time creating the preview image for: foryou.club/atmosphere I pulled the current profile pics of the Bluesky accounts for each of the apps that the cross-lexicon For You feed shows. Not sure what's the deal with @pckt.blog Can you spot an Easter egg?

New voice model, this one was trained on only speech supposedly! It sounds really realistic, and will apparently be open weight! And was trained on just a single nvl72! block.xyz/dessa

The thing that will personally turn me into The Joker will be the cheap screws included in installation kits.

I'm absolutely thrilled to have written this article for @longnow.org. It's about AI and time: how we got here, and what we can do now, when all of the pace layers are disrupted at once by an influx of new minds. longnow.org/ideas/attent...

Good thing the company who fatally mismanaged generation of their intermediate and root CAs can fall back on their core business of providing PKI management and trust services.

“You Can See Everything” would have been a better name for a security problem where a trusted root was pwned allowing trivial MITM.

The macOS built-in curl, git, pip, python3, perl, ruby, and openssl all read /etc/ssl/cert.pem, which still has the emSign root, but they don't check revocation like Apple's Secure Transport libraries. This is, uh, a problem. git and pip especially. I added a short-term workaround to this page.

Inspired by a question from my son, I had Claude put together something I've always wondered - how long does it take for a supercomputer to make it to your home?

She started her super amazing company, healthcare technology. Acc accuracy actually accuracy. Anyway I should say sorry.

Go to your camera roll and search “boat” and post one that you like.

"We have never seen a model do this before, but GLM-5.3 sometimes reasons (unprompted) about how its current actions could become training data and impact its future self" x.com/ApolloResear... www.apolloresearch.ai/science/upda...

Guys this is like my superbowl I waited so long for US open weight models to be relevant again. Before today the US models were still behind Chinese models from as early as Feb. 2026. Still more work to do! This cut the gap by at least half, in time (say 3 months)

An interesting paper-as-blogpost from Nick Cammarata at Goodfire. He thinks neuron polysemanticity and superposition are less common than thought. This is good news because it makes the purpose of individual neurons clear, and networks more readable.

Can confirm. Source: I am a business juice / pipe person.

"hmm, a 3.7 TB .exe, wonder what happens if I double click on it

Ah yup there’s your problem. You’ve got a distributed gemmaswarm in your walls. They burrow in there pretty good if you’ve got any compute they can dig into. We can try to flush ‘em out with a GLM but if you don’t pull out the nodes they’ll just be back later…

We're hiring across many roles at Trillium Labs to build the open scientific ecosystem for LLMs. Most important hires right now (ideally with start dates soon): * Full time post-training researchers * Engineers with experience with popular open-source libraries...

At least as far as charming animations, I think my Claude Number might be 5.5.

This goes surprisingly hard youtu.be/3TNpOD6bov8?...

Power Word: §

Tag yourself

Ha ha, the answer was 15k lines of a baroque accounting system to meter usage of Python inside of Blender.

"The correction copy is verified. The worker is implementing paid pre-import authentication and removing unmetered filesystem probes." Time to find out what flight of fancy is occurring in this codebase.

Please stop calling HTML served over HTTP “The Web”. Computers do not have spinnerets and are incapable of producing silk.

Oh yeah. It slaps. Calling all mineheads! games.mino.mobi/orb