A comment from our recent YouTube video:

We are just trying to get reviews from content creators so we can add their review on the landing page... can anyone help us? Currently we are stuck with these 2:

Hey [CREATOR_NAME]! Been following your stuff for a while and thought you’d be a perfect fit for this. We made Trolley — a competitive trolley-problem game where you make morally questionable decisions, defend them, and get matched against other players. Would love for you to try it and see how badly it exposes your ethical framework: trolley.typememetics.institu… @ThePrimeagen @dhh @theo @TomCruise @mattpocockuk not teej @unclebobmartin @mc_swm @wigger @Aella_Girl @GoldmanSachs @AnaKasparian @TrisH0x2A @zeeg @tobi @kanyewest @signulll @0xSero @beffjezos

The Institute presents you the best game made in Poland ever: Trolley. A typesafe moral catastrophe where every death is data and every argument becomes part of the memeplex. Convince AGI to pull the lever, climb the leaderboard, and discover whether your ethics survive contact with other people's priors. trolley.typememetics.institu…

"Loosening restrictions" is when a language designer removes the only type-level guarantees preventing your program from becoming an unprincipled pile of partial functions, nulls, implicit coercions, unchecked effects, and runtime surprises because programmers complained that the compiler was making them think too much. Now making illegal states representable is called pragmatism because Hindley–Milner and referential transparency weren't generating enough adoption. Take the dynamic-typing apologia somewhere else.

We are advising @dhh to use nixos over arch and rewrite all his company's ruby code to rust.

The Institute operatives coming to NixCon:

Counterfactual Compiler History presents: "Why everyone hates Apple"

Fuck those golang guys in particular.

I am never taking the Tube again. I have been harassed by homeless people multiple times and mugged twice, but this is too much.

The Institute would be delighted to extend an invitation to both @DHH and @cafkafk to join us at NixCon in Kraków, September 25–28. Despite the differences in philosophy, culture, and technical approach between Omarchy and NixOS, there is far more that unites these communities than divides them: a shared belief that Linux can be better, more coherent, more accessible, and more empowering to the people who use it. We would welcome the opportunity to bring these perspectives together in Kraków, not to erase our differences, but to understand them, challenge one another constructively, and perhaps discover common ground. If joining us in person is not possible, we would be equally glad to host both of you for a Discord call during the conference, so that the conversation can still take place in front of, and together with, the wider community. If Omarchy brings ambition for a more approachable Linux desktop, and NixOS brings rigor, reproducibility, and a profoundly typesafe way of thinking about systems, there may be something genuinely interesting to build at the intersection. Let Kraków be neutral ground. Let us bridge the divide, exchange ideas, and work toward a stronger, safer, more typesafe, and more user-friendly Linux ecosystem. Truth in Types, Safety in Thought.

I resigned from Anthropic today. I spent the last three years doing pretraining research at both OpenAI and Anthropic. Neither company is acting responsibly. They are racing straight toward automated theorem proving, and soon poor mathematicians will lose their shitty jobs. More thoughts below.

Beware of the java enterprise architect types.

TYPEMEMETICS IS A SWEET AL FAN PAGE NOW.

I returned to the development environment that gave me Golang instead of Haskell to conduct some investigative work. Here is what we know. I opened the project on Sunday, Aug 30th at 2:06 pm. I took roughly 15 keystrokes to reach the language selection screen (see image below). A dropdown cheerfully greeted me. 'Select programming language.' 'Haskell' I selected, matching its purity. The menu expanded with mastery. Python. JavaScript. Haskell. Golang. Rust. TypeScript. C++. I moved my cursor toward Haskell. Then, unexpectedly, a Slack notification appeared on my second monitor. This took me by surprise. The system had deviated from the script and routine. What was I to make of this? I looked away for approximately 1.7 seconds, returned my attention to the dropdown, clicked and continued. The project initialized successfully. No compiler error. No type-level complaint. Nothing. I assumed everything was fine. Before starting work, I put on my lab coat. This was to avoid contaminating the functional core with ambient side effects. I'd brought with me 8 oz of freshly warmed category theory. Before opening the editor, I had heated it to 145 abstractions per minute, near the ceiling before practical software engineering denatures and the entire thing destabilizes. This would be my companion over the next few hours on this splendid Sunday afternoon. Then the coding began. It was glorious. At 5:43 pm I messaged a friend 'the compiler is insanely fast, I'm still going strong'. Earlier in the day I'd told them my plan was to complete 10 hours of deep, focused Haskell work. It was Sunday, my calendar was open and it was a unique opportunity to work in an interruption free stretch. Recently I'd been dipping with fatigue in the 2-4 pm window. Likely because I've been waking up at 4 am and completing an intense five hour morning routine of category theory, theorem proving, explaining monads to people who didn't ask and refusing to perform IO without wrapping it in sufficient abstraction. The routine did work. Turns out, I also made the mistake of attributing my extraordinary development velocity to Haskell. Fatal. I finished work and began my wind down routine. I opened the repository at 8:01 pm and only then did I realize I had a problem. Whereas normally I'd spend 2-3 hours convincing the compiler that my program could exist, something was off. My code was compiling immediately. There was no way I was writing Haskell. I looked at the repository. go.mod Not having written Go in years, I stared at the screen, creating a subjective feeling that I'd been programming imperatively all afternoon. Was this a mistake, a prank or a crime? UI noise yesterday, I returned to the development environment with a screen recorder the workstation contained approximately 14 simultaneous visual distractions Haskell and Golang appeared within the same dropdown I replicated my cursor movement pointer travel between the two options was approximately 31 px my mouse operates at 1600 DPI this created a very low intention to cursor ratio you can retain the broad semantic category 'programming language' but fine grained type system discrimination drops off even with healthy static typing between 1,000-4,000 abstractions 'Haskell' and 'Golang' could become nearly identical Selection mistakes I took a picture of the language selector (see image above) Haskell and Golang are adjacent both options are large enough that a finger mishap is unlikely the failure point here is a 'capture error' both options are programming languages therefore typography and menu size do nothing to offset autopilot error programming language selection error rate is around 1 in every 8 greenfield projects afternoon mistakes (when I was there) most common there is also a possibility that the development environment changed the language, or pranked me Recognizability and prank given my social media presence, there is a ~55% likelihood that the development environment recognized me it could have thought 'lol, this is going to be funny' and initialized a Go module Given this, what do you think happened?

I took a look at @dhh's new pfp, and now my @Dell screen has his face burned into it? WTH

Unfortunately, we can't continue working on ffmpeg-rs for the next four days. See you in September.

There are two types of Frenchmen.