MarceColl

I understand that, but why can't the train self-limit to the 40kmh that the law requires when the security systems are down.

My question is, why is driving by sight with speed limited (40 kmh afaik) not something that is enforced by the train itself when there is no connection to the other controls systems. How is that not a hard limit instead of something the train operator needs to remember?

Oh, you dont have to worry. You'll be obsoleted waaay before him.

Reminds me a bit (although less interesting) of the 2000s JK Rowling website.

I disagree they are fundamentally at odds. The design I'm exploring with capabilities means that if you want you can do all of those things, but you can also scope them. An actor can create another actor and pass the capability or restrict it further. Imagine something like codemode with capabilities, where you want your agent to be able to shop for you. You ask your OS agent, it creates a couple different actors, one maybe manages your money, with limits you can set. Only that actor can do that. Another gets a restricted HTTP client only for those shops you whitelist.

Then a third is an agentic actor that gets access to those other two. That one then has access to two things:

* HTTP requests to specific websites * A limited way to use money

Since you can compose the primitives to limit the impact of any agent doing stupid shit you can trust it more to shop for you or book travel for you.

Of course it's more friction, but you can also choose how involved are you. You can potentially trust your coding agent to do most of it and once done everything that can be deterministic is deterministic, and only agents come in for the non-determinism and adaptability.

As the system evolve you have a library of actors that act as restricted capabilities, and an agent cannot create them out of thin air.

IMO, an agentic operating system should be not only about "integrating the agent everywhere" but also about providing the compartimentalization and tools to allow working safely with agents when they can reach into every part of the system. It should allow to easily extend the system safely.

I also built an agentic OS but the underlying system is very different based on a custom actor runtime with capabilities and a way to arbitrarily create and compose restricted capabilities.

The idea is that you can sculpt the OS to your needs using agents and fully restrict what they can or can't do.

The core is also I think kind of interesting because at it's core it's an actor runtime, where each program receives and sends messages, and only when processing a message they use memory. Each program has managed persistent state as well. So it's very easy to create actors that persist data and you can have millions of them without using a lot of resources.

I have been experimenting with agents inside the application itself as it's being developed. The application has a harness and then that harness can self-evolve and evolve/introspect the application.

The EUV LLC research was US-funded, agreed. But the light source is Cymer, which ASML has owned since 2013. Its drive laser is Trumpf (Germany), and the optics are Zeiss (Germany). Nikon and Canon had access to the same public research and dropped EUV. The US had GCA, Perkin-Elmer and SVG, and lost the market anyway. If integration were the easy part, someone else would have done it. You are underselling what it means to perfect the machines by calling it "just integrating".

By that logic SpaceX is a German company. Liquid-fuel rocketry was researched and developed at Peenemünde, and the US brought in von Braun's team wholesale to build its program. SpaceX just integrates German IP. Or Google is Swiss. The web was invented at CERN; Google is just an integrator of European tech. Or Apple Silicon is British. ARM was designed in Cambridge; Apple just licenses it.

That's why it has so many competitors, right?

Fair enough, but I think there is utility beyond this. Sometimes humans do stupid shit too.

I have no relation to the project, but calling this problem solved and then offering a vastly inferior solution to the proposal (PITR) is a bit unfair

It's a similar methodology, my problem with things like this is that they do give you comprehensible input, but it's very boring. I don't engage at all. The idea of Katarineko is to create comprehensible input from content that interests you. Say you want to watch Spirited Away but it's a bit too advanced for you at the moment, we create stories at your level for Spirited Away to teach you the core words you need to achieve your target comprehension level.

Yeah, that's bullshit.

Depending on my mental state, status with the person and many other factors each of them may trigger both many different internal thoughts, looks, body expressions and even outcomes.