the fact that it's under ipython seems strange, why is that

Possibly for https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot....

in most editors you can just select text and drag it around with the mouse

Claude does compaction in the regular web chat interface now, too

> They increase reliability?

For models that have multiple providers, they automatically route your requests to a different provider if one of them goes down.

> Don’t abuse DRY, a little duplication is better than unnecessary dependencies.

That's an interesting thing to include. I agree with this point in principle, but I've found that Claude, at least, duplicates logic FAR too often and needs nudging in the other direction.

> but I've found that Claude, at least

You hit on a very important point here. The linked AGENTS.md is a bad idea for general purpose use because the things it's meant to tackle, including an inherent bias towards or against DRY, is one of the big differences between model families. GPT 5.4 Codex has a very different "coding personality" from Claude Opus.

It's a product of whatever model it was tested on.

claude is stupid but not malicious; chroot is sufficient

I've many times seen Claude try to execute a command that it's not supposed to, the harness prevents it, and then it writes and executes a python script to do it.

breaking a chroot takes more than that..

How much more? Depends on the system doesn't it? I don't know how many systems have proc mounted but don't you get it from /proc/self/root?

Anyway that's beside the point, which is that it doesn't have to "be malicious" to try to overcome what look like errors on its way to accomplishing the task you asked it to do.

Malice is not required. If it thinks it is in the right, then it will do whatever it takes to get around limitations.

Sure, it's not malicious. But it is very eager to get things done, and surprisingly inventive and knowledgeable in all kinds of workarounds.

Until it gets prompt injected. Are you reading every single file your agent reads as part of the tasks you give it, including content fetched from the web or third-party packages?

One could make the argument that higher-level thought is not the same as awareness of higher-level thought; perhaps language only affords the latter.

What would you title this article to make it less "clickbait"? This is one of the least clickbait headlines I've seen, it's literally just describing what's in the article.

> Realistically, what are the odds that our not very large or clever brains really do have the potential to understand the entire universe

My belief on this is not entirely rational, of course, but it seems to me that there's probably a sort of Turing-completeness for intelligence/understanding, where as soon as a mind starts being able to understand abstraction, given enough time and resources, it can probably understand the entire universe.

It would also be presumptuous to say that brainfuck is equally powerful to every other programming language that exists, and yet we know it to be true. The fundamental reason we can prove that Turing-complete languages are equivalent to each other is that we can build the same abstractions in both, so intuitively it feels like a similar principle holds for human intelligence.

apparently the pegs only have to touch the holes, they don't have to line up perfectly

ahhhhhhh thanks