> We turned on crash reporting on the way.

I haven't burst out laughing like this in a while! You'll probably make for some horror stories to a poor Mozilla team.

These changes baffle me. It's not even enshittification because I cannot see any benefit to YouTube at all.

> NEVER include in commit messages or PR descriptions:

> [...]

> - The phrase "Claude Code" or any mention that you are an AI

EDIT: I just realized this might be used without publishing the changes, for internal evaluation only as you mentioned. That would be a lot better.

And some of the entries are too short and will create false positives. It'll match the word "offset" ("ffs"), for example. EDIT: no it won't, I missed the \b. Still sounds weird to me.

https://github.com/chatgptprojects/claude-code/blob/642c7f94...

I like the tradeoff offered: full access to the current directory, read-only access to the rest, copy-on-write for the home directory. With stricter modes to (presumably) protect against data exfiltration too. It really feels like it should be the default for agent systems.

This might not be the right frequency, though, and only accepting post/fax is bullshit. Doubly so for short deadlines.

And the thesis linked above seems to go beyond simply "use a trusted compiler to compile the next compiler". It involves deterministic compilation and comparing outputs, for example.

I think they started banning unauthorized API users around the time that "WhatsApp For Business" was introduced, because it was competing with that product. Unfortunately WhatsApp For Business is geared toward physical products and services with registered companies, so home automation and agents are left with no options.

I wonder how people are using them in a way that is not distracting to the main content. I've found that high-frequency patterns (small details with sharp transitions) can be a bit distracting, but I haven't found a good solution that doesn't compromise the beauty of the backgrounds.

We can debate whether that's real intelligence, and whether the question is fair, but this is still a real, measurable capability, that just eight years ago was a pipe dream. This capability is what OP is tracking, and what I believe is impressive but hamstrung by harnesses.

On a different note, LLMs are still not very wise, as displayed by all the prompt attacks and occasional inane responses like walking to the car wash.

And search engines are narrow tools that can only output copies of its dataset. An LLM is capable of surprisingly novel output, even if the exact level of creativity is heavily debated.

But LLMs are hamstrung by their harnesses. They are doing the equivalent of providing technical support via phone call: little to no context, and limited to a bidirectional stream of words (tokens). The best agent harnesses have the equivalent of vision-impairment accessibility interfaces, and even those are still subpar.

Heck, giving LLMs time to think was once a groundbreaking idea. Yesterday I saw Claude Code editing a file using shell redirects! It's barbaric.

I expect future improvements to come from harness improvements, especially around sub agents/context rollbacks (to work around the non-linear cost of context) and LLM-aligned "accessibility tools". That, or more synthetic training data.