I think that icons hold value so long as they have mostly distinct colors (which none of his examples do, so his point stands). At least for me, colors make vastly superior landmark than words do (once i know the interface).

I'm not sure what a typical ketamine dosage looks like, but I have discussed it with my psychiatrist as a fallback if TMS doesn't work out, and she said that ego death is one of the primary mechanisms (alongside reintegration therapy, spending a weekend tripping balls won't fix a thing - at least perpetually).

For humans the antidepressant dose is very low - I think less than 50mg. You might feel some minor dissociative effects at that dose, but only minor, yet in some patients it can apparently produce a pronounced antidepressant effect which is interesting and suggests a non NMDA mechanism

They mean to resell them in a different form: as part of their PaaS or SaaS. Per the article, OpenAI is just hoarding the wafers, not purchasing the final product.

Thanks for the AMA Peter!

Do you think there are risks involved with leaving (and hence returning to) the country on a Green Card?

I don't but a quick discussion with an immigration attorney to make sure that there are no hidden issues probably isn't a bad idea.

Thanks for the response, greatly appreciated!

> The students at America's elite universities are supposed to be the smartest, most promising young people in the country. And yet, shocking percentages of them are claiming academic accommodations designed for students with learning disabilities.

What the actual... Lack of journalistic integrity rears its head once again. Executive function and social challenges do not make a person "not smart."

Going back to the core of the problem, I feel that this does need to be controlled. It's one thing to disability signal online to gain clout, it's a completely different thing to drain resources from genuinely disabled folks. Disabilities need to come with diagnoses.

Maybe the smartest, most promising young people in the country realize it is smart to claim a disability.

I am not saying they don't have one. I am saying some people have realized ways it helps to point it out and maybe not everyone is clued into that.

Fundamentally I would be fine with this, the system exploits us so it's only fair to exploit it in return. Practically, however, my concern remains for people who need resources to support them.

These people are defacto exploiting others who are judged by tighter standards.

This attitude of it being OK to abuse some people because you feel like you're being abused needs to go. It's #$@%#@ childish.

They are taking the gun out of USA's hand and unloading it, figuratively speaking. With this strategy they don't have the compete at full competency with the US, because everyone else will with cheaper models. If a cheaper model can do it, then why fork out for Opus?

It's sans-io at the language level, I like the concept.

So I did a bit of research into how this works in Zig under the hood, in terms of compilation.

First things first, Zig does compile async fns to a state machine: https://github.com/ziglang/zig/issues/23446

The compiler decides at compile time which color to compile the function as (potentially both). That's a neat idea, but... https://github.com/ziglang/zig/issues/23367

> It would be checked illegal behavior to make an indirect call through a pointer to a restricted function type when the value of that pointer is not in the set of possible callees that were analyzed during compilation.

That's... a pretty nasty trade-off. Object safety in Rust is really annoying for async, and this smells a lot like it. The main difference is that it's vaguely late-bound in a magical way; you might get an unexpected runtime error and - even worse - potentially not have the tools to force the compiler to add a fn to the set of callees.

I still think sans-io at the language level might be the future, but this isn't a complete solution. Maybe we should be simply compiling all fns to state machines (with the Rust polling implementation detail, a sans-io interface could be used to make such functions trivially sync - just do the syscall and return a completed future).

> First things first, Zig does compile async fns to a state machine: https://github.com/ziglang/zig/issues/23446

Maybe I'm missing something, but that's still a proposal, which also assumes an implementation for the other proposal you linked and that also doesn't exist yet.

For now I would refrain from commenting on non-existing functionality.

> I still think sans-io at the language level might be the future, but this isn't a complete solution.

I'm not sure what about this is really at the language level (only stackless coroutines appear to require language level support, and it's still unclear if it's really possible to implement them). However I do agree that a sans-io, or at least dependency injection for I/O is a great improvement on the library side, and it's something I'd like to see in Rust too.

> I still think sans-io at the language level might be the future, but this isn't a complete solution. Maybe we should be simply compiling all fns to state machines (with the Rust polling implementation detail, a sans-io interface could be used to make such functions trivially sync - just do the syscall and return a completed future).

Can you be more specific what is missing in sans-io with explicit state machine for static and dynamic analysis would not be a complete solution? Serializing the state machine sounds excellent for static and dynamic analysis. I'd guess the debugging infrastructure for optimization passes and run-time debugging are missing or is there more?

Exactly the caveat that they themselves disclose: some scenarios are too dynamic for static analysis.

I wouldn't define it as Sans-IO if you take an IO argument and block/wait on reading/writing, whether that be via threads or an event loop.

Sans-IO the IO is _outside_ completely. No read/write at all.

Oof, you're completely right. I'm not sure where I got that wire crossed.

I coincidentally watched BasicallyHomeless's video on his 100+ day Linux experiment and he made a really good point: because everything on Linux can be done with the CLI, it also has a working natural language interface (Claude Code). He ran into several issues, such as sound (allegedly that's no surprise, but not my experience), and Claude fixed them all.

If it doesn't wipe your drive.

Still, interesting thought.

Yeah, setting up my router with VLANs/Firewall/NAT etc was so damn frustrating with Ubiquiti compared to the Mikrotik router I had before.

While I could just export my config file with Mikrotik and ask ChatGPT to make whatever changes I wanted in seconds ("here's my config, make a vlan 20 with all my iot devices") and get a fully working config back, with Ubiquiti you just get a bunch of inaccurate "click here and there" instructions back instead since the UI changes slightly all the time.

The switchover was still worth it, as the Ubiquity UI is nicer in daily use (and Mikrotik wifi sucks ass, so I had to use other APs). However, every time I want to change something I wish I had an easily ediable config file to edit, and get LLM help with, instead of a confusing UI to click around in.

With OpenWRT, it's likely even easier :)

Indeed, large language models have much easier time working with a real written language.

I wonder if the modern GUI conventions could be reliably translated to machine-understandable text representation, operated on, and then mapped back to the GUI picture.

Right, it could be a lack of competition in the direction of the reactor. It's a giant petri dish for anything able to withstand radiation.

AI is currently summoning market activity during what should be a recession. Now this bubble isn't only filled with hot air, it's also filled with sewage.

Short-term gains are great, though!