esclerofilo

I can't pretend to know how things work internally, but I would expect it to be involved in model updates.

Every time claude code runs tests or builds after a change, it's collecting training data.

From the Readme

I made some progress with hyprland using a set of Guile bindings I developed called hypripc, but I found that Hyprland isn’t as stable as Sway.

The HN crowd probably knows that understanding maths concepts is a very rewarding experience. It is also very useful for $Nth grade, because it allows you to apply the principles to different problems than the ones you've seen solved.

However, if you try teaching a "bad at maths" kid how to solve a quadratic equation, explaining how to factorize and why it works is a bad strategy. They _prefer_ using the formula IME. They don't want to understand the concepts, they want to pass their exams, and rote memorizing the formula is a faster/more reliable way to do it.

I'll admit that maybe I'm just a bad teacher. After all, I'm not a teacher, I'm a CS grad who has done some teaching.

But think of the average class, with 40 students. The teacher needs all of them to learn to perform some tasks. He'll choose the one-size-fits all solution, even if it's less beautiful, less motivating for the kids that find maths to be fun.

I'm not saying we should go back to having schools for "gifted kids". They have well-documented problems, and I'm not at all qualified to pick one side of the tradeoff. All I'm saying is that boring biology classes are there for a reason, not just teachers without passion.

Wanting someone who will stay for the long haul makes sense, but it leaves me as an unemployed engineer with two options:

- Stay unemployed until I find a job that seems interesting enough to work there for years. Apply, get rejected because companies don't like hiring unemployed people.

- Get a job that I don't like, then be constantly on the lookout for an opportunity to leave. Get rejected because I haven't been on my job for long enough.

What should I do?

> Is VSCode really all that "controlled by Microsoft"?

Yep. VSCode "core" is open source, but many of the official extensions aren't (Python, C#, Live Share, Dev Containers IIRC). Besides, accessing the extension marketplace from a non-official vscode build is against their terms of service, meaning forks like vscodium don't have easy access to most extensions.

> hard to imagine an "AI girlfriend" that could be similarly short-term and bring positive changes.

Feeling lonely has negative consequences, there's a famous study comparing it to smoking cigarettes. Lonely people can get fixated on that feeling. Mitigating it may help people look for other hobbies, that they kept putting off because they were so focused on their need of a girlfriend.

Hell, a realistic enough "woman AI" (not "girlfriend AI") could make you lose the fear of flirting, be useful as practice, to have more confidence when you have a real date.

> the most profitable way to run such a product is not to help, it's to eat up as much user time as possible

If you charge per-month (like OP), it's more profitable is to eat up less time.

Talent and experience are not the same thing, and I would say experience is a lot more correlated to demand of talent than to supply of it. More demand means more opportunities to get experience.

I believe that's not what they're saying. It's signing hardware, like a camera that signs every picture you take, so not even you can tamper with it without invalidating that signature. Naively, then, a signed picture would be proof that it was a real picture taken of a real thing. What GP is saying is that people would inevitably get the keys from the cameras, and then the whole thing would be pointless.

They already listed foveated rendering in the features (which I believe is what you're describing). It use the graphics performance budget efficiently, but it can't physically add more pixels.

It's really cool technology anyway, and according to PSVR2 reviews, it seems to work well.

Highway inevitability is a fallacy. They could've built a railway.

> the workers coned off the bike path

Why didn't the workers cone off a lane from the street to make space for the bikes? Cars usually have easier alternatives, they can just take another lane.

EDIT: But anyway, the real viable alternative to cars is probably not bikes, it's trains and buses.