That’s why you put those devices in a separate VLAN with no routing access to the rest of your network

The fact that this reponse has 1 downvote is *chef’s kiss*

Agreed.

If you can do your work on a 13" laptop with no mouse or external monitor without your productivity dropping off a cliff, you were never productive to begin with.

Decades ago I ran an “rm -fr *” as root, I thought that I was ~/bin, but I was in /bin. That was a fun lesson.

While true, it still means you’re locked into only being able to log in from a browser that has the password manager extension installed and logged in. Sometimes I want to log in from another machine, or another OS, or another browser, or even an incognito window that doesn’t have access to my extensions.

I want to like yaml, I really do, but why are there so many different ways of specifying the same thing?

It’s an MoE (Mixture of Experts) approach. An 80B-A3B model has 80B parameters total, so that dictates the size of the model and the VRAM+RAM you need to have to hold it, but only 3B of those parameters are active at any given time. This reduces the intelligence of the model compared to an 80B dense model, but improves the speed. In the end it’s the size of an 80B model, with the intelligence of a ~40B model, that runs at the speed of a 3B model.

Pretty much all state of the art models either have already, or are in the process of switching to an MoE design, since it significantly reduces the hardware required to run big models at usable speeds. You can often get usable speeds on MoEs without a GPU at all.

https://en.wikipedia.org/wiki/Electric_current

The conventional direction of current, also known as conventional current,[10][11] is arbitrarily defined as the direction in which positive charges flow.

Current is defined as the flow of positive charge. The fact that electrons, which are negatively charged, actually flow the opposite direction is irrelevant. The diagrams are still correct per the definition of current.

Like a baby’s arm holding an apple

The second half is a little rough, but the first half is fantastic IMO

Nah I’m on that guy’s side. His experience lines up with my own, namely that vibe coding is not useful for people who don’t know how to program, but it can be useful for people who do know how to program, and simply aren’t familiar with the specific syntax used in a language they’re not an expert in.

In that case, the queries to the AI model aren’t, “write me a program that can do X”, it’s more like “write me a function in this language that can take A, B, and C as inputs, do operation Y with them, and return Z”, or “what’s the best way to find all of the unique elements in an array and sort it alphabetically in this language”. Then the programmer can take those pieces and build up a proper application with them. The AI isn’t actually writing the program for you, it’s more like a customized Stack Overflow generator, without having to wade through a decade of people arguing back and forth in the comments about inane bullshit.

Does it save a ton of time? No, but it’s still helpful, and can get you up and running in a new language much faster than the alternative.

Context Switching

It’s why I hate when middle managers get a hold of my time allocation. “You have 8 hours a day, so you can spend 1 hour each on these 8 different projects and move them all forward together!” Sprinkle 3-4 pointless meetings throughout the day, and then they wonder why nothing gets done.

Kicad should be great, but they’ve made a number of insane UX decisions that make it really unusable in practice. Horizon is actually based on the pretty good Kicad engine but it fixes most of the UX mess.

Do you have an example? I use KiCAD pretty regularly, and while they do have some odd defaults for a lot of their tools and keybindings, I rarely run into one that can’t be changed/fixed in the settings.