Easily prevented by doing a few hundred unmanned tests at full depth for progressively longer periods of time, and using a portable CT scanner to see if there are any signs of hull damage after each descent.

I mean, for real companies, the free market does sort it out, because regardless of the payout for the deaths and whether it bankrupts the company, nobody is going to ride in a submarine knowing that the company's last design collapsed on the first try.

The problem is the existence of concepts like shell companies and the corporate veil. Most people don't like the idea of rich billionaires being able to create products and services that kill people without meaningfully getting punished for killing people, but the corporate veil is strong, so there's a real chance that the punishment would bankrupt the company, but the person who set up that company could be almost completely unaffected financially.

Worse, the company that goes bankrupt could be a shell company that's deliberately designed to fail, at which point the larger company that owns that shell company and all of its IP rights could then move on to a similar project with a similar shell company under a different name, and go on to kill again. Think of it as the murdering version of what Chinese companies with random 5- to 7-letter gibberish names do on Amazon when they get too many bad reviews, and you'll understand the problem.

Now imagine importing cars under similar conditions. Car catches fire and burns your family alive? Your one remaining living relative leaves a negative review, people stop buying from that company, and MIXFLIP motors goes under, and MIXFITZ motors is born, and has only five-star reviews, until the next family is toasted, and MIXFITZ dies and GENFLIP spins up. And because of jurisdictional boundaries, there's no accountability.

And this is why we have safety laws, and this is why companies going out of their way to avoid being regulated is so dangerous to everyone. The market can only work things out if there is actual accountability for bad enough failures, and corporate law is designed to limit accountability in ways that could easily turn them into mass murdering machines in the absence of regulation.

By forcing products that could be dangerous to undergo certain levels of testing and certification before they can be sold or used in the U.S., you ensure that the cost of entering the market is high enough to make those shell company tricks infeasible, thus ensuring that there's only one name for the company when it sells in this market, and that if they screw up badly enough, they'll genuinely be destroyed by the market.

Doesn't matter. The absolute worst case heat situation should be reentry. Apollo's ablative heat shield is only three inches thick. Putting a three-inch ball of phenolic epoxy resin, wrapped around a half an inch of lead, wrapped around something the size of a golf ball is well within the realm of what can be done.

you're crazy.

Remember this: all AI knows comes from stolen data. Why would you interactions with it be treated any differently?

his Synchron implant isn't owned by a Neo Nazi, being that Synchron is a competitor to Neuralink.

This alone is worth something.

Steering things in the wrong direction, apparently — solar panels, the country, etc.

I don't think you realize how little radioactive material we're talking about here. 1 kilogram of U-235 would power a 100 kW reactor for more than two decades, if my math is right. That's about the size of a golf ball. You're telling me you don't think they can put enough lead around a golf-ball-sized chunk of uranium to ensure that it doesn't end up exposing anyone if the ship explodes during launch?

Hitler's was Germania. Trump's is a nuclear moon base.

And of course, on top of that, the orange utan needs all the distractions he can throw at his idiot MAGA base to make them forget the Epstein file.

Once again, the environmentalist fringe has set standards so high that they are impossible to meet so that they can berate folks for not meeting them.

Meanwhile, PLA 3D printer output can be trivially mechanically shredded and extruded into new filament several times. It's hard to say that PLA isn't recyclable with a straight face.

+1. In my home town, the pay phone by the high school was used for exactly two things: calling parents to pick kids up after away games and calling in fake bomb threats(*) to get out of tests. I would expect similar behavior from public phones today, sadly, minus the kids calling their parents part.

* When I was a freshman, this is what the seniors told me people had done in previous years. I cannot corroborate the story with any actual evidence. Also notable: this was in the early 90s, before school shootings and bombings were really a thing.

Lots of things are pointers under the hood. But that's really irrelevant to the point.

Yeah, EVERYTHING is implemented at the base level in assembler, so pointers are in use everywhere. And I learned assembler first. But if that's your idea of where one should start, someone else can say we need to start with transistor theory, with just as valid an argument.

Pointers aren't required for most purposes. They're often just an optimization, frequently a questionable optimization. It's true that C pretty much requires pointers, but in C++ references can generally be substituted with greater clarity. Pointers are almost never used in Java (are they ever?), and certainly not in Python. Or many other languages I could name. (Yeah, they still exist "under the hood", but that's not the point of an exam of early or intermediate programming skill.) For that matter check out D https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fdlang.org%2F . That's a language that would be my favorite if they had a better way to document your code (last I checked Doxygen didn't do a good job) and it it had a slightly better library. (As it is I currently prefer C++ except for stuff that's heavy in unicode, where I'll switch to Python.)

C is actually easier than Java, though doing anything complex is harder.

A C++ subset doesn't have to be any harder than Java, and can be at least as capable. (The problem would be documenting it...and for students, learning to use only the subset.)

But it costs 100x as much, if not more. Running an LLM can be done on a notebook these days. But training one requires an entire data center of expensive GPUs. Not to mention that the notebook will run a reduced (quantized) version. Go check huggingface how large the full models are.

And also, LLMs are still suffering from a number of issues. For example, on many non-trivial tasks, the LLM is still unable to follow simple instructions. If you use LLMs routinely, you likely found cases where it has zeroed in on one - wrong - answer and no amount of prompting can convince it to give you a different one. It'll even totally ignore very clear and explicit prompts to not give that same answer again.

A human will understand "if you give that answer again, you're fired". An LLM... well you can tell it that it'll get shot between the eyes if it repeats that once more and it'll tell you where to get help if you have suicidal thoughts.

These things are both amazing and amazingly dumb at the same time.

that it should not be used for important stuff.