I see you've never been in one.

Google and Waymo have explicitly and repeatedly said that they are fully liable for accidents caused by their vehicles.

You're the one claiming it's well-known, but you are unable to provide a single link.

Figures.

Worth it for what? If the answer is a different financial motivation, then it's just a different form of corruption.

NONE of them have done this, not in 250 years. Not until Trump. Presidents make a lot of money after they leave office, yes, especially in recent decades. They command incredible speaking fees (Obama famously got $400k for one speech), they write books that make millions, etc. There are lots of people who are willing to give them money to get close to them, get their advice, etc.

But until Trump, none of them has used the office of the president to make money. No one else has sold pardons. No one else has set up a public stock and a memecoin to facilitate billions of dollars in bribes. No one else has brazenly sold insider secrets to stock traders.

Trump's corruption isn't just new in scale, it's a completely new thing that the United States of America has never before seen. We've seen plenty of corruption elsewhere in government, at all levels, but never in the oval office.

Right, the land rightfully belongs to whoever owned it second-to-last! Not to the people the second-to-last owners forcibly ejected from it, of course, and not to the people who own it now, only to the second-to-last.

</sarcasm>

I think we should have more respect than we do for the indigenous peoples, but pretending that there was no war and no forcible land grabs before the Europeans arrived is not just ahistorical but ignorant of human nature. Polynesian history, which includes the Hawaiians, is incredibly bloody. The Europeans dominated when they arrived not because they were more violent -- there's a good case to be made that they were less, though it's a question of degrees, not kind -- but because the Enlightenment gave them much better tools. And, of course, because centuries of city living had bred up a bunch of diseases, but that's part of the same story.

Did European arms and power make it "right" that the natives were conquered? No. But it wasn't "right" that the natives conquered each other, either. Humanity -- also as a result of the Enlightenment and the long, slow development of humanism -- has become more moral and less violent, and we now have more respect for other cultures and other peoples. And that's a good thing! But it doesn't make sense to project our own moral views back in time, and it definitely doesn't make sense to do so selectively, demonizing the winners and lionizing the losers.

The only thing that really makes sense is to apply our improved (though still very imperfect) morality to the situation we have now, messy, complicated and conflicted as it is. It's nice to give the native Hawaiian culture some respect, but it's wrong to give it more respect than that of Hawaii today.

Thatâ(TM)s the problem software developers arenâ(TM)t seeing: most software for internal facing employees only needs to be good enough. If I can drop in a bunch of loose files and data connections and get even half way to a better analysis and answer to a business question, then this paid for itself. The vast majority of IT departments do three things: keep outlook running, keep excel running, and keep windows running. I know thatâ(TM)s an oversimplification of IT work, but for some reason us techies have created an illusion that the only companies out there are all in the tech space. Case in point: if the company you work for doesnâ(TM)t make and sell software (e.g., they make widgets), go tell your CEO you think the company should stop paying for Excel and youâ(TM)re going to hire an army of developers to properly code every stupid spreadsheet. Donâ(TM)t be surprised when they hit you with that door on your ass on your way out. All these AIs are doing are making better spreadsheets and presentations. But guess whatâ(TM)s happening in the business side that makes the decisions about your salary: thatâ(TM)s all theyâ(TM)re doing too. Yeah, you got some analysts that are busting their assess making live dashboards that only a few people use, but even then that data gets exported to Excel and manipulated to tell the story that needs to be told to make a business decision. Should we build this new building or enter into a long term lease. Should we spend X on marketing on Y platform vs Z platform? Should we hire 60 more people in the factory or buy a couple robots and an engineer or two where the capEx is spread out over 10 years? These types of questions are exactly what MS CEO is alluding to. Heâ(TM)s not selling to you experienced coders, heâ(TM)s selling to the people who actually write the checks. What heâ(TM)s selling is the fact that an army of IT departments arenâ(TM)t needed anymore because you only need a couple people to run Azure and copilot will connect youâ(TM)re data, help you analyze it, and present it to someone else to make a business decision. A similar project in the past would require 5-20 people and take weeks if not months to deliver that well thought out, maybe properly specâ(TM)d and documented code, that can run forever that will get two glances before the business shifts directions.

Let me put it this way: youâ(TM)re delusional if you think that an MBA in a suit wonâ(TM)t put you right on your ass the first chance they get the minute they can get âoegood enoughâ for their IT needs. AI is inching them closer to that point, and there are way more non-Tech companies than there are Tech companies run by those suits.

To be clear, the language would have to change if you wanted it to outlaw corporate surveillance, because the existing one applies only to government. And I suppose the new one would also have to explain "corporate" and how that's different from other kinds of organizations or surveillance by individuals.

Everyone has to make their own decisions. I would discourage people from believing that SLTS kernels will be anything like secure by the end of their lifecycle. If that doesn't matter for some application, then fine. If it does matter then arguing that the SLTS kernel must be secure because recertifying is expensive is... not really an argument.

FWIW, in my day job I'm making this same point to people building cars with an expected 20-year lifespan. I think the right conclusion is "Don't use Linux". If you need long-term assurance, use something smaller, simpler, with much less attack surface. And then buy an ironclad long-term support contract for that.

It's thankless, yes. Necessary, or even a good idea? I think that's debatable. Actually LTS is probably a good idea... but I'm very skeptical of SLTS. I don't think CIP is going to be successful at keeping 10 year-old kernel versions secure. If you need 10 years, you should either plan to do the work necessary to upgrade to new kernel versions periodically or you should assume that it's just going to be insecure and isolate it or otherwise shield it from attack.

You're confusing the breaker sizing with the wire sizing. Code requires the breaker to be about 80% of the safe capacity of the wiring to maintain this safety margin. Breakers allow the rated capacity, continuously -- and even allow a little more for short periods.

Standard US outlets are 15A @ 120V. But, yeah, based on his numbers it looks like he was only getting a little over 8A. I charged my car on a US outlet for a few years and I got more like 5-6 miles (8-9.5 km) her hour of charging, which is pretty much what the math predicts for 15A @ 120V.

First a nit about units: You mean "you get about 1 kW".

"kW/hr" doesn't make sense as a unit because watts are an instantaneous measure; time is already divided out. "1 kWh/hour" technically makes sense but is kind of pointless because it's just "1kW * 1 hour / 1 hour". Also, a standard US outlet is 15A @ 120V = 1.8 kW, not 1 kW. There's a little loss so you get more like 1.7 kW into the battery.

I've also been driving EVs since 2011 (two Nissan Leafs, three Teslas), and I get more like 5-6 miles per hour, which is consistent with the math: 3 miles/kWh = 333 Wh/mile. 1.7 kW / 333 Wh/mile = 5.1 mi / hr. 4 miles/kWh = 250 Wh/mile. 1.7 kW / 250 Wh/mile = 6.8 mi/hr.

Maybe your mileage is worse than you think? Were you not getting 15A from your outlet? Your numbers fit if for some reason you were only getting about 8A out of your wall outlet, when you should have gotten nearly twice that much.

More like 50A @ 240V = 12 kW. My car will only take 48A, but my voltage is usually closer about 245 which the math says is 11.76 kW, and I normally get about 11.2 kW into the car.

Note that at 9 kW, 27 mi/hr = 333 Wh/mi, and 36 mi/hr = 250 Wh/mi, so your numbers line up there.

That's really not what's going on here.

What's going on is exactly what they're saying. OpenAI didn't realize their AI hacked HuggingFace until HuggingFace told them. That sparked a broad investigation in both Anthropic and OpanAI, scouring their logs to see if there were other cases, and they're finding a lot of them, and doing the responsible thing by disclosing.

Both companies are really worried.

Having an out-of-control AI is an anti-selling point, not a selling point.

I am an American, have been for all of my 57 years. I'm actually a certified pistol, rifle and shotgun instructor, a certified range safety officer, and a concealed carry instructor. And while I wouldn't recommend it, it's not at all hard to get away with shooting outside of one of those situations.

Also, it was a joke.