Appropriateness of the response to the emergency is part of the legal considerations. Congress granted the power for a reason. Taking that and assuming it means arbitrary power is not operating within the law, not for Trump, not for Biden.

In what way? Please correct me.

Bah. If a human driver increased their safety and reliability by avoiding certain situations, would you call them a worse driver for it?

They provide full access to the regulators, and they've allowed academic researchers full access. Putting it all online would be more transparent, but they're a business and they have up and coming competitors.

is and always will be impossible.

I see so many names in the commit logs, but some standouts include: Blum, Cook, Torvalds, Solodai, Tyragu, Stitt, Bergmann, Wysocki, Panda, de Mello, and no doubt some I missed who have a large number of commits fixing this problem.

Thank to all who undertook this Herculean chore!

Do you honestly believe that mass debt forgiveness -- after COVID was already over! -- was a necessary emergency response to the pandemic? Suspending payments (and interest) during the pandemic made perfect sense, and that was not struck down. I don't recall that it was even challenged.

No, the debt forgiveness clearly had nothing to do with the (already-ended) emergency, it was just an attempt to skirt the law, and the courts were quite correct to strike it down as executive overreach. If Biden wanted to do that, he should have lobbied Congress to change the law. He didn't do that, of course, because he knew Congress would refuse -- even though his party held both houses.

Your comment mischaracterizes what has happened. The Supreme Court has absolutely bent over backwards to let Trump do what he wants in temporary rulings, including jumping in to to stay lower-court orders that no previous court would even have responded to. But their on-the-merits rulings, when they have to issue a full opinion, have been much less friendly to Trump. There have been some incredibly bad ones (e.g. immunity) but Trump has lost more than he has won in SCOTUS final judgements.

What, the best you can do is bring up a show that went off the air fifty year ago?

And even then, when Robert Young did TV commercials for coffee, he was identified as "Robert Young," not as Doctor Marcus Welby, and they didn't pretend it was real. (In fact, the line from a series of advertisement of around that era, "I'm not a doctor but I play one on TV," is still a meme today.)

Buy indie games.

It's only the big players who have these delusions.

I'm not saying the right answer is to get a refund. The right answer is to not make the license revokable.

For the theater comparison: If the theatre would invalidate my ticket and throw me out mid-movie, you can be sure that I'd ask for a refund. And in any sane jurisdiction, I'd get it.

Mostly, the difference is some legales, but the kicker is: "revocable". That is an insane difference. I'm quite sure it doesn't say you get a refund if they revoke your license.

You're treating the current law as a standard handed down from on high, incontrovertible and guaranteed-correct, which must be applied verbatim. And, indeed, laws must be applied as written... but that doesn't mean the laws are perfect forever. Laws are written within a context, and when the context changes, the laws have to change.

In a world where all cars are driven by humans, if you want to protect highway workers one way to do it is to attach serious prison time to killing one and to publicize that fact loudly so that all of the drivers know that they should be especially cautious around highway workers, even more than they would around other sorts of pedestrians (let's put aside the moral debate about whether we actually should protect highway workers more than other pedestrians).

In a world where some cars are driven by software systems, that strategy doesn't really work -- as your question correctly points out -- but the right conclusion isn't "Therefore self-driving cars shouldn't be allowed", or "Therefore we must identify some scapegoat human at the company to put in prison". The right conclusion is "Therefore we need a different kind of regulation to keep highway workers safe from self-driving cars". What should that be? I can think of lots of possibilities, both pro-active (e.g. require self-driving vehicles to demonstrate in rigorous testing that their vehicles stay far from highway workers, with whatever minimum distance you want to specify) and reactive (severe penalties, up to heavy fines and/or immediate loss of permission to operate). The point is that the law should choose an approach that works with the new context.

And? They still drive 10% better than the average driver. And I realize that number is just an example, not intended to be accurate, but I still feel like I should point out that, statistically, it's too low.

The fact that the self-driving cars will all concentrate their worst behaviors in the same regions of the space of all driving conditions doesn't change the fact that, on average, they're quite a bit safer than human drivers. This wouldn't be true if the roads somehow changed so that the problematic-for-self-driving scenarios predominated, but they don't.

So you consider most human drivers too dangerous to be on public roads.

You're taking a binary position (too dangerous to be on public roads), so you should expect people to evaluate your position in a binary way. Your other position, trying to position Waymo safety within the wide continuum of driver safety, is more nuanced.

My position is that if they're statistically safer than average human drivers, which makes them far safer than the worst human drivers on the road, then replacing human-driven cars with self-driving cars makes the roads safer. This is straightforwardly obvious. It doesn't mean the companies shouldn't be held accountable for their failures, and certainly doesn't mean that we shouldn't expect them to to continue working on improvements.

This might work when there is a simple, easy search that can verify a fact. But that's often not the case. In my experience most cases of hallucination are cases where the LLM needs a fact mid-response, and the fact check requires both a non-trivial query and complex evaluation of the response data, sometimes involving judgement calls. When that happens, the LLM just gets lazy and goes with its guess rather than doing the check.

I'm speaking in the context of advanced models, mind, not the kind of thing that was available in 2022, nor the kind of thing that is available in Google search's limited-capability model, or open source models. Those are far more prone to hallucination. I won't say that, say, Claude Opus never hallucinates, because it does... but the hallucinations are common only when the models is being pushed hard, operating near the limits of its capacity, which makes it prone to taking shortcuts.

The original guy got to keep using it. There was someone else hired for a brief time.

I remember the author's name but he really doesn't want to use it, so that's OK to respect. He's given me a lot to think about over the years. I remember when he wrote on his PBS site about unicast becoming cheaper than radio broadcast for TV, predicting that it would overtake by 2012 (IIRC). Youtube became huge around then. We were smart folks around the water cooler in the late 90's who could follow the math but had nagging skepticism. He wss right.

I think I have one of his science writing books under his real name about atomic energy somewhere. You can find it if necessary.

Nice to see Bob back on the Dot.

IIRC, a "recall" is a notification of a defect in a car that needs addressing. If that can be done OTA, it's still a recall.