To be clear, I meant the sensor suite and steering rack and support parts, not necessarily that there would be a working brain available to the general public by then. Leaning towards yes, but no guarantees.

I'll grant you that the liability issue is a giant question mark.

Why would you think that? Cameras a cheap, and LiDAR prices are coming down, too. As companies build them in larger and larger quantities, economies of scale and competition will drive the price down rather quickly. Best guess is that in five years, self-driving hardware will add about $15k to the price of the vehicle if they use LiDAR, or $6k if they don't. And that's including the cost of stuff that a lot of cars come with already, like the electric steering rack. I'd be shocked if it were significantly more than $20k.

So as drivers replace their cabs or semi tractors, they'll spend the extra $20k or whatever to buy versions that are self-driving. For that matter, once the tech is reliable enough, you'll likely see retrofit kits come on the market. There's no good reason you'd replace a working tractor unit when you can just swap out the steering rack, bolt on cameras, and add some electronics, and that's true whether you're an owner-operator or the manager of FedEx's fleet.

Oh, absolutely. Most local delivery uses people who already work at the business, and delivery is just a small part of that person's job. So that impact is likely to be close to zero. But that still leaves probably probably around 5 to 10 million taxi drivers and probably three or four million truck drivers.

But taxi and truck drivers won't be replaced overnight. Most taxi drivers and many truck drivers own their own rigs, and although they may eventually replace themselves with robot rigs, they would continue to earn the revenue after doing so. They certainly have no incentive to fire themselves.

Ultimately, somebody has to own the rigs. There's nothing that necessarily requires that robotaxis be fleet vehicles owned by some big company like Uber, no matter how much companies like Uber might prefer it to be that way. Replacing all of those taxis with robot cars costs money, and Uber isn't capitalized that well. Uber's cash on hand wouldn't even be enough to replace all of the taxis in the United States. So while this may shift things around some, I wouldn't expect a taxipocalypse.

Sorry, was thinking orders of magnitude, not orders of complexity. :-)

I'm not sure why he thinks it will be a problem for drivers. A study a few years ago showed that something like 96% of all Uber drivers quit within the first year. So worldwide, we're talking about only O(350,000) people who will have to find something else to do. The world economy can easily absorb such a tiny number.

That's no moon. It's a space station. :-D

Wine runs in user space. I don't see how Wine could ever run drivers, such as peripheral drivers required by things like the iPhone sync functionality of iTunes or kernel-level anti-cheat required by major online games supporting pickup matches with strangers.

People who use software that requires Windows are "forced" to buy Windows along with a new PC. An OEM Windows 10 license that came with a PC is locked to that PC. A new PC purchased to replace a PC that cannot be upgraded to Windows 11 will come with a new Windows license.

Start with stage0 (whose binary seed is about 1 KiB) and GNU Mes. Use mescc to build tinycc, then GCC 2.95, then GCC 4.7, then fairly modern GCC, and then use mrustc to build some version of Rust. The time-consuming part is that each version of the Rust toolchain uses fairly new features in the Rust language, so yes, you'll probably have to build the world a couple dozen times starting with the most recent version supported by mrustc.

Having control is very different from allowing a third-party company to send down arbitrary definitions at any time that suddenly render arbitrary software nonfunctional. The whole concept of Crowdstrike can be summarized as "McAfee Antivirus on steroids". I mean, this sums it up.

This. The "someone says 'Hey Siri/Okay Google' on TV/radio/loudspeaker" problem is a well-known failure mode, and if they don't have reasonable mitigation in place by now, they don't know what they're doing, and their product shouldn't be taken seriously. Whether that mitigation is blocking it during meetings, doing handshaking to limit commands to the nearest device when multiple nearby devices detect the hot word at exactly the same time, making it recognize your voice and not other random people's voices, or any of dozens of other strategies for coping, having some mechanism in place to handle this should be considered a base requirement for any voice-based assistant.

If I had to guess, we let this happen by the USA electing George W. Bush in 2000. His incoming Attorney General settled United States v. Microsoft Corp., 253 F.3d 34 (D.C. Cir. 2001), with a wrist slap instead of a Windows division breakup.

You probably guessed correctly: equitable relief in the form of an injunction against Disney bringing a trademark lawsuit. I haven't read the complaint, but I'd be surprised if it didn't cite Kellogg and Dastar.

The Supreme Court of the United States has decided a few cases about the interaction between the Lanham Act, which inclues trademark law, and exclusive rights pursuant to the Copyright Clause. Key cases includes Kellogg Co. v. National Biscuit Co., 305 U.S. 111 (1938), and Dastar Corp. v. Twentieth Century Fox Film Corp., 539 U.S. 23 (2003). In both cases, the Court ruled that the Lanham Act cannot be used to extend the effective term of exclusive rights in an invention whose patent has expired or a work whose copyright has expired. Disney's legal counsel ought to be familiar with the latter case, seeing as it involved a company that is now a subsidiary of Disney.

From the article:

The Go project recently arranged for Go itself to be completely reproducible given only the source code, meaning that although a build needs some computer running some operating system and some earlier Go toolchain, none of those choices matters."

[...]

The Multics review is famous for pointing out the possibility of adding a back door to a compiler to insert back doors in critical system programs during compilation [...]. Reading the report inspired Ken Thompson to implement exactly that attack on an early Unix system, probably in early 1975. He later explained the attack in his 1983 Turing Award lecture, published in Communications as "Reflections on Trusting Trust."

David A. Wheeler described a defense against a back door that propagates through the compiler in a 2009 PhD dissertation titled Fully Countering Trusting Trust through Diverse Double-Compiling . Diverse double-compiling (DDC) involves choosing two or more other independently developed compilers A and B for a language, bootstrapping compiler C from source code through each of them (building C with A or B and then building C with itself), and ensuring that the output is byte-identical. This relies on previous effort to make builds reproducible.

However, DDC also relies on having more than one implementation of a particular language. Go and Rust each have only one widely used implementation. This means someone trying to wrangle a supply chain has to do one of three things: trust a particular old version of a compiler not to have a back door, compile every version since the dawn of the language (such as when Rust was prototyped in OCaml), or implement a usable subset of the language in a more widely implemented language. This is why mrustc is so important, as it's a way to skip forward by several years' worth of versions when bootstrapping a Rust compiler.