Yes, and it needs to apply to absolutely everything, including computers, cell phones, etc. Having hardware locked to a single software vendor is barely tolerable while the manufacturer is supporting it, but absolutely should not be allowed once the manufacturer drops support.

But that's not really enough. Reverse engineering the hardware for nontrivial stuff is usually not worth the effort. The companies should be required to also do one of two things:

• Release their internal documentation telling address space layout, programming info for any custom ASICs, etc.
• Provide complete source code for their firmware, along with the toolchain required to build it.

This serves two purposes. First, it makes it much easier for a third party to bring up a usable firmware build without having to figure out how to unsolder the CPU and install a jumper harness under it to expose the JTAG pins. Second, it greatly increases the amount of effort required to drop support for a device, which may be enough to make companies think twice about doing it, and also may make companies think twice about releasing a product without a reasonable long-term plan for how to support it indefinitely.

Devices that require the cloud should be banned, period. Any device sold should be required to provide a non-cloud-based alternative control path that never leaves your network. This falls under "minimum security requirements" even if you ignore the whole "the cloud server could go away" problem and the "I can't turn my lights on because the Internet is down, and I can't see to reboot my router because I can't turn the lights on" problem.

But that's a separate law.

That only helps if it is actually possible to build the firmware, which isn't a given. Also, that's opening all users up to a lot of risk if it gets installed automatically, because if there are any security holes in the device, someone would then have the global signing key and could upload new firmware that permanently compromises every unpatched instance of that device all across the planet in a matter of days.

IMO, a much better process is something like this:

• Devices must be able to have properly signed firmware side-loaded, either by shoving the data on a USB stick and plugging it into the device (strongly preferred) or by making the device look like a USB mass storage device when you connect it to a computer and the user dragging and dropping the firmware bundle onto it.
• The manufacturer stops shipping automatic firmware updates.
• The manufacturer releases a manually downloadable firmware update that allows the user to provide a replacement signing key by attaching a mass storage device or uploading it to the mass storage endpoint.

The use of an external mass storage device (rather than the device looking like a mass storage device) is strongly preferable for security reasons, because it would be much harder (and, if done correctly, impossible) to exploit remotely and install a malicious party's key onto the device, then trick it into downloading a new firmware bundle from some other server, etc.

At that point, the user would provide a new user signing key that would sit alongside the manufacturer signing key and would be treated as an authorized signing key for firmware uploaded to the device. This would also have the benefit (from a corporate perspective) of making the product something that would only be used by hobbyists, and not turned into a tool for their competitors, because the installation process would be just enough effort to limit interest.

As we always used to say, "If you think having a backup is too expensive, try not having one."

Mark Zuckerberg voiced by the guy who played Milton: I make the [expletive] [expletive] decisions so the engineers don't have to. I have decision skills! I am good at making decisions!

IMO, a minimum requirement for anything other than a virtual serial port would be a microcontroller that has 32-bit addressing. That way you can put some shared dual-port RAM between the microcontroller and the main CPU and use it for a pair of ring buffers.

That wouldn't really be feasible on a 16-bit chip like the ATmega2560, because you'd be cutting into your program memory too much, and program memory in those things is already very limited.

What I'd like to see is a full Linux-based ARM setup with a fast CPU and a Mega2560-like controller on the side, with interprocessor communication channels designed to support message passing between one or more threads in the Linux app and code running on the 2560, i.e. the Mega's address and data bus would have to exposed externally and they would have to share some RAM.

But maybe that's just me.

To be fair, I mostly am as well, just with my cynic hat firmly in place. :-D

Yes and no. A transformer does have a gap, but it is a really small gap. Unless you have some robot arm that moves up and touches the underside of the car while it is parked, which, being a moving part, would be a frequent point of failure, possibly even making it very difficult to move the car without causing damage, you're going to have a very *large* gap, measured in inches.

The difference in minimum ground clearance between a Model S and a Cybertruck is 3.9 inches. That's a LOT of difference. To make this a general solution would almost certainly involve either major suspension changes to allow for standardizing the minimum ground clearance height or some sort of moving arm that reaches down to the charging pad, which is going to be a moving part that will end up being a frequent point of failure. So realistically, I don't think you'll see something that is 90% efficient across all cars. There's just too much variation.

Also, large HVDC charging stations actually convert to DC very early in the process and use batteries to level out the load over the course of the day to minimize demand charges from the power company. A wireless charging solution would involve an extra DC-to-AC conversion step and the resulting losses from that, so it's going to be significantly worse than 90% efficient when you factor that in.

That's why both vehicles and chargers are typically designed to let you quickly change out the charge connector. At least on Teslas, it's something like a half-hour job to swap out the charge port, and Tesla Mobile Service can do it without you even bringing in the car. And commercial-grade EV chargers typically have cords that are field-swappable. (Whether cheap consumer-grade chargers do or not, I couldn't say.)

Even if it is 90%, that's still an 11% increase in power consumption. Now consider a car. If you are driving two or three hundred miles per week, that might be 50 to 100 kWh per week. At California prices, that could mean an extra $2.50 to $5 per week, or $130 to $260 per year.

This is not a small amount of power loss we're talking about anymore.

Nah, there will be plenty of them working at every cafe in Hollywood just like they are now.

But in all seriousness, the real problem with acting as a career is that only maybe one or two percent of the people who graduate with degrees in drama actually end up acting as a career. Maybe a quarter of them manage to get some odd jobs on the side doing a little bit of voice acting or hand modeling or whatever, but it doesn't pay the bills. The rest of them end up doing other things. There are far more people who want to act than there are jobs that can pay them.

And the reason for this is because Hollywood rakes in huge amounts of money for a small number of elite actors and actresses, and leaves no market for much of anything else. That's changing somewhat with the rise of streaming, but even there, the number of companies doing production is relatively small, and the amount of content they produce is quite limited. In other words, acting wasn't ever really a realistic career in the first place.

And because most of the money went to the wealthy people at the top instead of being turned into more production, and because copyright has made it so that studios can reissue old films and sit on their laurels and make money off of their back catalogs instead of being forced to compete with the public domain by creating new content as the creators of copyright law originally intended, we're seeing more consolidation and less production. Every year, seasons get shorter, fewer shows get greenlit, etc.

So the way I see it, this is destroying a job that was already in decline and hard to get, while in the process creating a giant pile of new jobs for content creators at a different level. I'm not sure that's a bad thing. And there will still be people who choose to act. They'll just be doing it on stage, in schools, in short films that people create for fun, and so on.

Hahahahah. Receptive. Sure. When I'm wasting half an hour or more of my time trying to find a very specific product that meets very specific requirements and can be delivered in a tight timeframe, I'm going to be happily distracted by an ad and say, "Oh, maybe I should buy toilet paper, too." If Amazon's search didn't suck harder than a black hole, maybe, but as long as it's an uphill battle just to buy things on Amazon, and as long as 75% of my purchases end up with me doing a carefully tailored Google search to get to the products that I'm trying to buy instead of the higher-margin, but useless products that they're trying to push, their ads are just going to piss me off even more.

I leave when I make a purchase or conclude that I could not find what I needed. And every extra distraction that gets in the way of finding what I needed makes the latter more likely than the former. It's self-defeating. See also my comment above about seeing products locked in a cage and giving up before someone came to unlock it and ordering it on Amazon. If Amazon becomes too inconvenient, I'll buy it on AliExpress.com or Walmart.com instead.

Do you know, out of all the times I've shopped at Amazon, what percentage of the time an alternative was worth considering? I always have a list of requirements, and products that don't meet all of those requirements are of no value to me. When I'm looking for 6mmx30mm hex pan head machine screws, knowing that Amazon also has 5mmx30mm machine screws and 6mmx30mm countersink Phillips screws is of no value, because I need the exact product that I'm searching for. Again, all that does is make it more likely for me to leave and buy it somewhere else.

Maybe 1% of the time, I don't know exactly what I'm looking for. And even then, I know enough to reject large swaths of products in a particular space. What's missing is the "Don't ever show me products like this one again" button. That's what makes Amazon so miserable is that there are so many hundreds of vendors selling the exact same products with different name badges, and you can't find the legitimate products by established vendors for all the dross.

And I've literally never, in all the years I've spend many thousands of dollars per year on Amazon, bought anything because of any of that. So clearly it doesn't work, it isn't smart, and it isn't data-driven, because if it were, they would have realized that they're just wasting my time and pissing me off, and they would have stopped showing me that crap.

He's also wrong about Fulfilled by Amazon. It might be more expensive, but there are no comparable fulfillment services. None. Amazon has 1200 fulfillment warehouses and logistics facilities. And assuming the Google summary isn't wrong, FedEx Logistics has 130, and UPS has 250.

That's why I can order a lot of products and get them on the same day or the next day, depending on when I order them. Literally nobody else can do that unless you want to spend a hundred bucks on next-day air shipment. And when you look at it that way, you just might find that it's not so expensive after all.

Whether that speed translates into increased sales or not is, of course, a different question, and that would probably depend on the vendor and what they're selling.

If true, that's pretty open-and-shut false advertising, and gas companies have been hit with very large fines for playing very similar tricks. This is arguably tantamount to drip pricing, which is explicitly illegal in California (SB 478). And on the federal level, it also appears to violate my casual reading of Title 16 Chapter 1 Subchapter D Part 464 section 2a. Why? Because if you are non-member, that's not the total price. The total price is the other, higher price, or alternatively, the lower price plus the price of Prime membership. And that higher price must be clearly and conspicuously disclosed under 2a.

So to the FTC, if you're reading this, please nail Amazon to the wall.

That's kind of like mine, but backwards. I need something ASAP, so I look online and figure out if they have it at Lowe's or Home Depot. Then I drive there, find that it is locked in a cage, and either hit a call button or wander around looking for someone to unlock the cage that it is locked in. After about five minutes without getting any help, I check the price on Amazon, find that it is $20 cheaper and I can have it the next day, and I order it from Amazon as I walk out of the store without having made a purchase.

Between those two stores, I'm pretty sure they've lost at least a thousand bucks in purchases just from me in the last year alone because of those cages. The whole reason for buying locally is because it is faster and more convenient. If you throw obstacles in my way so that it stops being faster and more convenient, I'll just order everything from Amazon and you'll get no business from me whatsoever, because realistically speaking, they are almost always cheaper. And as more and more stuff from Amazon gets same-day delivery, my willingness to put up with local merchants and the related hassle diminishes.

I mean, it's not terrible. Acceleration is comparable to my 2017 Model X or a current-generation Model Y. (The current generation of Model X leaves it in the dust.) The main problem is that it has as little as 68% of the range of a current Model Y, despite being smaller than a Model Y, weighing less, and having slightly less powerful motors. And despite all of those compromises, it is still only $6k cheaper than the Model Y.

When you're spending about $40k on a car, the difference between $38k and $44k is a lot less than the difference between 222 miles of range and 357 miles of range. If this car were $25k, it would sell very well. But it's competing head-to-head with the Model Y on price and is getting stomped into the ground by the Model Y on range, size, and performance.

The remarkable thing is that the battery isn't significantly smaller than the Model Y in terms of capacity, the motors are less powerful (using less power), and the vehicle weighs less, yet they're still getting less range per charge. Unless they're reserving a lot more of the capacity at the top/bottom of the battery, this suggests to me that either they're wasting a *lot* of power on poor accessory efficiency or their drive motor efficiency sucks. Am I missing something?