Either way, miles isn't, because it's an arbitrary metric. Is that Prius miles (as little as 180 Wh per mile) or Cybertruck miles (800 Wh per mile while towing)?

Try a different USB-C cable. Either that or you have one of the dodgy models of Sandisk SSD that's about to die. Maybe a good time to buy a new one.

Also because of coal. Honestly, more as a result of coal, though they certainly have built a lot of solar. But the reason they've been building coal plants like crazy, so much so that many of them are idled from the day they go into service, is because it was their insurance against problems with the oil supply.

I'm a fan of solar power and happy to see the world is building a lot of it, but intellectual integrity demands that we also acknowledge China's investment in coal generation capacity.

Ok, to clarify a few things:

Current designs I've put up:

1. A modernised version of the DeHavilland DH98 and Merlin engine, where I basically fed ChatGPT and Claude with all of the known historic faults and some potential solutions to various problems, then let them run wild, feeding off each other to fix, refine, and clarify the various design. The premise here is that we're using known designs with known properties, changing only materials but doing so carefully so as to ensure that the balance is unchanged from the historic design. The aircraft is probably the least interesting part, as it would be very hard to make that safe, but a fully modernised Merlin that starts where Rolls Royce left off is something that could be built with minimal risk and could be quite interesting in its own right.

2. A High Dynamic Range microphone. This basically riffs off assorted physics technologies for measurement and the basic idea in many HDR schemes that you can split an input into the fine detail (essentially an equivalent of a mantissa) and a magnitude (essentially an exponent), producing a design that aught to permit (if it works) the same microphone with no adjustments handling everything from a nearby whisper to the roar of a jet engine -- but with all of the fine detail still captured from that engine.

3. An electric guitar that operates not by magnetic pickups but by accurate mapping of string behaviour in two dimensions via lasers, where this is then turned into an accurate representation of the sound in an external device. So it's not a synth guitar in the classic sense, it's actually modelling the waveform for each string in two dimensions precisely. The reason for doing 2D modelling is that this has the potential for novel behaviours but without an obligation for it to do so.

4. A synthesiser/wave processor that looks at everything that they knew how to do, and allows you to link it together arbitrarily. It is designed in two forms. The first is engineered to match the components, materials, and knowledge available in 1964, so it is something they could have built if sufficiently insane. The second is a modernised extrapolation of that, using modern digital electronics, where I can show that the modern version is a strict superset of any existing DAW, simply because I started with none of the assumptions and metaphors around which DAWs were subsequently designed.

5. Multiband camera. An attempt to build a digital camera that is far smaller and more compact than a 3CCD camera, but (like the 3CCD design) produces a far better picture than a conventional digital camera, where I don't stop at three frequencies but support many, albeit with the limitation that the time required for a photograph is abysmal.

Each design I've put up has a detailed hardware specification (including wiring where appropriate), validation/verification documents, and testing procedures. Software is defined by means of formal software contracts and occasionally Z-like forms. The designs are extremely detailed, although not quite at the level you could build them right there and then. However, the synthesiser is described right down to the level of individual transistors, diodes, and connectors, and the Merlin engine specifies precise materials, expected temperature ranges, material interactions (and how they're mitigated), and other such information.

Again, it's precise but not quite at the point where an engineer would feed comfortable feeding the specifications into an AI, having it order the bits online, and be sure of building something that works, but it's intended to be close enough that (provided the AIs actually did what they were supposed to) that an engineer would feel very comfortable taking the design and polishing it to working level.

If, however, an engineer looking at these designs comes to the conclusion that the AIs were utterly deluded, then obviously they can't handle something as simple as selecting candidate items from ranged data.

I wouldn't call that "supporting". He plays the game by donating to both sides, so that whoever wins, they see him as an ally. And in particular, Trump responds to obsequiousness. There is a perception that by giving money, businesses can curry favor. Not saying that doing so is a good thing, but it definitely should not be interpreted as evidence that the business leaders are necessarily in favor of Trump's policies.

There actually was a push in the UK a few years ago to outlaw pointy kitchen knives, but it met with great resistance and was dropped.

However, the point remains that stabbings in the UK are actually less common that stabbings in the US. This points out that while many think that guns are the cause of the US' violence problem, the real problem is deeper: US culture is just more violent.

I post the AI-derived designs to my gitlab page:

https://gitlab.com/wanderingne...

I really should add this to my journal.

Not to mention a Time Machine clone for local NAS storage. I don't want my phone backed up in anybody's cloud. I use iCloud because Apple makes it d**n near impossible to back up locally except over a wire.

That sounds nice and all, but there's really no legal way to do that. The money was collected illegally, so it has to be returned (with interest) to the people it was collected from -- the importers.

It's going to take years to find out just how corrupt, and we'll never get the full story. What we can see isn't even the tip of the iceberg.

Heh. The usual metric is "mission success". For a manned flight, that includes getting the people down safely. For a typical unmanned flight the mission is "get the payload to the right orbit". If you manage to land the rocket after that, that's gravy.

Putting a hardware guy in charge of Apple might help the company return to its roots as a hardware-first company. They've been so distracted by silliness like trying to squeeze more money out of the App Store, iBooks Store, etc., resulting in fines and antitrust decisions going against them in the EU and the U.S. on so many occasions, mostly because the company has strayed too far away from its core mission — to make great hardware and build operating systems to support that hardware and produce a great user experience.

Build a great product, and everything else will follow naturally.

To be fair, that's nothing against Tim Cook. He always struck me as having a good head on his shoulders and being generally a decent human being. And he held the company together through a tough transition, losing one of its founders. That's not an easy task.

But Mr. Ternus has, in some ways, an even tougher job, showing the markets that Apple is more than just a company that sells phones. I don't envy him. But I do look forward to seeing the direction that he takes the company.

Zipline is still around, still making deliveries in Africa and some other places, and also doing some in the US, for Wal-mart.

There isn't any launch platform with no failures, ever, that's not how you measure reliability. Reliability is measured on percentage of successful launches (payload reached target orbit), and Falcon 9 is, indeed, the most reliable orbital launch vehicle ever, by a wide margin. Here are the platforms with >= 100 launches (the 100-launch line is kind of arbitrary, but you have to draw a line somewhere and platforms with very few launches don't have meaningful statistics):

#1 Falcon 9 (including Falcon Heavy): 637 successes of 640 launches, 99.5% success rate. If you focus only on the block 5 variant (most-flown version, currently flying), it's 572 out of 573, 99.8%.
#2 Atlas V: 106 of 107, 99.1%
#3 Delta II: 153 of 155, 98.7%
#4 Space Shuttle: 133 of 135, 98.5%
#5 Long March 2/3/4: 503/521, 96.5%
#6 Ariane 5: 112 of 117, 95.7%
#7 Soyuz: 1889 of 2014, 93.8%
#8 Kosmos: 559 of 610, 91.6%
#9 Proton: 382 of 431, 88.6%

Soyuz has to get props for the sheer number of launches, of course, though that's probably mostly because the Russians couldn't afford to build another platform. Soyuz isn't a particularly great rocket in any way -- smallish payload, good but not great reliability -- but they kept using what they had. It's also worth noting that assuming Falcon 9 maintains its current launch cadence (which it won't; Starship will probably start taking its launches eventually, and if that doesn't happen, the cadence seems likely to increase), it will match Soyuz' launch count around 2033.