I'm assuming they have some physical backups somewhere, yes. But they'll probably be at least several days out of date.

The *daily* backups were almost certainly the 10TB of backups that were found and destroyed by the attackers. Which makes sense: you want your most frequent backups to be fully 100% automated so they're as up to date as possible whenever a hard drive dies (which, for most organizations, happens considerably more often than this kind of successful malicious attack). So your continuous and daily backups go onto media that are online 24/7. So when something like this happens, you're going to have to go back to the last time a secondary backup was made, and that's less automated (among other things, someone has to physically swap the media in and out, and if we're talking about 10TB of data, that's probably going to have to happen multiple times, over the course of a couple of days, to complete the backup), so it generally happens less frequently. Since this was a munitions factory, we can charitably assume they would have known they were a potential target for this sort of thing, and so probably would have at least done a secondary backup weekly? Probably. Most organizations don't have their sysadmins practice restoring from secondary backups on anything resembling a regular basis, so they won't really know what they're doing and will run into all sorts of minor-but-annoying setbacks and delays. Software that's needed won't be installed, and there won't be a complete list of it anywhere, so they'll have to fool around by trial and error figuring out why blah-blah-blah won't run, oh, we forgot to install foolib on the design department's database server, have to do that, ok, now why does it still not run, oh, it also wants the foolib extensions for Postgres, install that, rinse, repeat. Some data that are stored in oddball locations (typically, configuration stuff) will have been missed, and will have to be recreated. And so on.

It's hard to predict exactly how long that stuff will take, but my first guess would be more than a week.

Granted, that's a far cry short of the timeframe if the factory had been, say, bombed into craterdom. But this may have been cheaper, and in any case it also gives Ukraine a significant amount of information about the factory's operation, which could be valuable in other ways.

Superintelligence might be a ways off.

Tesla has had its own custom-designed AI chips running in cars since HW3 in 2019, replacing an earlier setup based on NVidia GPUs in order to get better performance at a lower unit cost and lower power consumption, with onboard redundancy, by optimizing the hardware to their tasks. They don't build the chips - they don't have a fab - they just design them.

Since 2021 (officially) they've also been working on the much more architecturally ambitious Dojo training supercomputer project, but I'm not sure if this contract has anything to do with it.

Sounds ripe for a class action lawsuit.

I mean, in principle, you could also buy a much cheaper firearm (say, a shotgun) from Wal-Mart, and use an affordable consumer-grade 3D printer to print up a cheap plastic shell that it fits neatly into, that makes it look like a toy guitar or light saber or whatever. Granted, that would get noticed by a metal detector.

The argument I find most interesting in this debate is the economic one, though. 3D printers that are good enough to print a practical firearm, are *outrageously* expensive, closer to the price of a house, than the price of a normal, legal firearm. Fundamentally, mass produced items are always going to be cheaper than one-off equivalents, because the whole process of making them, including the purchase of the materials, can be optimized. So in theory, anybody whose interest in firearms is of a practical nature, should just go buy a mass-produced one (or three, or seventy-three, or however many you think you need for whatever hunting trip you have planned or whatever). The 3D-printed ones are mainly appealing to enthusiasts and hobbyists and whatnot, for reasons that are not pragmatic in nature. I'm not quite sure which side of the debate over their legality this most argues for (which is probably why nobody brings it up much), but I think it's worth taking into consideration one way or the other, and also I think it's more _interesting_ than most of the issues that do get discussed to death every time the topic comes up.

I want to be able to take out 2-4 screws, maybe pull a latch, and the case opens. At that point I want to be able to easily swap out things like RAM or the hard drive (either for higher capacity, or as a repair). I want to be able to easily replace the parts that break most often, such as the power supply and cable and, most importantly, the _hinge_ with standard items that aren't specific to a particular product line of laptop. Ideally, I'd like to be able to swap out the case, the display, or the motherboard.

Basically, I want a standard form factor for laptops, the equivalent of what ATX is for mid-towers. I don't want to have to throw away the laptop if one thing breaks, three months after the warranty expires.

Show me all of that, and decent specs, and a full-sized keyboard, and all of it works with open-source software, and I'll pay premium prices.

I mean, the chlorine solution will kill any cell it touches, so if you apply it specifically to the cancer cells, that *will* work. Heck, if anything, chlorine is overkill, iodine would work.

The issue is more that there are likely to be better options. If you can identify the edges of the tumor with the required level of precision, you can probably just surgically remove it, or cauterize it, or whatever.

Oh, there's *lots* of boatloads of evidence that high concentrations of chlorine like this, will definitely kill cancer cells.

It'll also kill bacteria, and most fungi, and most parasites. And destroy viruses. Heck, it'll kill most insects on contact. Halogen ions in general are powerfully anti-organic, and chlorine is the second strongest of them (behind only fluorine). The medical community has used iodine solutions as topical disinfectant for decades; that works great, and chlorine is stronger. In fact, pretty much the only reason bleach is safe to handle at all, is because you have a nice thick (by chemical standards) outer layer of skin, that is already dead anyway.

Oh, you wanted evidence that the treatment is _safe_ ? Oh, that's different then. Though honestly, a lot of cancer treatments are kinda dubious in that regard. Even quite hawkish regulatory agency like the US FDA, tend to apply a standard somewhere in the vicinity of "statistically at least marginally safer than letting the cancer keep going unchecked, we think". Word of mouth among cancer survivors is that if you're tough enough to get through all the chemo and radiation and whatnot that they can throw at you, without crying uncle, then you can probably beat the cancer too.

Sterile water is relatively cheap and easy to obtain and to store. If you can acquire and store this new product in case of emergencies, you can also acquire (in advance) and store the water.

The scenarios where this is potentially going to save lives, compared to the current system, mostly involve remote areas with sparse population where it's impractical to maintain blood reserves. Everywhere else, what it's going to do (assuming eventual full success), is reduce the quantity of blood donation society requires, and the expense of keeping it all fresh and rotated and so on. Which sounds a lot less dramatic but is valuable nonetheless.

I think the article was asserting that the instructions were "dangerous" because they involved burning and cutting oneself on purpose, among other things, not because anyone thought ChatGPT's instructions might actually result in a demon being summoned.

Though, if you believe in Evolution, then you should probably consider instructions like that a net good for the genome, because anyone who would follow them is, ipso facto, extremely gullible and suggestible and not very good at making decisions.

Not true: viruses didn't crash nearly as much.

He probably means the small amounts of various metals (lithium, transition metals, lanthanides, ...) found in anything electronic. Many of which are toxic in high concentration, but totally natural in low concentration. He presumably just hasn't thought through all the implications of where they come from in the first place and what it is about the human activity surrounding them, that makes them toxic. Either that or he vastly underestimates how expensive it is to get any significant quantity of anything into orbit, and therefore over-estimates the concentration that's achievable in the atmosphere by human activity. Perhaps both.

Almost the same here, but in Wayland, so Sway instead of i3.

Yeah, but it's important for the Chinese domestic market, because it's enough of a pain to get foreign companies to cooperate with the Chinese government when they have to have physical infrastructure (like cell towers) on Chinese soil; they find ways to manage it, but it's a hassle. If the foreign company has *no* meaningful assets inside mainland China (because they're in space), it would be even more of a hassle, and China would like to avoid that scenario.

We can expect foreign options like Starlink to be illegal to import to China, as soon as they have a domestic alternative up and running.

This is also why Baidu and Weibo and so on exist.

Yes, I came to the comment section to say essentially this. Economists commonly call this condition "Dutch Disease", for historical reasons involving the mid twentieth century. The short explanation is, if one specific sector (commonly, resource extraction) becomes disproportionately profitable, it drives demand for your currency overseas, shifting the exchange rates in a way that makes your other products less competitive, making it difficult to maintain or develop a balanced or diverse economy. (This is, of course, a simplification; but it's a good simplification, very useful for beginners to wrap their minds around the basics before digging into more detail later.)