You not mentioning LLMs doesn’t mean the post you were replying to wasn’t talking about LLM-based AGI. If someone responds to an article about the obvious improbability of LLM-based AGI with a comment about the obviously make-believe genie, the only obviously make-believe genie they could be referring to is the one from the article. If they’re referring to something outside the article, there’s nothing more to suggest it’s non-LLM-based AGI than there is Robin Williams’ character from Aladdin.

AGI being possible (potentially even inevitable) doesn’t mean that AGI based on LLMs is possible, and it’s LLMs that investors have bet on. It’s been pretty obvious for a while that certain problems that LLMs have aren’t getting better as models get larger, so there are no grounds to expect that just making models larger is the answer to AGI. It’s pretty reasonable to extrapolate that to say LLM-based AGI is impossible, and that’s what the article’s discussing.

He’d already contributed to humanity by writing the definition of wanker for the Oxford English Dictionary by the age of 28, though.

Many of the claims of improved efficiency over natural meat are based on projections and rely on technology that doesn’t exist yet, or they just ignore things like heating (most of the calories mammals eat go to keeping them warm, but incubators for lab meat need heating electrically, so if you leave out the electricity cost, it gives lab meat an unfair advantage.

It might be more efficient eventually - there’s more wiggle room to change things with a machine, but a cow is a cow - but it isn’t yet.

Windows permissions are more flexible than basic Unix ones. A file doesn’t just have an owner and a group, it can have individual permissions for arbitrarily many entities, so taking ownership doesn’t remove any of the permissions from anything that already had access, it just adds more. The command shown here is closest in effect to deciding you’re always going to log in as root from now on, although Windows has a way to effectively do that without modifying the ACL of every file. Either way, it’s silly, and usually people who suggest it are under the impression that XP did permissions right by not meaningfully enforcing them and not having an equivalent of a root account you can temporarily switch to, and Vista only changed things specifically to annoy people, and not to be more like Unix.

As I said, I fundamentally disagree. Even if you can make a nearly-teenager-proof website (and so far, your example has been something that most of the people I was at school with could have beaten aged thirteen), teenagers can just go to a different website, so the system is only ever as teenager-resistant as it is difficult to find a website that doesn’t care. Most vaguely competent teenagers know how to find pirate sites with illegally-hosted TV, movies and music (even if they’re not techy, one of their friends just has to tell them a URL and they can visit it). Governments have had minimal success stopping online piracy even when aided by multi-billion-dollar copyright-holding companies, so there’s no realistic reason to think they’ll have any more success stopping porn sites with non-compliant age checks.

My point is that you can’t build a completely teenager-proof system. Even if most parents uphold the most unimpeachable password discipline, someone’s going to put a password on a post-it note near their computer, and have their child see the piece of paper, or use their dog’s name despite their child having also met the family dog.

The original comment I was replying to was framing the issue as teenagers being allowed to watch porn versus no teenager ever seeing porn and maybe some freedom is sacrificed to do that, which doesn’t match the real-world debate. If freedoms are sacrificed just to make it a hassle for teenagers to see porn, that’s much less compelling whether or not you see it as a worthwhile goal.

As for what a teenager with access to their parents’ bank password would do, if they’re not a moron, they’ll realise that spending their parents’ money will leave lots of evidence (e.g. that they have extra stuff, their parents have less money than expected in their account, and there’s an unexpected purchase from The Lego Group on the bank statement), and so they’re guaranteed to end up in trouble for it. It’s not any different to a child taking banknotes from their parent’s wallet. On the other hand, using it to prove adulthood, if it was truly untraceable like adults would want, wouldn’t leave a paper trail.

• Teenagers can find out their parents’ passwords (or their friends’ parents’ passwords) if they really want to, and if things are anonymous enough not to leave a paper trail that would allow spouses to see each other’s porn usage, they’re anonymous enough to let teenagers hide that they’re using their parents’ credentials. 2FA helps, but it’s not like teenagers never see their parents’ phones.
• There’s not anything that all adults in the UK have that could be used for everyone. There’s no unified national ID or online government identity. There’s no one-size-fits-all bank login system. You’d have to build and secure tens of independent systems to cover nearly all adults.
• As I said in the post above, if it’s too much hassle for teenagers to access mainstream, legitimate porn sites, then there’s very little anyone can do to stop them accessing obscure ones that don’t care about obeying the law or can’t do so competently. If governments could stop websites from existing and providing content, there wouldn’t be any online piracy.

There is no possible way to actually stop teenagers accessing online porn that doesn’t require such a massive invasion of privacy that it leaves no safe way for adults to access it. To go with your adult video store analogy, it’s like if the store staff would have to accompany you home and watch you watching the porn to check there wasn’t anyone standing behind you also looking at the screen, and while they were there, they were supposed to take notes on everything they saw. Even if they had no interest in doing anything nefarious, a criminal could steal their notebook and blackmail all their customers with the details it contained, and there’d be enough proof that there wouldn’t be any way to plausibly claim the blackmailer had just made everything up.

If you want to prove someone on the Internet is a real adult and not a determined teenager, you need lots of layers. E.g. if you just ask for a photo of an ID card, that can be defeated by a photo of someone else’s ID card, and a video of a face can be defeated by a video game character (potentially even one made to resemble the person whose ID has been copied). You need to prove there’s an ID card that belongs to a real person and that it’s that person who is using it, and that’s both easier to fake than going to a store with a fake ID (if you look young, they’ll be suspicious of your ID) or Mission Impossible mask, and unlike in a store, the customer can’t see that you’re not making a copy of the ID card for later blackmail or targeted advertisements. No one would go back to a porn shop that asked for a home address and a bank statement to prove it.

Another big factor is that if there’s a physical shop supplying porn to children, the police will notice and stop it, but online, it’s really easy to make a website and fly under the radar. It’s pretty easy for sites that don’t care about the law to provide an indefinite supply of porn to children, and once that’s happening, there’s no reason to think that it’s only going to be legal porn just being supplied to the wrong people.

Overall, the risk of showing porn to children doesn’t go down very much, but the risk of showing blackmailable data to criminals and showing particularly extreme and illegal porn to children goes up by a lot. Protecting children from extreme material, e.g. videos of real necrophilia and rape, which are widely accepted to be seriously harmful, should be a higher priority than protecting a larger number from less extreme material that the evidence says is less harmful, if at all. Even if it’s taken as fact that any exposure to porn is always harmful to minors, the policies that are possible to implement in the real world can’t prevent it, just add either extra hassle or opportunities for even worse things to happen. There hasn’t been any proposal by any government with a chance of doing more good than harm.

You’re well into Ship of Theseus territory once you’re replacing major chips. You’ve not really recovered the phone, you’ve just replaced big chunk of it so it’s not entirely the same phone anymore, just with extra hassle because you’ve not replaced the main board as a whole.

Putting "false" in a YAML file gives you a string, and just false on its own gives you a boolean, unless you tell the YAML library that it’s a string. Part of the point of YAML is that you don’t have to specify lots of stuff that’s redundant except when it would otherwise be ambiguous, and people misinterpret that as never having to specify anything ever.

Most of the problems can be totally avoided by telling the YAML loader what type you’re expecting instead of forcing it to guess (e.g. provide a schema or use typed getter functions). If it has to guess, it’s no surprise that some things don’t survive the string to inferred type to desired type journey, and this is something that isn’t seen as a dealbreaker in other contexts, e.g. the multitude of languages where the string "false" evaluates to true when converted to a boolean because it’s non-empty.

Sometimes when a product is whitelabeled and can be bought directly from its real manufacturer for much less, there’s still a good reason to buy the expensive one. The main one is that sometimes the ones that pass QA are sold via the whitelabeler and the ones that fail QA are sold directly, so the cheaper ones are known to have something wrong with them and you’re gambling that it’s something without symptoms.

The new law allows you to have more than one charging connector provided that either the USB-C one is the best one, or the USB-C one is as good as the spec allows. If the new connector’s genuinely better, then it’ll beat a maxed-out USB-C connector, so devices will provide it in addition to a maxed-out USB-C connector.

Male to female A-to-A cables are pretty common (they’re just basic extensions) and totally legal under the spec provided they’re limited to a certain length or contain a powered repeater. It’s just the rare male-to-male (which my keyboard stupidly uses) and even rarer female-to-female that aren’t legal. There’s also the exception of USB-on-the-go cables with a micro-B end and a female A end for devices like smartphones that are capable of being host or connecting to a host, back before they switched to USB-C.

It doesn’t have to have been effective. They might just have overestimated how many people would think killing health insurance CEOs was unacceptable.

It adds the executable permission (without which, things can’t be executed) to all the files in the game’s directory. You only need to be able to execute a few of those files, and there’s a dedicated permission to control what can and can’t be executed for a reason. Windows doesn’t have a direct equivalent, so setting it for everything gives the impression that they’re trying to make it behave like Windows rather than working with the OS.

Selling old games and new games isn’t mutually exclusive, and more money tends to be spent on new games than old ones. It’s not unreasonable to expect that selling new games too could subsidise the work to make old games run on modern platforms.

Like other commenters have said, start by asking the upstream developer (whether that’s by sending a message with a link to the fork or by sending a mega-PR that says you don’t expect it to be merged as-is in the description). They should be the best judge of how they’d prefer to handle it. The thing I’d add is that you should try to avoid taking it personally if their preferred approach isn’t one you think is a good idea. Sometimes good fixes end up never merged because of disagreements becoming too heated even if everyone’s basically on the same page about the fox being good. There’s also a decent chance that your refactors are things the upstream developer explicitly doesn’t want and would otherwise have done them themselves and implemented the same fix, too, or they don’t agree that your fix is good enough. They won’t want to be on the hook for maintaining contributions that use approaches and code style that they don’t like, and that’s okay. They also might know something you don’t about their project that would make something that’s obviously a good idea to you obviously a bad idea to them.

Basically, just try and remember that if it’s a hobby project, it makes progress when the maintainer is having a good time, and gets abandoned when they’re not anymore, so try and avoid making a mess and having arguments when they’re the one that’ll have to deal with any fallout from any mistakes.

You’re not throttling between 0% output and 100% output, as that takes weeks or months, and instead throttling within a limited range at the upper end of the output power. Because a nuclear reactor puts out so much power compared to a combined cycle gas turbine, going down to 80% power has a comparable impact to totally shutting down a gas turbine. It doesn’t need to be instant to be used for dynamic load - throttling a gas turbine isn’t as it takes time for the heat exchanger to warm up or cool down after increasing or decreasing the fuel flow, and time for the first turbine to speed up or slow down after the flow of the Brayton-cycle coolant changes, and then more time for the second heat exchanger to heat up/cool down and more time for the Rankin-cycle turbine to speed up or slow down as the flow of steam changes, and only then is the new desired output power achieved.

Wikipedia puts the average emission time for delayed neutrons at fifteen seconds, which while ludicrously slow compared to a bomb, is really fast compared to the day-night cycle that represents most dynamic load variance in a country with plenty of renewables or heavy industry that doesn’t operate at night time, so there’s plenty of time for the power output to respond as long as you’re restricting the range that it’s operating in.