Posts by doublelayer 11409 publicly visible posts • joined 22 Feb 2018
"On a day when the UKs air traffic control borked the Reg has nothing to say."
Oh, that's a shame. Tell you what, here's a different article that looks interesting instead, published eight hours before your comment:
UK flights disrupted by 'technical issue' with air traffic computer system
I have no numbers on it, and I'm sure they have a great incentive to exaggerate. However, I see it used on so many sites that don't need it. A lot of small companies or organizations seem to think that it's the perfect CMS for their websites, and I've downloaded firmware update files from a wp-content directory about two hours ago. I have two sites with it myself, although technically only one of those is mine, the other being one that someone else designs and maintains and my job is to make sure the server is still paid for and not broken. It is very common on small sites, both personal and business, so whatever the number is, it's probably quite a lot of domains that have a WP backend somewhere.
I think about this sometimes when I find sites that haven't been updated since 2004 but are still available without going through the Internet Archive. Sadly, this isn't very common, but I do stumble on such things from time to time and start to wonder how the author kept it going. Of course, the simple answer is that they are still alive and pay for the domain renewals and somewhere to host, but if they were doing that, I would assume they'd update at least a few things on their site. I'm thinking about personal sites that use ancient presentation tags that no longer work, partially because they were browser-specific, or sometimes it's a person with their personal freeware that "runs well on Windows 95, 98, ME, and probably 2000 but I haven't checked yet". Sometimes it still would run on Windows 10, sometimes it is broken on everything except a VM, but either way the files will make it obvious that they're the same as they were decades ago.
Since the sites I run will go down as soon as I stop paying certain regular bills, I kind of like the idea of one that's designed for more longevity. I wouldn't pay this much for WordPress to try to do it, though. My cheaper solution has been to make sure the archives are aware of public sites that I want to archive in case the proverbial bus comes for me.
The problem is that, if you put people in places where nothing grows, you will be spending a lot more resources doing things like delivering stuff to them and insulating them from the reasons that stuff won't grow. Large cities in the desert need lots of water, which obviously isn't there since it's a desert. Some ones will be lucky enough to be large cities in a desert near the coast, so desalinization is an option. However, with a country like the US, there is a lot of desert that's not near the ocean, meaning that if you put too many people there, you'll quickly drain the water that's nearby. A few large rivers, when people try to use them to provide water for millions of people and, where possible, agriculture, is going to cause serious damage to that river. You also have increased risk of fires in a place like that, and it is difficult to prevent fires from being dangerous or to keep rebuilding stuff when the fires have destroyed it.
At current levels of food production, do we really need to use all arable land for agriculture to the extent that cities are preventing us from doing so? There is more arable land in other parts of the continent which isn't farmed for various reasons, including environmental concerns and more complicated transportation routes. This is only relevant if we're building entirely new cities, since most existing cities were put on arable land because, when they were founded, they were getting more of their food supply either from the surrounding land or near the river that served as the primary method of long-range transportation, and now that there is a city there, it's not worth moving it.
"I'm even less impressed by handwaving appeals to res cogitans to try to claim human cognition is magically different."
It doesn't impress me either, and I've argued against it on multiple occasions. I question whether you correctly parsed my comment, as I didn't say that or anything like it for this case. I don't even have to argue by neuroscience here. My argument is purely technical and concerns what we are capable of doing as independent organisms and what an LMM is capable of doing.
At a very basic level, we are capable of taking an action and seeing what happens. We can drop an item and notice that it falls, and from this arrive at a guess that every item will fall when dropped. A bit more experimentation and we can start to guess that items will fall faster if they're more compact and start to create ideas about air resistance. An LMM is not able to drop any items or watch what happens, so any information it has about what objects do has to be told to it. It is able to answer questions about kinematics only because someone has written them down and it is repeating what someone else said.
Similarly, with my discussion of whether LMMs reason (they don't). This does not mean that a program is incapable of reason. I think that, if we got a computer powerful enough and a program designed to make more general logical connections, it would be capable of making generalizations the way we do, and we definitely do as it's one of the core parts of education. LMMs, however, are not written to do that. Their only purpose is to write, and they have not been set up with logical constructs that are more advanced than writing likely passages of text. I am not saying that human brains are more powerful than computers in some way. I'm saying that human brains do things that this program was never intended to do, which is why our brains will be capable of things that this program never will. Give me a new type of program doing different things in a different way, and that could all change.
No, we analyze correlations between a lot of things, not just tokens. That's why we are more capable than LMMs. We can make correlations between actions and consequences, intentions and results, processes and products. It can only make correlations between chunk of text and other chunk of text. We can also experiment on our own to gain new correlations, without having to have a new training run feed it to us from someone's exogenous information. Oh, and we have the ability to generalize which it mostly cannot; this is why a lot of children can solve any mathematical equation using operations they know, and any computer can do it extremely quickly, but GPT cannot solve a simple equation if it hasn't already memorized the answer.
That is certainly optimal, and it's common as courses get more difficult, but it is sometimes difficult to find a new problem that can be solved by students that haven't learned much yet and has never been considered by anyone on the internet. The basics of education will unavoidably be things which are known and discussed by lots of people, and students will need to get past those levels in order to get to the places where they can be set more complex problems that haven't already been solved.
I think it says about as much about the subject as the program, since what it says is whether the solutions to problems asked at that level are likely enough to have already existed on the internet. At low levels, the answer is yes for most subjects. A computer science question asked of relatively early students like "Write quicksort on arrays of doubles in C++" is a question where someone has written that code and published it on a page with most or all of those keywords on it, thus it's relatively easy for a program that has mashed up a lot of websites to find and reconstitute something close to it. A similarly difficult history essay is likely to be on a topic that has already been discussed by some people online, maybe even in these forums, and the bot can copy from that. A political course that focuses on modern events means that the bot can mash up some newspaper articles and make an essay that approximates an acceptable response. A mathematical problem, made from randomly chosen numbers, is probably not written on a website since the sites will have been written with instructions on solving such problems in general, not that one in particular. Hence, the bot can't solve it, since it hasn't been built to reason or to understand, just to repeat what the internet generally says in a format that looks like natural language.
"Doesn't that then immediately link an anonymous site to an entry in a global unchangeable ledger"
In a way, but three factors limit how bad that could be for them:
1. Wallets can be created easily, so in the worst case, it's linked to an empty wallet. Some Tor sites already accept donations through cryptocurrency, so known wallet IDs aren't that rare as it is.
2. There are some cryptocurrencies that are designed not to make public information about previous transactions, so they could use one of those.
3. It wouldn't be very hard to make a pool of a lot of different people, including but not limited to the hidden service. If you mined in the pool, you'd know that one of the twelve thousand wallet IDs in the pool belongs to the operator of the service, but not which one. The pool wouldn't have to know either; as far as they know, they get hashes from someone and pay out, but they don't know who did the hashing.
I'm still glad they didn't go that way. It would inevitably lead to more abuse and, by using a more complex system, there would be more capacity for vulnerabilities in a protocol that already has a few but to which no alternative at a similar scale really exists. Oh, and it would be more unpleasant for the users.
This isn't about exit nodes. They are not that easy to DDOS since, just to work at all, they have to be quite large with a lot of bandwidth. If you're going to DDOS them, you'll probably have more success doing so by flooding their open side through the internet than trying to swamp them through the Tor side because you can deliver a lot more packets the former way.
This is talking about hidden services which are only available through the Tor network and, because their bandwidth is restricted by the relatively slow network and the costs of creating circuits, they are more vulnerable to DDOS attacks running through Tor than an exit node would be. This challenge mechanism is being suggested as a resistant measure that those services could employ.
That may be true (before I go farther, I know neither Catia nor Solidworks, nor any CAD really). However, that information could be passed through to the manager for them to rank applicants. Perhaps they got several applicants who all know Catia, in which case Solidworks guy is probably not going to get invited. If nobody with existing Catia experience is available, they might be more willing to try someone with less experience than they want, but more experience than not hiring anyone. HR doesn't know how easy or hard it is to pick up a skill, nor do they know what related skills convey valuable information. The person who wants the job filled should understand those things and can better parse the available candidates for them.
While I don't know the word, it would seem to be quite distinct from "asylum seekers", which would include anyone who had requested asylum, no matter what state that application was in. This term would appear to be limited to those who applied for and were granted asylum. I don't really have a word or phrase to describe that particular set that's shorter than the eight I just used, but it's still valid to distinguish them from all seekers.
I don't know whether that is supported, but it really can cause problems. If you're planning to download a 110 GB file, you don't expect it to finish any time soon, so you might not bother watching the download. If you come back a couple hours after starting and realize that it stopped after twenty minutes and has been doing nothing, you now need to restart it and keep checking back to see if it happened again. I've experienced this once when trying to download a file about 700 MB from a server that did support resuming downloads but kept failing for some reason every 12-18 MB. Continually going back and pressing the resume button got annoying, but it's not like I could just write a script to pull the partial download from Firefox and do it there. I kept considering whether I should write a script to do it automatically and start over or to keep switching windows to press the button every five minutes (it wasn't a fast server).
"One could argue that a speed limit is a limit."
Yes, which is why you have to be careful about checking what they said was unlimited. Technically, the speed is going to have a limit if only because the hardware they've connected to it has a maximum cap on how fast it can go. If I have a 1 GBPS line coming in, it won't matter if I have 10-gig capable hardware, the bits won't move in or out of my network that fast. For ISP contracts, the thing most likely to be advertised as unlimited is data consumption, not speed, and they'll often have something in the contract about how speeds may be decreased after a certain amount of data transferred depending on load. I consider this to be a valid limit if they don't throttle too aggressively, because they haven't removed my ability to move data through and because they told me at the start. I take more exception to ISPs who have an "unlimited" thing that still has a data cap somewhere after which either they charge more or degrade the service to uselessness, which I last saw with a mobile provider who had an unlimited plan where unlimited meant 15 GB. This was a few years ago. I'm hoping they got hit with a false advertising complaint.
Dropbox's unlimited appears to really have been unlimited, but now that they are adding a limit to it, I think they need to find a new name for it.
They're very common for nearly all SBCs. It is not at all unusual to get an image that's pre-built for some SBC, especially one that's built into another piece of hardware, which comes with an OS, software, and the drivers for that hardware rather than instructions to install the drivers and software from somewhere on a basic version. Of course, they usually make a compressed image and expand filesystems to fill in the rest of the space so you can use as small a card as possible. Still, an image that big is possible if there's a lot of stuff or they didn't compress it.
There is at least one system I know of for getting SD card images which are actually that big. There is a program called Kiwix which provides compressed versions of websites like Wikipedia that can be read offline, and they have a version which turns a Raspberry Pi into a hotspot where people can connect and view the content served from it. The images with Wikipedia included are around that size, though they're not distributed through SharePoint.
That would be my guess. Chrome OS and Android are pretty clean if you use the standard image, except you can't use the standard image outside a virtual machine, so in both cases you're stuck with whatever mess the manufacturer put on including any bloatware or malware they decided would come with it. I'm guessing money changed hands for that, although with some preinstalled software I've seen, I start to wonder whether companies are really getting much value from having their apps preinstalled on cheap phones or laptops.
No, the people who don't stick to the rules are a different category who can be hit with the terms of service they violated. Even after that, an unlimited program will have some people using a lot more than others, with the possibility that people following the rules and using more are being subsidized by those who opted for unlimited but don't use that much anyway.
In Dropbox's case, resale of storage is probably already in the terms as not allowed, and unrelated individuals pooling storage could be added if it's not. People who want to store 50 TB of their own data and bought the plan that said you could do that are not breaking any rules, even if Dropbox now wants them not to. Dropbox can change the products they offer, but they shouldn't call that kind of user an abuser because they chose a product Dropbox offered based on what Dropbox said you could do with it, then used it in the way they wanted and the product let them, and have not broken any rules. If Dropbox was still running an unlimited service, they would probably lose money on that person and make money on the person who bought unlimited but only stores 4 TB in it. They probably knew that at the start but didn't expect that they'd have so many users using lots of storage. Instead of trying to crack down on those who are using it for explicitly disallowed things, they're just ending the unlimited service (the article makes it sound like it will still be called unlimited though, which I hope they don't do).
It depends. What was in it? If it was an image that came with a lot of big files, then it's inconvenient to distribute it as one big file instead of multiple files you could put together yourself, but the size might have been unavoidable. I have seen system images around that size when there is really that much data on them and they find it more convenient to distribute the image as a whole rather than instructions to recreate it and hoping that users will follow them faithfully. This is especially the case if a specific version of Linux is required because you know some users will try to run it on whichever version they already have or if a lot of building from source is needed.
If not, I'm guessing it was someone who has a disk of a certain size, likely 128 GB with a recovery partition of some type on it, who just copied the entire thing and didn't bother using a better image format or running it through a compression program before uploading it. I would hope that anyone who knows how to do that also knows that xz exists, but there are no guarantees.
I also take exception to the term. I get the point and it's valid to some extent, but it isn't abuse if they agreed to provide it. What they really mean is "more than we were expecting". Changing the product based on that is reasonable, but calling people who bought and used a product as specified abusers is unreasonable.
This is, however, why I tend to be a bit nervous about unlimited packages. I understand how much I could use one of those and if the provider hasn't specified some limit* to it, I start to wonder whether they really planned for people to use them as much as possible and whether they'll find a need to change things when people do.
* There are ways to limit an "unlimited" plan without immediately being contradictory or hypocritical, such as the "unlimited data" internet plan which eventually limits the speed but doesn't cut you off or charge you more.
The owner may have more of a stake in the survival of the business, but nothing guarantees that that will be compatible with running the business well or treating employees correctly. I get why the stereotype makes sense, but small companies also often do things like assuming they can get away with not paying for things they require and trying to put a lot of responsibility on some person since they pay them, often badly. At least a large company is more likely to realize that they really do need a backup system that's not built during spare time out of old parts and that, if they didn't back up a computer, it's not the IT guy's fault. Smaller companies will make that decision by personal whim, so there's much less of a guarantee about it. The same is true for firing people as this article discusses. A large company is subject to investors wanting a higher share price and may fire people unwisely as a result, but if the owner of a small business doesn't like the way the profit statement looks, they may also take a similar cost-cutting approach. Their ability to fire at scale is lower, but that will make little difference to the people who end up in the affected group.
Your example of Musk doesn't make much sense to me, either. A person who has just founded a company has to work hard and plan well or they'll lose all the money and time they've put into the business, which could likely be catastrophic to their life plans. Musk can also lose money, but going from many billions to fewer billions isn't really going to prevent him buying more stuff, so he doesn't have that incentive. There is certainly a similarity in that both are ruled by personal whims, but with a small business they're whims that have to bend to pragmatism whereas for a billionaire, there is no such requirement. Also, the companies Musk owns aren't small, even if they're not as large as some companies out there. Musk isn't working in a place so small that he knows most or all of the employees there.
Of course it's easy. It's also pointless. If someone likes having a wider window, breaking up the lines just makes things harder. If they compress the window to 60 characters for some reason (portable device with narrow screen, maybe, or so they can have another window next to it while still seeing the message somewhere), they're either left without the right end of every line or the lines split in two and leave a lot of pointless whitespace. For decades, we've had software, in CLI or GUI, capable of putting some text in whatever size of rectangle you wanted it. We should allow people to have that without structuring our text such that only one configuration is practical without them writing a program to undo what we did. Even though I find writing such programs quite easy and have done it repeatedly for lots of different formats, I don't think everyone else should have to. This is especially because there are many people who don't know how to write something that simple but are still fully capable of using email, and it's our job as programmers to make software that doesn't require every user to write their own code to make it work well in configurations that aren't the same as our own.
"More to the point, why doesn't Outlook provide the option to read the thread in the correct order?"
Because email is not a structured format. Yes, they could take a guess at it and try to reorder the message. However, it won't always work. For example, for the various clients that append the older message to the new one, the dividing line can look like anything. I remember seeing the following dividing lines:
----- Original message -----
On Tues, 22 Aug 2023 12:34:56, somebody <somebody@example.com> wrote:
In response to: [lots of headers for this one for some reason]
Re: [subject] [nothing else, really I have seen it]
As well as translations of those lines into several languages, which I have also seen. By far the largest group of those are the ones that have the date because you have every combination of language and date format to choose from. Any client that tries to split up a message on those lines will eventually fail to find one of them. The other problem is that a lot of clients don't append the whole thread, and that's before you introduce automatic mail handling software which almost certainly won't. If a certain email only includes the last three messages from the thread, reading it in oldest-to-newest order won't give you the whole thread because you'll need the earliest message, but for space or organization, they will have cut it off. Someone may also reply to any part of the thread. This should be obvious from the posts here, where our nested thread interface tracks which reply button someone pushed but means that responses to the top level aren't always near the thing they're responding to because we're effectively doing a depth-first read. If someone deletes a comment in the middle of that, that branch of the tree can stop making sense. Mailing lists often keep online archives in part because it is the way to ensure you find the old messages if you need them.
We could make a new format for the contents of email to change this, but while we leave it as a single blob of text in which any program can paste whatever it likes, we won't be able to change the format when the source may have been generated by any program in whatever way it wanted to.
Forum threads and emails are quite different. When I come to this forum for the first time, I haven't read any of this so I want to read from oldest to newest (in thread order). However, in an email thread where I have already read the rest of the messages, I want to see the new stuff. If I need to refresh my memory on something from the archives, I can find the old message I already read in your reply or when it originally showed up. For a lot of lists that trim the replies, as the article suggests, I'm only going to get the new message anyway which works the same as top posting. I even do that in these forums; if I'm returning to a topic that has some new posts but I've already read most of them, I will sometimes change the order to "newest" using the handy selection box at the top of any list of comments so I can quickly read the posts that were added. Of course, if there are a lot of them, this won't work very well. The ideal email setup would be a structured format where I can order messages in whichever order I like so if I want to read only the newest message or the whole thing in chronological order, I just hit a button, but unfortunately our standards haven't gotten over the idea that an email is a single blob and parsing semantics from it is the human's job.
At least the subscription to Spotify is something that can be used and you presumably enjoyed what you got to do with it. An NFT combines the weirdness of collecting something physical that doesn't have any real value except maybe someone else also likes it with the added benefit of the thing not being tangible and anyone who wants to have one can, just not with the collector's stamp attached. I'm surprised any of them still have nominal value.
Yes, he probably was. I take a more limited view: you should do things that aren't perfect answers, but you really shouldn't claim that they are perfect answers or someone will take you at your word and be disappointed when it doesn't perfectly solve the problem. Of course he should have contributed to solving this problem, which I'm quite sure he never would have done no matter how beneficial it could have been. It doesn't change the fact that the person he was responding to was inaccurately stating the problem and its solution.
If you read the rest of my comment, you will see that I said the same thing. I think that, if he was presented with evidence that solving world hunger was actually that cheap, he would find some other reason why he didn't have to keep his pledge. The person who couldn't demonstrate it was able to name a lot of useful things that such a donation could do, and that clearly didn't sway him. I'm not defending his selfishness, but I think it is valid to understand that he was not promising to donate anything, thus his failure to donate, while caused by selfishness, is not what it is painted to be.
I hate to defend him in any way. I'm literally feeling a bit of revulsion typing these sentences. However, he didn't exactly promise that and he had a point in doing what he did. Someone made the suggestion that a payment around that level, I think it was $6 billion, could end world hunger. Musk responded claiming that he would pay for it if they could prove their point, but they, perhaps unsurprisingly, were unable to substantiate the hyperbolic claim that you could eliminate famine and all its causes with that much. I don't think he ever intended to pay it, even if it could have permanently ended it, but he was making a point in an argument, not a promise to donate that amount.
His selfishness aside, I think the general act is logical when someone claims a problem is much easier to solve than it is. Those who like to simplify problems to the point of inaccuracy make it harder to really solve them. Like a cryptocurrency adherent who thinks that adopting Bitcoin as legal tender will eliminate fraud, someone who thinks that a specific piece of software will mean that there will be no more security vulnerabilities, or a politician who says that electing them will mean nothing can go wrong in the country, we shouldn't assume them to be correct. The claim that was made about hunger is equivalent to telling you that you can solve the problems of someone living in a civil war by paying them £50. If that worked, you'd probably be happy to pay it, but you know it won't actually fix their life up. This doesn't mean we should avoid donations, and in fact knowing what can really be done is a reason to donate more if you can, and we owe it to ourselves to understand what we're really doing instead of assuming problems are easier than they really are. Failing to do that will likely result in people giving up when what they were told would solve it proves insufficient.
I agree with you. At least previously, it was possible to see what someone had posted without logging in. This was <del>useful</del> used by me as I didn't have an account but sometimes people would post links to Twitter posts to show what someone had said. In my experience, it always took a long time if I needed to read more than one tweet at a time, but it was an option. As such, it would seem that blocking someone wouldn't stop them from viewing what you'd written very long as all they'd have to do is open a private window and view it anonymously. Maybe I'm missing something, but otherwise the block and mute options sound basically the same from harassment potential.
It looks like it will affect everyone, not just those using accessibility tools. If someone included a link and didn't provide a label of their own, it would show the title of the destination page. Now, it's going to be some opaque mystery link which probably shows a shortened URL. People can either click through to go who knows where or get the link and use the various tricks to get the shortener to tell you what the long URL is. I wonder how many people know those tricks and how many people know that it's useful. I only know the methods for three of the shorteners, and one of them is tinyurl which I don't think people use too much anymore.
My detector is a bit more sensitive, often making a guess in 5-10 words. I'd say I'm 90-95% correct, but I do occasionally glance and see that it's someone else and the comment is worth finishing. I find it a bit weirder when people respond to such posts. If they're really new, maybe they didn't recognize what it was and are really trying for a conversation, but when they've been here long enough, you'd think they are aware what this is. Some of the responses try to respond as if the post was trying to make a point, and some respond to question why it didn't make any sense, but neither is getting anywhere.
Yes, and some of us might think that this was not a good thing. Part of living in a democracy is that certain rights shouldn't be temporary, if we feel like it. However, even in that situation, the UK during the world wars had an existential risk, which it really doesn't have now. If I had to argue in favor of their actions at that time, I would say that the emergency procedures when bombs are falling on your citizens every night should be very different from non-emergency procedures when basically nothing like that ever happens. In my real opinion, I think the UK probably would have gotten through the wars equally well had it left its citizens' mail alone.
Of course they are part of it, but that doesn't stop them being isolated. A lot more stuff passes over the border between Russia and China than passes across either North Korean border, even though they do have imports and exports. Outside that block, there are ships from China going everywhere in the world nearly always, and ships from Russia did the same until recent sanctions, and even with those, there is still much more traffic from Russia to the rest of the world than there is from North Korea. North Korea's isolation isn't about having no friends, but about not interacting very much with anyone, friends or otherwise.
"Add a little bit of "AI" to the mix and you have the risk that copyright in a movie or tv production could be challenged on the same basis: that humans didn't do the whole thing, only a part."
That won't work that way for three independent reasons:
1. Generally, if humans did some part, that part will be covered. If the film as a whole is in question, they can point to the human-made parts which will be copyrighted and use that to prevent anyone copying the film legally. The protection they're looking for from copyright is usually either to prevent people from making copies available for free or to prevent someone from using their characters and setup in an unauthorized sequel. The former only requires that enough of the final film is copyrighted, and the latter requires that the starting kernel of the script is, both of which are easy even if part of the dialogue is not.
2. Studios will not be trying to claim that an AI program itself is the copyright holder. For each part that is automatically generated, some studio employee or contractor spent time running different prompts, judging the result, and editing the final thing. Those actions are creative, so even if the output from the computer isn't copyrightable, they'll have a claim for why they did put human creativity back into the process and should get the protection anyway.
3. They are copyrighting a lot of things, and they're not going to tell someone how each one was generated. Copyright already applies to things generated with computers, such as animation models and the video created from them. It will not be easy to identify those parts that were specifically created by a program without sufficient human input. In order to challenge their copyright, you'd have to be able to say that this part doesn't count and provide reasoning, but they won't provide you the information necessary to prove it.
That's somewhat true, because everyone in IT knows that, even with two servers, there is some chance that something will affect them all. However, it's less difficult to find someone who says they're reasonably confident that they can make the server stay up without having a second one, especially if the manager keeps reducing the number of facts they tell them about how critical the thing is.
This assumes that they knew about that and that they didn't mind losing their jobs that day. If they created the software, made the test systems, then handed it to someone else to deploy in production, they wouldn't know whether the servers were present. If the person who received the task of putting it in production is aware that there will be no backup server and assumes the developers were informed, they may assume that it has been handled in some way. It's possible nobody was in a position to make that ultimatum. Which brings us to the problem with ultimatums: managers who like commanding don't like being threatened by employees, and employees know it. People who aren't ready to leave tend not to be that blunt with people who won't listen, so it's more likely to have sounded like this:
Manager: I've decided we don't need the two servers. One should be enough and save our budget.
IT: The system is designed for two, so I suggest we go along with those requirements.
Manager: I've decided that we won't. Make it work.
IT: If something goes wrong, this might have safety implications.
Manager: Then do your job well and make sure that server stays up.
IT: We can't guarantee that, though. Two servers would
Manager: I can find an IT person who will guarantee it. Should I look for them?
It depends on what they're supposed to be redundant to. A UPS and generator are not supposed to be redundant to mains power for the entire office but is designed to be redundant for the important servers. Other roads are generally designed to be redundant for emergency vehicles and average traffic, not all the cars that can go down a larger road. If it's designed to be redundant for every purpose the original one was used for, then you're right. If not, you may not be.
It wasn't comparing 3G to wired networks, but to the previous services. I didn't have a data-capable phone at that time, but I remember people trying to use browsers on phones with 2G services (GPRS, I think, so not really 2G) and it never worked, to the extent that I didn't see any need to get a phone that could do it. Nobody ever used that except as an attempt to see if they could. With 3G, it certainly wasn't fast, but it became practical to retrieve at least some information when not connected to WiFi. It wasn't fast, but compared to what came before, it was quite an improvement.
And what will those be? I get that 5G standalone means even faster speeds, which is great, but I can't think of any products where someone said "It's just impossible with 4G speeds and we can't possibly get any faster network there", where 5G makes a serious difference. As it spreads, there will certainly be more use of the system and new things will show up, but I don't think those were impossible using previous technology we already have, nor will 5G have jump-started them.
That is basically what they're doing, although they've chosen to do so by putting weird requirements designed to be impossible to meet in the license and then allowing someone to pay a fee to get out of them. Either way, the license no longer qualifies as free or open based on FSF and OSI definitions because they're restricting on the basis of a user's business. I support that limitation of free and open.
I dislike such terms because one of the major benefits of an open source product in my mind is that I'm not subject to an Oracle-style audit of what I'm using it for, how much, on what systems, etc. This means that I don't have to fear that lawyers will want to see a list of systems with it on and descriptions of their use or that programmers will insert license protection or auditing code to check on it. I don't have to send them an email asking whether my use of the software to provide a service for a charity for which I'm volunteering, not an employee, and it's running on my systems, not ones they built counts as production use or running it for someone else. I don't have to wonder whether a specific use means that I owe them a fee or negotiate how much that fee needs to be every time I change the plan for how much I'm going to use it. They can do it, but it will make them not open source anymore.
I can't say I agree with your suggestions. Open source software means you are intentionally forgoing a lot of the nice options that come with proprietary software you own and can use as you like. It has always meant that, it should always mean it. In turn, it also excuses you from a lot of the problems that proprietary software tends to have, like your customers thinking you owe them support and having consumer protection law to use against you, and it also tends to bring in a lot of users who appreciate the software and contributors who improve it. I'll respond to a few of your points in particular:
"From what I remember of Elastic's licencing model change it was basically "We're not getting enough on the premium / managed service, and AWS is riding on our coat tails taking alot of this business without any financial compensation coming our way" Granted that there was no obligation for AWS to give Elastic anything, but elastic still has to survive."
The article addressed this. Elastic was worth about $14B. This is not the picture of a company that just can't survive because mean cloud providers weren't handing over more cash. AWS didn't give them money, at least as far as I can tell, but they did contribute code changes which are expensive for Elastic to make. They did get value from the relationship, and as you correctly said, they chose a license where nobody had to give them anything and still managed to be successful. Of course, they have a right to do what they did, but it's not going to make me think well of them.
"Perhaps some of the open source projects should have non-mandatory suggested licencing costs to bring social pressure on some companies to pay a fair amount for open source projects."
Sure, they can and do suggest donation amounts. It doesn't really change what I choose to donate. I donate based on my interpretation of the work they do and the value I receive, and one of the great things about open source is that I don't have a responsibility to pay their costs just because they feel like it. They can't claim to be surprised, since this has been clear since the 1980s with tons of examples in the past. Oh, and one other thing I tend to consider before donating to an open source project is how well they're doing already. I have limited funds, so if I have a choice to donate to projects with a small budget or ones that have billions, it's usually the small one that gets my cash. This isn't hypothetical. I have a few regular donations to open source projects, and they tend to be projects where the contributors do a lot of poorly-paid work which my donation can help with. Projects where the official organization doesn't do the complex work because some other contributor is doing it or organizations that already make a large chunk of money from their commercial operations don't need it as much, so I donate less or nothing.
There are a lot of methods you can use to try to find your library in someone else's program. This ranges from the easy to check for and resolve: do they include a copy of the library in their program's folder as a .so, .dll, or similar format. Some of them are a bit harder: if they have statically linked something and removed names, do they have similar or identical messages, function names, etc. Some of them are difficult to check for and hide: do their code have machine code that would be produced by a compiler running on your code, indicating that they're using either the same or similar instructions. In most situations, you don't have to do the hard part; if you already have reason to suspect that your code is in there, you can probably find out quickly.
The reason that this doesn't happen often is that we rarely get told about it. If I have something public under a GPL license and someone violates it, I will probably not be told that they're doing so, so proving that they do will fail at the first step of knowing anything about them. That works for a company that wants to avoid license requirements or more frequently a company that doesn't bother thinking about license requirements, but anything big enough will consider their obligations and at least have a legal argument if they're not going to be following that license.
In this license situation, people could always just ignore the new license and use it as before. The problem is that many of these are using software that is popular among large companies, either for their own use or because they rent hardware and people like to run it. In either of these situations, you don't have to look too far for a list of places to check because providers will have to use the name of the software so people know what is available. If you had done this license swap with a program called ExampleDB, you could probably just do a search for "cloud provider ExampleDB" and get a list of targets.
Nothing stops you, and sometimes that happens. However, when it does, it tends to result in stupid fights between different groups. For example, when Amazon started a project to fork Elastic into OpenSearch. Some people assumed that Amazon was being malicious, a lot of people attacked them for making money on renting out hardware on which that software ran, and Elastic took as many steps as they could to make sure the new version of Elastic Stack wouldn't work with that version. This meant that people who used either version was now dealing with a bunch of people who had strong opinions about what should be used and a company intentionally trying to break things for one side (doing so by breaking things for their side as well because changes to one program will always affect its users). So yes, it can be done, but there is a reason why people aren't always pleased about the opportunity.
Equivalent from a product perspective: probably not. Equivalent from an app running at all times, most definitely. Apple doesn't have to do any tricks because they have the authority to tell their operating system to never stop this bit over here. Tiel doesn't have it, hence the tricks you saw. Both of them manage the products they have with the same software persistence. The "questionable tactics" are because Apple doesn't allow them to more simply ask for the permission to continue running but reserves that for their own software. Blaming Tile for something that Apple chose to do (and incidentally could be the subject of an anticompetition complaint) suggests you may be unfamiliar with the similarity or biased on one side.
The attempts to continue running were intentional, because one of the features their system has is having phones look for trackers that have been lost, including yours, so they can point out their location to you if you ask later. I don't mean that you have to like that or even use the product. I don't have Tile, Apple, or any trackers myself. However, if your complaint is about the app trying to continue running, it's useful to know why it is and it is also useful to know that AirTags are exactly the same. The service running them doesn't have to do any tricks to continue running because Apple baked it into the device, but everyone who has AirTags or hasn't specifically opted out of the Find My network has a service on their IOS device that is running at all times, exactly the same way that Tile's app would.
If you can manage it, I'll consider it. At least with that, it would enable someone to take something that's quite large and ungainly and make it smaller. With a television, you take a screen and make it about the same size, so nothing gained. A screen and a battery are not that many things, so it wouldn't be that hard to get those things and put them into a box yourself.
Since you mentioned it, when you've opened them up and retrieved the magnets, what do you do with them? I have a few of them sitting around, and while they're a bit handy if I ever need a magnet that's relatively strong and I don't mind having a really ungainly shape, that kind of situation doesn't happen very often.
That's a safety limit, not a safe background level. Long-term exposure to radiation can expose you to health risks even if the short-term exposures weren't high enough. For example, the daily limit for a member of the public is 1 mSv/day, but for radiation workers, who are more frequently scanned for problems, it is only 50 mSv/year, and at 100 mSv/year, there is at least some statistically significant increase in cancer risk. 1 mSv in a day is safe, but 1 mSv every day is less safe.
If you were experiencing that level of radiation unshielded routinely, you would be exposed to the equivalent of a couple hundred dental X-ray scans each day. It is also about three times as much as a nuclear power plant is allowed to release in a year, which is why they're so safe to be around. You wouldn't want to go unshielded routinely.
Admittedly, so would equipping the airlock with safety systems to prevent that from working, for example by not allowing multiple people to fit and requiring that the person inside the airlock activate the open out function. Old fashioned methods of murder like stabbing or striking are still available though.
