Posts by doublelayer

Why do cloud titans keep building datacenters in America's hottest city?

Re: 4 cents?

Please note that this is not for the entire country, but for one particular state. The reason that Washington, in the northwest has such a low price are a few factors. The largest one is hydroelectric power. Washington contains the last parts of the Columbia river, which is the second largest river by volume in the US. Washington is therefore able to have much larger hydroelectric plants than most other locations, comprising about 25% of the nations hydroelectric generation. It also has a significant amount of wind generation compared to other parts of the country. A nuclear plant doesn't hurt either. That area has more power generation capacity than is needed to serve its population and is often providing power to neighboring areas in the western parts of Canada and the US. Other parts of the US have higher electricity prices.

You're too dumb to use click-to-cancel, Big Biz says with straight face

Re: ever mindful of the limits of netizens' mental capacity

I'm sure you're jesting, but it proves my point. People like Rust, and it's starting to become more popular and used in more places. However, there have been articles posted here which suggest that people should stop using C and start using Rust instead, and the comments on those are not so positive. If we're going to adopt a new system and completely replace an old one, we should have some pretty good benefits over the one that's being replaced because we're going to have to overcome the costs of annoyance from those who are used to and like the old version. Rust versus C has some benefits with memory safety and attendant security benefits. So far, the benefits from binary or other powers of two are just that it would be nicer, with nary a practical benefit nor a convincing problem with decimal provided.

Re: ever mindful of the limits of netizens' mental capacity

Multiple people, including me, have explained why we're in favor of the decimal system. Basically, because it's already a global standard that everyone understands and we've seen what a comparatively minor change, like changing some units, brings up. If we told the entire world that they now had to use binary for everything, they would refuse. I also find it disappointing that a suggestion to use binary always seems to crop up when someone is defending the imperial unit system, which fails to use binary. I'm not sure if you are supporting it, but even if you're not, that's how we got here in the first place, with someone trying to explain how the imperial weight system was better because it was binary, while ignoring that only one of the parts happened to be.

This isn't really that ironic. Most of us are also from a system where we have various standards that we don't intend to change. In many cases, something that is functional and works everywhere is better than something that's nicer and works in only one place. Do we like C-type languages because it's the best way of writing code that could ever be envisioned? No, we like it because it works and is commonly recognized. Do we like Linux because it is the perfect kernel, which couldn't possibly be improved, atop which runs perfect software? No, we like it because it can run on most computers and it provides a stable and consistent operating system. If I built a new OS from scratch, written in a new language of my own design, and I tell you that they are the epitome of logic and elegance and you must now use it everywhere, it's unlikely that will get adoption. For example, I know several researchers in programming language design who have a few languages they are particularly enthusiastic about because the mathematics are so nice when they write their proofs, and the one common feature of all these languages is that nobody uses them in production.

Re: ever mindful of the limits of netizens' mental capacity

I explained two ways to do this. The long way where you try several things and do some calculations, or the short way where you find something that weighs a pound and can be molded into a shape where you can accurately cut it, from clay to metal. Both were possible four thousand years ago. The latter is really easy today. Hence, this is not a problem for us.

Re: Requiring 'simple' cancellation is a difficult standard for businesses to implement

Why are three choices needed here? Yes initiates the cancellation. No returns you to the previous page without cancelling. What does the cancel button do here? Although the labels could have been clearer, the options in Microsoft's save box were required and made sense: you had asked to close, it asked you about saving and allowed you to cancel the previous command to close.

Re: ever mindful of the limits of netizens' mental capacity

And this would be why people manufactured weights in multiple increments. For example, when people wanted to start dividing things by 3, they didn't tend to try to make a simple one-weight balance do it for them, but found something that weighed a third of a pound and used that. Anyone with a balance could even do that on their own: collect a large number of smallish rocks, see how they affect the balance, put them in useful clusters and label what weight they represented. This wasn't necessary even in early history, because if you could get someone to cut chunks of metal, you could manufacture more convenient weights. Of course, that's not too much of a problem today. A lot of things that might have caused a problem for our centuries-old ancestors no longer cause one for us today, so we can have the luxury of something that's more convenient even if it would add some relatively minor inconvenience if we were teleported back to the 6th century. If we ever get teleported back, we'll probably die of disease anyway, so even the possibility of time travel doesn't require us to use old units.

Re: ever mindful of the limits of netizens' mental capacity

Sure, I'll accept that, as long as you also start using hexadecimal or another power of two as your base for every numerical thing. As long as you're comfortable expressing a price as £1F3 and a distance as 67 kibifeet (approximately 20 miles), then we can also divide by the same base. I think the people who complain about switched units are going to have more complaints about new units, even if the names are similar, and a new numerical system.

As for foot and cubit, one obvious problem with them is that they're not at all standardized. Nearly all humans at least started out with ten fingers. Those that did not due to genetic differences are aware that others did. And of course, basically every society that developed a system of mathematics used base 10, on multiple continents, without having it imposed on them. The length of a foot is very person-specific and changes over their life as well. Of course, we did standardize the foot as a specific length, but it doesn't have a connection to the feet at the end of our legs. Had we been deciding to make a new unit system from scratch, there is no reason to prefer a new unit of length to a standardized foot, as long as we started using a common base (maybe that one we've used for millennia) with it. Basically any arbitrary length you want could have been the base unit of length as long as we agreed to use it. And we have an arbitrary base unit which is accepted nearly everywhere in the world, so let's just use that. Other than the theoretical benefit of "The base unit is sort of near the length of your foot if you happen to be an adult male with somewhat larger feet than normal*", what reason do we have to prefer it? You still can't use your feet as a ruler unless you've measured them.

* The average length of a foot of an adult American male was 26.3 cm (approximately 10.3 inches). I don't have the full distribution, but the standard deviation suggests that feet measuring 30.48 cm or larger would be about 10% of the set. If it's supposed to be connected to part of the body, they did it wrong.

Re: How hard can it be?

Admittedly, that would mean a lot of people signing up for free trials all the time, because it's much easier to make new email addresses than new payment methods. Unless they start collecting some other deduplication method, they're likely to have a problem with increased use of free trials by the same people, and if they have a problem with that, only time will tell how they decide to react to it, likely to the user's detriment.

However, you could have the rest of this without having to break anything. You sign up for a free trial, and they can collect a payment method if they want, but they can't make renewal automatic. You have to opt in to that when the trial is ending. All signing up to such things when you're buying something else must also be opt in. Canceling must also be easy, though if they're so concerned about one-click, I'm willing to let it be one click on a prominently located cancel link which leads to a confirmation page where they can put any information they want, as long as the cancel button is located on the top, they show it on the first screen that appears when the page loads and they don't move it later*, and immediately cancels the subscription when pressed without needing to do any other actions.

* And anything else we need to add to prevent them from hiding it. Two clicks is acceptable if both are extremely obvious to the uninformed user. If they find themselves confused by this concept, I'm willing to act as a UI consultant for a reasonable subscription charge which will be no easier to cancel than theirs is.

Google, DeepMind accused of 'stealing the internet' to create Bard AI chatbot

"Either it's legal or not. Doesn't matter if it's Mom or Microsoft."

That's not correct. Lots of laws distinguish between individuals and companies, and others distinguish between some companies and other ones. For example, I am allowed to go out and buy Activision, the big game company, if I can get the cash, without limitation. They're still fighting about it, but both the UK and US say that Microsoft is not allowed to do that. The difference is there because of Microsoft's different position in the videogame market to me. They can have competition risks because they already have a big chunk of the market. I cannot because I don't have any.

"Shouldn't matter if it's man or language model,"

I disagree on the same basis. Not just on arguments that the language model's incorporation of data is not at all the same as a human's reading of it, but also from the level of harm the actions could cause to others. That's often when laws that distinguish between different organizations doing the same thing come in. The problems that arise when I read something are different from when a large program does so because that program will be creating much more output data and will quote much more freely from its input data than would I.

"Either it's moral to do it or not."

Well, of course. The only problem, which I'm sure you know if you've ever been in a debate of morality, is that you will find a lot of disagreement about whether that or anything else is moral and that there's no way to prove that something is or is not. Not to mention that there are a lot of gradations of moral, including moral as long as it's used in the way we imagined, moral but people hate that it is, immoral except in specific cases, immoral but it's going to happen anyway and there's another discussion of the morality of fighting it, and the various options in the big category of "I'm not sure". Your opinion on whether it's moral won't have much of an effect on how existing laws will apply and, unless you try to convince people, won't have any larger effect on what new laws will look like.

They appear to think that, unless I've included a disallow googlebot line in my robots.txt, I am subject to their terms and conditions. That isn't the easiest legal position to be in.

However, if they succeed, I'd like to take advantage of this situation. The following sentences are a contract between the poster of this comment, hearfter referred to as "I" and "me", ant Alphabet Inc, hereafter known as Google, or its successor. I permit Google to store and analyze this comment and only this comment for use in a search engine and the training of large language models. In compensation, Google agrees to provide me with either ten billion dollars or all the shares of any publicly traded companies related to Google. If Google does not agree to provide this compensation, they are not permitted to read, copy, or process this comment.

The plaintiffs could write that down, but since they're suing an American company in American courts for violating American laws, that's not what the court was going to deal with. American privacy legislation, such as it is, does not apply to other people (and if they tried to make it apply, there would be a lot of people hating them), so this case really is only about what has happened to Americans. Europeans, on the other hand, can use their much stronger privacy legislation for the same effect assuming they can somehow get the Irish DPC to do something, although if they delay for long enough, a court case may help.

Ex-Twitter employees owed half a billion in severance, says lawsuit

Re: Nearly 600% reduction in staff and...

That's not how percentages work. I don't think it's a big issue; we all understood what you were going for, but just in case it comes up again. Reducing by a percentage is capped at 100% if the value is restricted to non-negative numbers. Thus, employee count can only be reduced by 100% from its current point to another point if it goes to zero. Money can decrease by more than 100%; if I made £100k and next cycle I made a loss of £200k, my profit decreased by 300%. To divide in two is a decrease of 50%, because only 50% remains.

There are some ways around this if you're willing to get technical and confusing; if I start the comparison at a level and then compare two other points, I can decrease by more than 100%. For example, if a company employs 5000 people in January, 1000 people in May, and 0 people in July, then I could say that their employment decreased by 500% of their May employment level between January and July. As the example demonstrates, that's often not the most useful way to describe something.

Re: Nearly 600% reduction in staff and...

It doesn't really say much about Twitter. If it had completely collapsed, that would indicate they had some serious quality issues with code. Properly written code should be able to continue running without modification for some time, as long as they don't need it to change very much. Of course, they are now testing various modifications, often pointless ones, to see how far they can push it. As I don't use Twitter, I can't say how well this is going, but I have been seeing more reports of things breaking which suggest that it's not going as well as it once was.

Senator trying to force Uncle Sam to share everything it knows about UFOs

Who are clearly aliens. If you believe that North Korea is a country, then the government has fooled you. Don't bother asking which government; I haven't thought that far.

Re: Give aliens some credit

If they find Earth animal protein as nutritious or tasty as humans do, it would make more sense to remove us and simply automate the process of livestock production. Unless they have a particular appetite for human, other animals are more efficient at turning things into protein while not fighting back or wasting time trying to build things on top of the input material.

It doesn't even have to be that close to plausible. The people who want every UAP to be aliens are not very smart, but they are extremely dedicated for no reason, and one thing they've made very clear is that whatever goes against their theory will not be considered. If an object in the air was photographed, then it was aliens. If the mock object has been found, the camera tricks explained, the motivation for the trick completely understood, and the original creator who is the figurehead of the story confessed on tape to making up a hoax, they'll still say it was aliens. They will mostly not talk about all of those things, and if questioned, they'll spend about five seconds coming up with their explanation, which will always be that someone (government/aliens/somebody bad) has made the evidence or forced this person to lie for some reason they won't explain.

If the declassified reports say that a pilot in a secret spy plane was flying above North Korea and saw some unidentified lights, and thirty minutes later the pilot and ground control and everybody identified it as a reflection of an explicable light source off an explicable atmospheric phenomenon, it will still be aliens as far as they're concerned. There might be less of a stigma to those who report UAP sightings if the field hadn't polluted by a bunch of people who insist on ignoring a lot of the ones that became identified and still talk about long-debunked events. Not that they'd be convincing if they only talked about inconclusive sightings, but at least it wouldn't be so obvious they're not even trying for accuracy.

Bizarre backup taught techie to dumb things down for the boss

Re: The old story, Employee is always smarter than the Founder

Yes, these stories are a bit of self-aggrandizement. I won't deny it. They're not really meant as a representative anecdote of every support call we've ever received. I'll point out that we have another column here which is about tech staff messing up themselves, even though the stories are often a bit simple (people don't want to publish the time when they completely broke something massive and it couldn't be recovered the same way a doctor probably isn't happy to relive the time they didn't think of something and the patient died a painful and preventable death), but we discuss those as well. These stories are something that we can enjoy, however, because we have all had to deal with similar problems in support requests, often from people who did not follow instructions or got angry at the person solving their problem.

I have never worked near medicine, so I can't give examples of the stories told there, but I think you may be misinterpreting the stories you describe from doctors. Those stories of a life saved in a tricky situation aren't saying "I was smarter than everyone else because I'm great". They're telling an uplifting story about a positive outcome which was unusual; stories of patients with an obvious situation which was successfully resolved are routine, and patients where they had something obscure and died from it are sad and probably similarly common, but stories of successfully beating the affliction of the patient reminds doctors that they can succeed and make their patients' lives better even in difficult circumstances. While I can't know what is in the doctor's mind when they tell such a story, the general one you've described fits as well with "I succeeded and you can too" as with "I'm better than you".

The IT-themed stories don't have the same character, possibly because we've all had to deal with support cases like this. You have, unfortunately, made up some facts in order to incorrectly characterize the response. For example, you start to praise the founder for building a business which could hire the IT person. I'll point out that there are many founders of businesses that don't actually know what they're doing and use others' money to pay their IT people, and I'll also point out that the article didn't say founder. It said "One of the top three execs in the UK office". Now executives always have the chance to have built their career from something useful, but not necessarily. It's akin to telling you that every director in a hospital in which you've worked must be better because how many hospitals have you directed and the director was intelligent enough to be able to pay you. If you've never found an administrator who didn't understand something or took a bad approach, then I'm happy for you, but pretty much everyone I've met has experienced such a thing at some point. We don't always blame the user for things they do wrong; about two hours ago, I was complaining with my colleagues that our software is generating an error message which is confusing users and that their mistaken actions are entirely our fault until we improve it. That is not representative of every user error we see, and the stories we tell will reflect that experience.

Re: Puzzled....

The email system may have already had a backup somewhere, but that wasn't guaranteed. When space was more limited, and indeed still today in some places, backups are limited to specific places where data is supposed to be, hence why people are told to store important files on network shares or roaming profiles instead of local directories on their computers. It's possible that the deleted items folders on the server were not fully kept but that people kept local copies of things the server had already deleted. Either way, the special backup procedure makes sense for this because, if there was a larger problem with the new software image, the old disk could simply be replaced to return the machine to its old state, something that a routine backup system wouldn't have done.

Re: Using the recycle bin as storage location.

I wonder if that was always the case. Of course, if you know the internal path to that file, you can still run things that are in there, but I doubt that's what they were doing because that level of knowledge about OS internals combined with that level of stupidity about OS basics doesn't make sense to me.

Re: This is I.T's fault , not the Execs

Well, if they're politicians, they're in their position. They had the skills to get there, even if those skills are no use to what they pretend they're there to do. They may also lack the skills to do what they wanted to do for themselves having achieved their position. Still, they had a goal of getting elected and they managed it, so they were successful in that particular achievement.

Re: Training?

Not every certification has value, and many of them exist either to enrich the certifying authority or to restrict the number of people who can do a task. This includes fields like computing and engineering where you can't just certify that I can do computer work; there are far too many subtypes of them, some of which I have skills in, a few of which, if I say it myself, I have particularly strong experience in, and many of which I've never done and could not do without spending some time learning how. Even that doesn't help, because there are some where that learning would be short (there are programming languages I don't know, but I have learned a lot of fundamental concepts and can write in a large collection, so learning another is pretty fast), and some where that does not apply (I do not write blockchain software, and I have not bothered with many of the proofs that have gone into them, so if you want a new cryptocurrency, I won't be able to give you a good one quickly).

There are ways to present qualifications in various fields on paper. Computing is no exception; there are educational papers which can be and are used to determine some level of experience. However, as people here can testify, you can know what you are doing without a degree and you can know what you're doing without having taken someone's certification test.

Re: It could be worse

No, the metaphor broke down, and email ended up scaling to a level that paper never could. After taking something from the in tray, you'd have to read it and put it somewhere else because, if you put it back into the in tray, you wouldn't know for sure which items in it you had read and which ones you had not. Email tracks that, so the inbox became a mix between the in tray and the surface of the desk, the stuff you were planning to work with today, or soon, probably, let's see.

Of course, the quantity of email that was sent also meant that people could have significantly more cluttered desktops than they could with paper, and the find features meant they could actually manage with that. That doesn't make it the most organized, but if you receive enough emails, it can be difficult to create a good self-managed organization system for all of it. My approach where my inbox stays relatively short but I have a few archive folders is not much different. Sure, my in tray and desktop are clean, but there's a big stack of papers somewhere that I occasionally refer to when it becomes necessary. The really important ones are stored elsewhere, but in case someone asks for something that I received recently, I should be able to locate it quickly enough because everyone is of the belief that I should be reading the things they send me, even when I don't need to.

Re: Recycle Bin is not permanent storage

You put some paper documents in the recycling bin, and you won't be seeing them again once they are retrieved. By the time the phrase was there, people had those bins and understood that items discarded there were no more available later on than items placed in other bins, if a bit easier to retrieve in the very short term. Nothing is wrong with the analogy. A user could figure out what was meant by thinking about what the system did for five minutes or by following the analogy thoroughly (the contents are available until it is emptied and not thereafter).

Re: You wouldn't file your important papers in the wheelie bin

Of course the rubbish bin is preferable to one full of water, since how frequently do you receive waterproof delivery boxes? Neither is acceptable, though. The delivery person is human, either possesses rubbish bins or knows many who do, and should be trained on where a package can be placed and where it cannot. In the trash is not acceptable. Neither is in a location that will destroy the package. I find it surprising you seem not so concerned by this.

Tax prep firms 'recklessly shared' your data with Google and Meta – senators

Re: Why do browsers allow this?

I realize I was being unclear. By "both these companies", I meant Google and Facebook, the two largest users of tracking pixels, not Mozilla and Apple, which by reading back I can see I did not make clear. My point was that, whatever efforts Mozilla (mostly just them) and occasionally Apple do to block trackers, Google and Facebook have stronger incentives and a lot more resources dedicated to bypassing those blocks. In Google's case, they also have an motivation and a method to try to prevent it being possible to block them in the future by adding things to Chrome and then to web standards.

Re: Why do browsers allow this?

Some browsers, including Firefox, do try to block it. There are also extensions that are designed to do so. The only problem is that both these companies quite like the idea of tracking all this information while pretending not to be going for personal data, so they will try to fight against the existing protections. It's an arms race between companies who make billions of advertising* and browser writers who don't always get to catch every change they make. The successes in blocking such things is one of the reasons Google pushes so much stuff through the W3C and why their incredible market share in browsers is so dangerous, since they're trying to expand the attack surface of possible fingerprinting techniques for the next time something they are using gets blocked.

* The companies would probably make about the same amount without the tracking, because they're not that good at tracking people with it, but they're afraid that some advertiser would pull their business on the theory that the tracking actually works.

Samsung’s midrange A54 is lovely, but users won't feel seen

Re: Budget phones

While some places have made devices kind of like that, they're generally not as powerful as you might want. The closest that I have seen is the Titan Pocket from Unihertz, which does have a QWERTY keyboard and a small screen (it is still a touchscreen). It's also solidly built and has a pretty good battery life. However, it still runs Android 11, I'm pretty sure, and it's from a manufacturer that only builds a few devices. Depending on your preferences, it may not be the right one for you, but I don't have other ones that are anywhere near as close to what you've described.

LG to offer subscriptions for appliances and televisions

That's great, but not everyone has that option, including people who either don't have a place to put such a line or people for whom the prevailing weather means that, instead of it taking a while during the winter, the winter would basically just freeze them into ice*. That said, I'm not sure what different settings you have to choose on an electric drier depending on the climate. There are a few settings available, but most of those are related to what you just put in it, not where you are, and the default of turn it on, wait for a while, turn it off will work for pretty much everything.

* Yes, I know, you could manage it inside as well, but eventually it makes some sense to use an appliance just as we don't need to boil all of our water over a fire.

North Korean satellite had no military utility for spying, says South Korea

Re: What was this satellite for?

It was 200 km off an island which is part of South Korea, which is within the 200 nautical mile exclusive economic zone of South Korea. It doesn't automatically mean that anything that lands there belongs to them, but "in your area" does typically include that place. Of course, it can also just include any place that isn't someone else's area. If something crashes in international waters, someone who wants it enough can go retrieve it if you don't.

By the way, that balloon was brought down off the coast of the US. That was done deliberately, and the location was within US waters. China did complain about that, but they were well aware that they would not get their balloon back.

I don't know much about salvage law, but it appears that the 1989 convention and those on which it was built don't really concern themselves with who can retrieve military equipment, but who can claim rewards for assisting a ship in danger or preventing environmental damage. Naval vessels are excluded, but not military things of all kinds, and there are also abandoned property, risks to navigation or safety, and environmental points which can be made should someone want to make a legal case of it.

Indian developer fired 90 percent of tech support team, outsourced the job to AI

"Amazon's Echo has been out for years now and its language abilities are outstanding"

My experience is different. I don't own one, but I have used others' ones, and they strike me as the voice equivalent of command line interfaces. Yes, they are pretty accurate at detecting what words I said (I hear people with certain accents don't get that either), but they don't parse a lot of different sentence structures. Those who installed custom skills had to instruct it to open the skills like running a program, I.E. saying a specific name. They never seemed to add new types of sentences that the machine would understand to mean to submit a request to that program. It also failed to understand the sentence "Do not tell me about Amazon Music Unlimited anymore", which it helpfully discussed every time that device's owner requested it to play a song. If the sentence was in a form they already expected, it was fine. Otherwise, it would just reject it and ask you to say something else.

Intel pulls plug on mini-PC NUCs

Re: I thought about it, I really did...

Out of curiosity, what did you want to be running? If you started knowing you would only use Mac OS, then why consider a NUC? If you hadn't decided that was the only software you would choose, what alternative had you considered? If you reject Windows because Windows and Linux because it's complicated (I'm guessing we can exclude most of the other options on the same basis), what was left?

Re: Small Edge compute clusters are ideal fodder for ARM already

I don't doubt that they exist, but I do doubt how economical they are if you're not buying thousands at once. The benefit of other boxes is that you can buy them a few at a time and, if necessary, from the retail market (though that should only be done if it's really urgent). If I am mistaken, I'd like to know what machines you have in mind and where I could buy one or two of them (not at a ridiculous price from someone trying to resell them to someone who got locked in).

I also doubt that the machines you describe support portable system images, simply because ARM has failed to get that even after hundreds of different organizations focused on Linux users, who would appreciate such things, have not provided it. I don't like being locked in to one provider when a cluster of two NUCs and some other X86 box works just as well with little technician work required.

I've dealt with computing hardware specifically built for niche business purposes before, and I have frequently been disappointed. They sometimes have an advantage in repairability and long-term support, but not in price, specifications, or openness. Oh, and that "long-term support" I was talking about means that you can call the support number five years after buying it and they'll answer and acknowledge that the devices exist, not that they will release any meaningful updates after they've made the next model.

Re: Small Edge compute clusters are ideal fodder for ARM already

It depends what you need that cluster to do. A cluster of Pis really doesn't have anything near the computing power that one of NUCs would have. Yes, there are SBCs with more powerful ARM processors on them, but not as much as you'd expect. A computer built around a flagship smartphone chip would probably be pretty good, but you won't find that very often (Qualcomm has occasionally released development kits like that, though not with the best they have, but they don't tend to sell them in large quantities or provide support). The most powerful ARM SBCs that can be purchased commercially are quite a bit faster than the Pi, but still not as much as an X86 box would generally be. The other problem is that many of those boards do not offer consistent software support or production guarantees. For an X86 box, you don't really need production guarantees because you can run the same software on another one, but if you use a variety of ARM-based boards, you'll have to have custom images for each one.

Make sure that off-the-shelf AI model is legit – it could be a poisoned dependency

You've missed a few points in sequence. The first one is that these models are not trying to assess writing quality, they are not grading tools, and they are not intended to judge people on their grammar. They are intended to tell the user whether the text was generated by AI. The fact that they have a significant amount of false positives indicate that these programs are flawed, and it is likely that this is not the only flaw. For example, the big AIs are generally pretty good at grammar now. That wasn't true with the previous generations, and it's not always true with the small ones you can download and run on your own system, but if you've seen the output of GPT4, you should know that, while the content is likely riddled with errors, they will be presented in a grammatically correct statement. A program that tends to mark worse grammar as AI generated may also grade the better grammar of a LMM as human-generated. That's not the only possibility of how this demonstrated failure could mean that the program is useless.

Then, you equate the false positive to a way of indicating competency in English. It doesn't mean that, and the result will not be used as that. A person who fails to make their writing understood will be marked lower. A person who fails an AI check may be expelled for cheating. As I've already pointed out, it's not guaranteed that the only people marked as AI-generated by this program are those who don't speak English well, who I assume you don't care about, so what do you think about the problems that will affect others who are also falsely accused of using an LMM?

Then, we come to the assumption that people will simply not enter an institution where they don't speak the language perfectly. Observation is sufficient to indicate that this does, in fact, happen. They may do it in part to improve their language skills as well as studying the content they would want to. There are some subjects where specific languages are needed. For example, studying computer science is more likely to go well if you study it in English, because English is far more common than other languages in documentation for the tools they will use. There are books on programming in basically any language, but when you want to look up the library docs, an open-source project's readme, or even some larger projects' updated information, you may find that English is the only option. I'll also point out that tools to attempt to recognize AI writing are also used on researchers' papers to try to catch dishonesty. A lot of researchers will publish in the language spoken by many colleagues, which for most of them is one of a few linguas franca, English common among them. I've read many papers written by non-native English speakers, and sometimes this results in different writing styles, but since they have been reviewed a lot of times, these style differences don't often prevent understanding. Punishing them based on an erroneous program, sold by some company who couldn't be bothered to test that their tool worked and who probably lied about its accuracy, is harmful and outweighs any potential benefit of the correct answers that program might at some point generate.

Fedora Project mulls 'privacy preserving' usage telemetry

Re: Stats please

I have a website. It's basically just a way for me to send files to people that have a URL, because file transfer is still bad (XKCD). It offers next to nothing to a public user. I also have a few other websites that do have something for the general public to read. The browser statistics on each of these sites are pretty different, since the personal website is heavily biased toward the browsers used by friends and family to whom I've transferred data recently. One of the other sites even has an interesting bias because it appeals to a specific hobbyist group who tend to use an unusual browser.

One site's traffic is an unreliable way to determine the frequency of browsers, and if your site is small, it's even more likely to have unrepresentative results. That's not entirely true, as I imagine microsoft.com gets more visitors using Edge and apple.com more users of Safari, but at least if you're getting millions of requests, there is some scale which gives more useful data.

Re: Stats please

Read the numbers again. Skew away from Chrome, yes, we appear to. Skew toward Firefox, maybe if we're being charitable, at least it is much higher, but trying to paint 6% instead of 3% as a major change is not so easy. Skew away from Safari? Did you see that the Reg statistics have it as 24% and StatCounter has it as 21%? In fact, we would appear to skew toward Safari at the same level (3 percentage points) as we do with Firefox.

Sarah Silverman, novelists sue OpenAI for scraping their books to train ChatGPT

Re: Analyzing isn't copying

We don't have a great idea of what this model contains within it. The point is that it is not purely an analysis or a copying program, but a combination of the two. You can get it to produce verbatim quotes, including of copyrighted material. That has been covered in articles here and elsewhere; I'm sure you've seen them. That wouldn't have happened if it was purely a lossy analysis tool. What we don't know as well is exactly how much you can get it to quote, because it's set up in such a way that it will not print an entire book in one go.

The data it has read is radically altered from the original format, and some of it has likely been discarded. Neither proves that it is not a copying tool or violating copyright. If I write a program to mash up data into a form that's nearly unrecognizable, it doesn't prevent me from violating copyright if it can be used to reconstitute the data I don't have rights to. If I take a book and discard a few chapters, but then quote the rest, I have still violated copyright. A perfect example is if I take copyrighted music and run it through a lossy compression algorithm. Some of the original sound is no longer available in the file I created and will never be retrieved by someone who only has my file. The original strings of bytes aren't there either because the format is different. Yet, because it makes the same general impression to someone who listens to it, I am not off the hook.

Music also provides a good example of how even small uses of copyrighted data can be a problem. Some people have decided that they like certain sounds produced by other musicians, frequently drum beats. It would seem pretty easy to make more whenever you want, because you just have to get a drum (or basically any object that makes a nice sound) and hit it with something, but still they value them and integrate them into their songs. These beats are very short slices of audio, and they are not the original audio because, when they're reused, processing has to be done to remove other instrument sounds that are unwanted. If you choose to do this, keep around some money because you aren't allowed to do that for free. You have to license the sound from the person who created it, even despite its brevity, your changes, and the fact that your song is likely completely different from theirs. You used something original to them for your purposes, and so has OpenAI. It isn't as simple as determining whether it produces the entire original work. Even if it can't, it could still be violating.

Re: OpenAI could have avoided all this

"my suspicion is that, as these things are being pushed as "good for generating texts for businesses to use", mayhap businesses would also be better off with just the older material as a style guide?"

I don't think they would be. I'm not sure if modern books are very useful, but the modern text will allow the models to generate text of dubious usefulness that at least sounds like modern text. People generally want to read messages from businesses that sound natural, not old, although those old books at least get unnecessary formality down. Also, it would make a more limited style guide consisting of centuries of English which didn't have any particular styles in common, also including translations from other languages which may not always be the best (check out the quality of Gutenberg's translations of Verne compared to modern translations that do such things as getting the characters' names correct). One style in a book which only uses that one is easier to understand than a mash of styles when modern English is expected.

It is my pleasure to inform you that the communication service which you, our esteemed customer, hath purchased is indeed without function as to your particular place of residence. This we have ascribed to the deficiency of a telegraph to be found betwixt your abode and the house of business. When we are granted a personage of adequate knowledge and a horse upon which to send him, the fault in thy resource shall be witnessed and, if it please he which to him is all glory, fixed. I remain your obedient servant, Local Internet, LTD.

The AI arms race could give us the cool without the cruel

Technology will fix this how?

climate change isn't going to be slowed in the nick of time by magic new technology. We have most of what we need there. Instead, it's existing interests trying to slow change by claims of economic necessity. Against that, models of what happens to agriculture, industry and populations with different carbon emission models. That can be seen as an arms race for strategies within an economic model with the winner getting to a sustainable target in a sustainable time.

This doesn't make any sense to me. Sure, adversarial models can be used to improve the quality of one of them, assuming that the models are being targeted at that narrow set of things that can use automatic means to train themselves. Maybe, if our simulations were very accurate, we could use that method to suggest ways to make our investments in climate technology more efficient. What it won't do is convince those who don't want to incur that cost to do so anyway. It will not change the minds of people who are convinced that there isn't a problem, that it isn't their problem, or that they don't want to do anything even if it is their problem.

This suggestion sounds like a dream where running some AI programs against each other will create some effectively no-cost solutions to big problems. It won't. There is only so far one can go to reach maximum efficiency within current limits, and there is a cap on how efficient something can be at all even if we had perfect technology. I, and possibly you, don't think about that very much since the companies and governments we work with so often operate with such a ridiculous level of inefficiency that it seems there is always a lot more ability to improve them than there is a desire to, but the limit still exists. Running a lot of simulations until we get one that looks nice will not change the significant costs of sustainability, and even if we cut some of them, it will not prevent us from having to do the hard work of either paying the costs ourselves or getting those who cause the problem to do so.

Turning a computer off, then on again, never goes wrong. Right?

Re: And Arnie? He will not be back.

It probably happens more often than not, but nobody writes in about the time they didn't find what they expected, walked away, found someone else, and it got fixed. It would only end up here if the second opinion didn't help and the box went down anyway.

However, having dealt with my fair share of boxes with uncertain support, I get why it happens. I've conducted tasks on systems that didn't seem right, but I either couldn't get a second opinion at all or the people I could find didn't know any more than I did about what the system was doing or how. They still wanted it fixed, though. For context, I'm a programmer and work mostly with programmers, so not everyone has as much administration experience as I would like them to, and not everyone on the team is a good choice for disentangling an ambiguous config. My general solution to this situation is backups and walking very slowly through the system to make sure I've thought of several things before I utter that sentence "Well, that's all I can think of".

China's openKylin 1.0 arrives. Our verdict? Not a bad-looking, er, Ubuntu remix

The copyright line is required and is present even when the GPL applies. The "All rights reserved" isn't required and was probably put there on autopilot by someone who has seen it after nearly every copyright line. It has relatively little meaning anyway, but when the GPL explicitly applies, it cannot counteract any of that license's provisions. Unless they've also changed the GPL, it has no effect on what the user can do. If they have replaced it, things get more complicated but we'd have to look at the specific changes.

Re: Playing fast and loose with the GOL?

Almost certainly not. All they have to do to comply with the original license is to include the license terms including the copyright statement somewhere in the system. I'm sure there's a file in /usr/share that does that. It doesn't even have to be unusual to do that, because that is what forks are likely to do; they no longer use the original name because they might make breaking changes and don't guarantee compatibility, so if this is going to be made into a new version by the Kylin team, they might have done the new version numbers to make that clearer. That doesn't guarantee that they have done what they need to, but their obligations are pretty easy to do and don't require them to produce useful compatibility statements.

Let's take a look at those US Supreme Court decisions and how they will affect tech

Re: This is somewhat weird...

The reason this doesn't make sense to you is that the article talks about two separate, unrelated court cases, each one involving different people and a different political point, and you've grafted pieces of the two into one Frankenstein's monster of a case. For example, this theory about what might be happening: "OR is she for/neutral to gay marriage, and "the Harvard College and the University of North Carolina" wanted to force her to write an anti-gay-marriage website?"

The universities are part of case number 1. The website designer (well, perspective designer at some point) is part of case number 2. The two do not connect. The only case that has to do with websites and their content is case number 2. I'm afraid that providing more information may just re-introduce the confusion, but you may want to search out more information on each of the individual cases to better understand what has happened.

Red Hat's open source rot took root when IBM walked in

Technically, just to the person with the binaries. However, that person cannot be forbidden from sharing the source or the binary using that license, so they can then publish it. Additional agreements can be used to get someone to agree that they have the rights to publish but that they will not exercise them, but that is more complex and must be created when they get the binaries originally.

"The directors will consider what shareholders want. And, as we've seen with the likes of Twitter, despite objections from hoi polloi in the cube farms, the management and board insisted on selling."

In Twitter's case, the directors and the board both attempted not to sell, adding extra blocks to the process. It was only after shareholders demanded that they change their mind and get them the Musk payout that they accepted it. Of course, when Musk allowed it to break down, they pursued the sale he had agreed to, but that was also something the shareholders would have made them do anyway. Sometimes, the management doesn't want to make a company crash, but if the people who own the company want the money now, they have the right to get it and to make the management act for that goal.

Now that you've all tried it ... ChatGPT web traffic falls 10%

Re: It will pick up soon enough

I'm not sure they need to actually use any LLM to write stories completely misunderstanding what it can and can't do and making unrealistic extrapolations from that. Using the model opens them to the risk that they actually see the kind of stuff that we, who have either used it for our own curiosity (I plead guilty) or seen the results of people doing so posted here, have already seen, which indicate that accuracy from a model like this is just a game of chance. Not that some people, who are more interested in spending some time talking than on real journalism, can't ignore that fact and talk more about how it's going to take all our jobs, but some of them might not feel so confident about it knowing what will actually come out of it.

BOFH: Lies, damned lies, and standards

Re: Association of Servicepeople for Software and Hardware Over the Lifetime of Equipment.

I think it only means that the boss gets the idea, but doesn't recognize the original acronym. Alternatively, they do, but they don't think quickly enough to come up with an alternate acronym that still fits the pattern closely enough.

From cage fight to page fight: Twitter threatens to sue Meta after Threads app launch

Re: Privacy? what's changed

They may be thinking of earlier versions of Android, where permissions were specified and granted at install time. This wasn't just the first versions of Android. Through many of them (I'm not sure when this changed), it was normal to see the message "Do you want to install this app? It will get access to:". IOS never did that, at least I've never seen it, and Android doesn't do that anymore, but for many years, it was structured that way.

UK's proposed alt.GDPR will turn Britain into a 'test lab' for data harvesting

The comment by EvaQ appeared to be primarily complaining about the use of the word "European" when the actual meaning was "European Union". This is not the first time a discussion like that has started when there is something real being talked about. For example, some people will describe the result of the Brexit referendum as "leaving Europe", and others will kindly reply that it really means "leaving the European Union", and that "leaving Europe" would be much more complicated. As correct as that pedantic reply is, it isn't really doing anything for the original conversation.

To illustrate this, I pointed out that, in the EU's statement where the offending "European" appeared, to be entirely correct about who is covered and who is not by the rest of the sentence, you'd probably have to put a paragraph of extra conditions in place of the one word, but that everyone reading the statement understood with perfect clarity what they were trying to say.

I find this quibbling about terminology rather strange. A lot of arguing over someone using the word "Europe" when they mean "The EU", even though we all know what they meant. There are a lot of terms in English where we could use a strict definition that few others are using despite everyone knowing what was meant, but it's rarely helpful. For example, people or things are sometimes characterized as British, meaning they're from or in the UK. However, technically British means from Britain, so if they're Northern Irish or from one of the non-Britain UK islands, I could argue that they're not really British. It wouldn't help anyone if I did that, because we know that is included.

In this case, the terms are even harder, because what the sentence really should say is "In light of the serious threat to the rights of those who have them under GDPR". Noncitizens resident in the EU have those rights as well. Citizens of Iceland, Liechtenstein, and Norway are also covered because they've adopted it, even though they're only EEA, not EU, members. Turkey and Switzerland have laws that are similar to GDPR in an effort to be harmonized with EU regulations, so they might have rights or might not. And, if we're going even further, there are threats to the rights of British, sorry UK citizens, even though the EU commission has no power or intent to do anything about those threats, they're still there if the law is weakened. When you read a statement, do you want a paragraph listing all these technicalities, or can we agree to accept text when we all understand what they meant?