Posts by doublelayer

Millions of age checks performed as UK Online Safety Act gets rolling

Re: Alternatives?

No, the number would have a link to the person, although presumably a unidirectional one so individual sites wouldn't automatically know from seeing your number that it was yours. The government who verified it would. Sites, if they checked that number among themselves, could still build up a profile, and you couldn't have another number, so that would be a nice profiling method. That last one would be illegal under GDPR, which is enforced so strongly that only 65% of sites would track it. That's the problem with most systems. Either they have a trackable identity connected to them, or they're very easy to bypass. I don't like the system, so I prefer easy to bypass, but either way you go, someone who wants the system to be perfectly locked and perfectly private is guaranteed to be disappointed.

Re: Madness

"Anyone who says parents should parent like their parents did [...] is probably thinking of their parents parenting by taking the TV out the child's bedroom. Tell me how this is done with devices that by definition are movable,"

It's not the only thing that needs to happen, but one important place to start is by deciding what devices the child should have. If that means no smartphone because they aren't trusted to use it properly, then don't buy them a smartphone. Buy them a simple phone, or maybe don't buy them any phone. It's not perfect, but it's a start that, when suggested, some people act like is an impossible idea that they don't understand how anyone could suggest.

"In no business on the planet is the IT department expected to enforce agreed company policies without the ability to administrate devices, software, or company websites, yet patents are told they have to do just that for their children."

Of course they're not. They're the parents' devices, and they can be locked down. In some cases, you might be able to manage without doing so, but there are restrictions. Use them. A lot of complainers have never looked at the options or decided that because they're theoretically bypassable, we'll just ignore those and demand others. There are many management options for mobile and desktop devices, many of them free, and they can be used.

"others will unfortunately have all their friends on Instagram so it becomes necessary for the child to have an account as part of having a social life with other children and for the parent to have a degree of control over what their child does in Instagram [...] This is not possible at the moment."

This is where you see what the various parental options on Instagram allow and decide whether they're sufficient, and if they're not, take it up with Instagram. I don't know what they are as I don't use or administer it, but they certainly claim to have lots of options. But before you rush to legislate, consider that Instagram is not a necessity after all, even if others use it. I had friends who used social media I did not, and somehow I still had a social life with them. If the risks of Instagram are too high, prohibiting it is an option. You have the decision whether it, with parental options enabled, is an acceptable risk or not. It is not our responsibility to lock down everything because Instagram doesn't have a feature you think it needs but you insist on letting your children use it.

One important thing is not to let technological control outweigh all the other important things. Preventing your child from being bullied on social media is important, but if all you do is prevent them from seeing messages, the problem is still there. If they're a target of a bully, they're probably in close proximity to the bully in question. You can do far more good by knowing and responding to that situation, whether that's by teaching the child about ways to respond to raising a complaint against the bully than any social media filter will ever do. When we ask parents to parent, it's often with non-technology solutions which they should use already.

Re: Alternatives?

Let's review this:

I submit my identification information and I get a long number, which I can now enter on websites to prove I'm of age.

When a website wants to verify that, they presumably submit that number to the tax authorities who verify that it is linked to an identity that has created it. That means the tax authorities know my identity and what sites I access, and the only question is how long they choose to store it and whether their systems are secure when they collect it. Privacy gone.

Any website can correlate that number with any other website to know that the same user has accessed that site. You specifically recommend they store and check trends on that number to identify reuse. If one of them has a login, they can now associate that number with an email address. I only have one of these as well, making it a perfect fingerprinting tool because I can't have multiple ones and I can't use their site without one, which is great for advertisers and data traffickers. Privacy gone.

Anyone who finds that long number can identify themselves as an adult even if they're not, and I can claim that I didn't know they did it, so those numbers, obtained from willing or unknowing adults, will be readily available to people who want privacy or children who are willing to put a little work into this system. If sites don't check location, those values should work nationwide until they're automatically revoked by being used a hundred times every day. If sites do check location, you'll need individual numbers for people near where you live, which will make it more annoying for people trying to obtain others' numbers but will also make it more annoying for people trying to use their own.

You are right, there are a lot of holes. There will always be a lot of holes. The system as described is trying to verify that a user is within a set of known people, and there is no way of doing that without knowing identities at some point. Even if I was setting it up with a fanatical attention to privacy, it would still break. I might choose not to log anything and take the substantial legal risks that come from that, but that is not enough. Consider, for example, all the systems that process payment cards. They are not supposed to and usually don't log payment card data, because that would lead to fraud. And yet, they get attacked, because if you have software running on them, you can collect that data when it's processed even if it isn't stored. Mathematical tricks to obfuscate the numbers improve things a little, but only against the basics like replaying the same value, not the rest of the problem.

Tech bro denied dev's hard-earned bonus for bug that overcharged a little old lady

Re: "...dev's hard-earned bonus..."

Unfortunately, most of the time bonus calculations for technical positions and anything with a similar profile are very discretionary from management down, meaning it's very easy for them to set them however they want them. At least in my experience, contracts tend not to guarantee a bonus on something you can prove you did, because it's based on performance to some undefined standard. Some jobs that have a more mathematical method, like sales or finance, can do this differently, but for a programming job, it's usually tied to subjective assessments of performance. That can be helpful, because if you thought something was going to be a quick task and it ended up dragging on, management can decide that this was reasonable for you to not know and give you a bonus anyway, but it can also go the other way.

In some cases, you might have a case. For example, if you had a goals list written six months before, completed every item on it, and they still denied you a bonus, that could be convincing evidence in a complaint. However, most tech jobs I have seen have these lists but end up changing the plans halfway through, meaning that inevitably some items will not be complete because they were deprioritized or even eliminated, and if the management say you were not productive on the new things, that's much harder to disprove in a complaint even if it's a complete lie. The same thing happens with promotions, pay rises, or anything else where it's at the manager's leisure.

Again, you're too happy to assign all responsibility to the boss. Let's try an alternative: I'm hired by a startup who has no tech staff. I tell them that I'm an expert on Linux and can definitely write the kernel module that they're going to make part of their major product. It turns out I'm a basic coder and don't actually know how to do that well. My attempt crashes the hardware, leading to financial losses. Who is at fault?

If the article included a sentence like "Ivan was not familiar with telecoms equipment and had warned the other two members of the company, but they decided that it would be fine and told him to write the software anyway", then that would have changed it. The article does not include that. We don't have anything near enough to assume that exists as you are doing.

In my example, I am at fault for making the mistake, and if the others are, it is for choosing me in the first place. In the case of the article, it's far less clear, because it sounds like Ivan was reasonably competent and happened to have one bug which was triggered exactly once. No matter how competent Ivan was, no matter if he had had a decade in telecoms and was as expert a person as you can find, that could happen. Although I'm in the position of arguing that the dev has responsibility, I'm also in the camp that bugs should be expected and, unless they indicate significant negligence, nobody should be penalized for them. Had they cleaned up properly after that bug, they would have had no problems, and that is why I say that the manager is at fault for the actual problem: the risk of legal action from failing to refund the customer when they made a mistake. But you're going much further and trying to assign automatic responsibility for anything, which is not how that should or does work. You're making up evidence that does not exist to make it more plausible, but your general statements suggest that, even if Ivan was a telecoms expert, you would blame the nontechnical boss for not intuiting that there would be bugs and allowing it to release, which in my opinion is unreasonable and unrealistic.

True, but they didn't have a problem with payment systems. All the testing they needed to do is printing "Would charge customer {customer_id} amount {amount}". The part they needed more testing of was the phone disconnection thing, and I don't know how that works, so I don't know if their issue was down to insufficient testing, testing something theoretical that didn't act like real equipment, or just very bad luck with an intermittent phone system bug. The last one would probably have been harder to patch with more failures, so that seems unlikely, but any of those is possible.

I'm not sure that's relevant. No matter how much testing was done, the important part to legal action would be who could and was supposed to refund that order. I don't know, but from most places I've seen, that's not something the dev is allowed or even able to do. Refunding orders is usually a manual process run by whoever is doing finance or customer management. The biggest problem here was not the bug, but the failure to refund when the bug was detected which was almost immediately. That is the reason I think the dev is not the blameworthy party here. If the dev did have the authority and responsibility to manage refunds, then I'm willing to blame him again because that should have been obvious; I just don't think that's very likely.

Both people are responsible to some extent. The boss doesn't take total responsibility for everything. I'm not sure how this bug cropped up, but if it was deterministic, it evidently didn't show up in the testing they did. Does that mean testing was insufficient? Maybe, in which case it's probably more on the dev than the boss who can't dev/test and hired this person to do it. And the dev was probably the closest thing to telecoms expert they had, and it's not automatically the boss's responsibility to question that. Just because someone said they're ready to turn on doesn't mean they take responsibility for any failures created by others.

In my opinion, using my best assumptions since most data isn't available, the responsibility goes like this:

1. For the bug meaning someone was overbilled: the dev.

2. For not refunding the customer immediately: the boss.

3. Therefore, for the risk of legal action: about 100% the boss, since if the refund was issued immediately, there would be no cause for action.

But in a situation where things worked out differently, if we assumed the existence of a bug that cost the same amount but wasn't mostly due to unethical business practices, that could go very differently. It still wouldn't justify lots of consequences for the dev, because the business has to cover the costs even if the dev is responsible, but it would likely justify losing a bonus, and some businesses fire people for that kind of thing. I don't think that applied to this situation, but your "The boss is paid to take responsibility" theory is unacceptably broad.

I agree that this needed to be considered, and I've written a comment to that effect, but I don't know if this is something the dev is responsible for. That sounds more like a finance thing. From the information in the article, I don't know if the dev knew that the money had not been refunded at the time. If that was not the dev's responsibility, I feel more sympathy for them.

Re: Lawsuit culture

Except they only refunded the payment after people showed up at their offices. They should have figured this out when the call didn't drop and issue the refund at that time. Keeping the money longer could have caused lots of problems for the payer, and depending on that length of time, that could be argued as a deliberate act rather than an oversight, and from the information we have, we know it was one because they had noticed the problem but evidently didn't choose to fix it then. A lawsuit wouldn't have been justified, and the people clearly didn't choose to pursue one, but there is a reason to think they might have been able to had they proven litigious.

China's IPv6 adoption takes a decent leap forward, especially on fixed networks

Re: “Let’s create a world where every interaction with governments is assisted by digital agents”

I doubt it. An LLM can "infer" some of this and can definitely write that down, but it won't be able to make things happen as a result. It's not hard to get an LLM to write a paragraph explaining the consequences of a cable on the ground, but if there is no procedure to report this as an issue, that paragraph is not going anywhere. From the comment, we can be pretty sure there isn't, because a human did hear this, did recognize that something happened, did open a ticket, and still nothing happened. The problem is the lack of response systems for that situation, and whether it's a human or an LLM in front of that, the problem is not fixed.

Microsoft gives in to Chromebook bullies and drops Windows 11 SE

From the post where they said they didn't support it, the user had managed to erase it and install Windows 11 which booted, but their problem was drivers for the original hardware. So yes, you can, and I wouldn't be surprised that those drivers can be found somewhere. The specs of one of these (4 GB of RAM, 64 GB of EMMC, and a Celeron N4020) probably aren't too fun to run Windows 11 on, but there are people selling machines with that spec running normal Windows as it is. Hopefully nobody buys them.

I mostly agree, although I expect most of the destruction is accidental and might not be blameable on a single person. Most schools I'm aware of aren't giving out laptops to every student; they just don't have that funding. If they move machines around from classroom to classroom, loan them to a student temporarily then get it back and reintroduce it, it's only a matter of time before someone drops it, which probably doesn't do much damage. Have it get dropped enough times, have it knocked off a table while connected to a power chord which puts pressure on the only power-capable USB-C port* and makes it land on a corner, and hardware will start to break in a way you can't easily find a scapegoat for.

* I don't know why laptop manufacturers sometimes have a single power port but multiple USB-C ports. I think it must be just to annoy me. Fortunately, plugging a charge cable to the wrong port won't break them, it just won't charge, but I don't like it.

Because the one thing where this is better than Chromebooks, and it takes a while to find it because this is pretty bad, is that these computers are not locked. Read that comment about installing Windows Pro on the machines, and what's the problem? It's not locked bootloaders; it's drivers. You can still boot to other media and install something with the primary challenge being hoping it has the drivers for the important hardware in the thing. Thus, you can arrange that petition, but complying with it will be very easy. If you want a petition that matters, demand the drivers, although I wouldn't be surprised if there is a way of extracting them from the existing SE build of Windows and getting them into regular Windows. Linux and Chrome OS probably won't have that option, but Linux's driver choices are quite expansive.

I was still surprised and disappointed hearing of this because I've helped get other school-focused Windows versions to unlock to regular Windows before, and when I did that, it wasn't hard at all and didn't even require a reinstall. It shouldn't be hard to make that happen, and Microsoft should have done that for all the devices they could having dropped support so quickly. It seems that the main desktop OS companies are in a race to see who can expire equipment most quickly, and I want that to stop. To Microsoft, Google, and Apple, haven't you noticed that Android manufacturers are being dragged slowly into supporting their devices longer? That doesn't mean you can go the other way with laptops because they already lasted that long.

German phone repair biz collapses following 2023 ransomware attack

Re: You paid them

It's probably not illegal for the prosecutors to hold the money because, by the time they took it, it wasn't this business's money anymore. That business had willingly paid it to the ransomware criminals, and it was confiscated from them. Of course, that's cold comfort to the people who thought the payment would go to getting a decryption key, didn't get it, and can't have the money back, and it probably feels more tantalizing because it could be but isn't being returned. I'm not sure the special case makes this any different from any other time when a payment is made and the criminals don't hold up their end of the deal.

Another one bites the dust as KubeSphere kills open source edition

Re: Two wrongs don't make a Right...

That is what I was saying. Back when it was open source, if they had to sue someone for violating that license, they could have gotten help, and they probably wouldn't have needed it. That they didn't do that suggests that they didn't need to, which suggests that they are now being dishonest about why they made the change they have.

Re: Two wrongs don't make a Right...

They have enough money to manage this, or they could outsource this to one of the many advocacy organizations that love to make this point. Or they could just threaten it and see if people will back down then, a move which is much cheaper and works very well with a public statement. So maybe they only did that. Let's find that public statement on their blog and ... I can't.

And your lock analogy is bad because their changing of the license doesn't prevent me from getting access to their code and using it for prohibited purposes anyway. To prevent me from doing that, they would have to do exactly what you think is too much work. Their required steps are not different now, but the number of people they could try to collect money from is larger. Do you really think it's unrealistic to expect that that was the change they cared about?

Re: Maybe there needs to be a new open source licence

You can write any license you like, including that one. It's not at all the same, because both free software and open source have been very clear on a few rights that are not compatible with that license. One is obvious; if you have those restrictions, then you're not getting the right to use the software for any reason. Some others are less obvious but perhaps bigger problems. For example, let's say that I have contributed, so I get free access to the code. Do I get the right to modify and distribute, and under which terms? Can I distribute a version which allows anyone to use it? Can I distribute a version under the original terms, but now they're paying me, because after all, that's who they got the code from? If the original author stops developing it and I'm now doing all the work, do they still pay the original guy? Pretty much every attempt, from ones where they're being clear that they want all the terms of proprietary but with the word "open" in there somewhere to well-meaning but still broken ones like Bruce Perens's "post-open" license or FUTO's "source-first" approach, deals with this problem by stripping those rights too and from everyone involved.

So by all means, write some software and put that license on it. I won't view it the way that I view open source, even if I end up with the higher tier of rights, because chances are that I'll be forbidden to do things that I'm allowed with real open source.

Re: Two wrongs don't make a Right...

If they knew that people were using their source against the license, they have a thing they could do that would earn them money, do something about the problem, and get them praise from fans of open source. That miraculous solution is: sue the violators in court. It's not experimental anymore. We've seen lots of lawsuits about violations of the licenses which can serve as precedent. Unless you've done something really stupid like Neo4J did (specifically, writing their own license without thinking for more than five seconds about whether they were making a legally valid one), then you know you're very likely to win your case, and if it ever starts to go wrong, foundations that don't want open source to die will quickly leap in to help. Rampant violations of the licenses is mostly down to people not bothering to do that, but someone with resources who wants more money from what they developed could easily do that.

They could, that is, if the problem they're talking about was real. That they haven't suggests that maybe, just maybe, they made this problem up. Maybe their real problem is that they want every user to pay for the software and they're unhappy that their own choice, a choice that got them users and contributors, makes this more difficult. It follows the example of many open source projects. The people who run them see that 20 million people downloaded that code recently and think "Hmm. If it cost $100 per user per year, that would mean $2 billion per year for me", without considering that, if it cost $100 per user per year, most of their users would have never started using it and nobody would have given them code to fix problems and that some of the people who downloaded it aren't using it now anyway. These people are not stupid. They understand exactly what they're doing when they put their code under an open source license. There are not two wrongs here. There is only one.

BOFH: Deepfake or just an idiot? We'll need an audit to confirm

It depends what the lock is supposed to lock. If it's supposed to prevent the computer from running anything at all, then it would need a case that's effectively a safe so you can't just open it up and disconnect it. But you could prevent it from booting without a separate device by encrypting the disk and requiring the additional device to provide the keys needed to decrypt it, and a BIOS patch could make that device unclonable by doing a key exchange when first set up and always communicating with that encryption and random challenges afterward. There are a few places you could lock without needing to harden the computer too much.

I recently learned on someone else's phone that Samsung no longer lets you update any of the builtin apps without one. Before, if you wanted an update, you could go to the Galaxy Store entry for the app and update it, and at one time, you could even install things, but now, both of those are gated by a login. Which makes me feel less pleasantly toward Samsung, but that might be one reason people set up such an account.

Re: Very interesting.

Generally, the recommendation is to choose the words randomly, and not your brain randomly. I use that approach for temporary passwords given to others because it makes it easier to type, with the words selected from a list of about 25,000 words which should be recognizable to about everyone. I tend to use four and put hyphens between them to make them more distinct, which results in passwords like this:

Predawn-Conferences-Ranting-Nice

Milling-Thwarted-Plummeting-Accomplishments

Beset-Typewriter-Vividly-Breezy

Mathematically, these are about as random as a 9-character random password like j2#reXq"D which, on the surface, doesn't look great. They're a little better than the typical password though because people with a length limit like that won't choose their short password randomly either. Also, until an attacker knows that happens, it's much more secure because they're not limiting their search patterns. Still, if you need more entropy, you need it to be longer.

Rampant emoji use suggests crypto-stealing NPM package was written by AI

Re: Attribution now easy?

"If the server where the upload happened actually knows the timezone where the uploader was located why doesn't it seem to know the uploader's IP?"

It does, but we don't know that server. What we know is the stuff that got left in the code, and given that this left in comments explaining how this was stealing things, the author was evidently not smart enough to delete such things. What this tells us is that the uploader's time zone was set to something UTC+05:00 when they generated the files that contain timestamps. That doesn't prove they were there, or that if they were they're usually there, but there's reason to give it some credence. That would put them in one of seven countries: Kazakhstan, Maldives, Pakistan, Russia, Tajikistan, Turkmenistan, Uzbekistan

"Attribution -which used to be next to impossible back in the old, non post-factual days- seems to be a standard procediure, nowadays"

And you'll note that they haven't attributed this one. Identifying a time zone is far from attribution. But questioning attribution because it conflicts with one's political beliefs is a similarly common procedure. The researchers who attribute more regularly now have put a lot of work into doing it properly. Those who question usually haven't.

"because pretty much every threat, malware or attack is immediately attributed on disclosure - usually to asia where the state-backed evil lives."

And there we have it. Nobody said this was state-backed. Also, they clearly weren't. It's a quick and dirty cryptocurrency stealer. That's not state-backed. That's someone individual looking for money. But thanks to putting words in someone's mouth, you identify yourself as someone with an agenda or someone who can't read what was said and what was not.

Reddit is people! Which means its search might not be so damaged by AI slop

Re: Have you ever searched for anything on Reddit ?

In my experience, both are true. Searching there was never good, and maybe this will be better though I'm not very sure. The content was another story, because quite often, there is useful information somewhere. The problem was that there is enough useless information that even a good search engine might not find the useful part for you. This depends on your tolerance for reading several search results until you find the place where a knowledgeable person chose to answer it, so it ends up working out some of the time, but it isn't a fast way to get a result, and since it's human-generated, it tends not to be comprehensive data when that arrives. It's a tool that's useful for some cases, but I don't think it will be useful to enough cases or people that Reddit will gain much search market share.

Long live the nub: ThinkPad designer David Hill spills secrets, designs that never made it

Exactly. Acceptable reasons to massively change the interfaces include and are limited to the following:

1. The old one is crap, and we're willing to admit it was.

2. We're adding something which the old interface will make far too complicated to make sense, and we've actually created a prototype of how it would work with that interface, but that would be untenable.

3. We have large amounts of evidence in the form of large stacks of paper or large files demonstrating how this is actually better from the experienced and new users' perspective, created by actual tests or people who only care about usability and tell designers to go away until they're done researching.

"Because we haven't changed in a while" is not and never will be on that list.

Court upholds Epic win in Google Play Store antitrust battle

Re: The security aspect.

That is possible, and I'd care more if Google Play wasn't already lousy with malware. The unavoidable implication of this comparison is that, if they didn't have to allow other stores, malware wouldn't be a problem when people stuck to the builtin store. That's not the case. It's not even the case with Apple's store, although the difference is significant. Google will still be allowed to ban any store for having hosted malware, which means that stores that want to stay around will have to be quite careful. I'd expect, for instance, that Epic will be monitoring their offerings quite closely because they're running their store to collect lots of cash, Google would like any excuse to throw them out, and "they hosted malware" is a really good one. Others will not have the same incentives and will host malware, and those stores will be gone very quickly after the first detection happens. Meanwhile, the normal store will continue to hold tons of it.

Florida jury throws huge fine at Tesla in Autopilot crash

Re: "thought the car would take care of things"

I mostly agree with you. Most drivers who think that autopilot works are idiots, and this driver is even more of one than most because of the numerous previous warnings. However, the problem is that, using your analogy, the manufacturers of the furnace have been repeatedly announcing for a decade that this will be the furnace that you can safely enter while it's burning and the furnace will protect you, and you don't even have to push a button. Admittedly, that decade included the point where they said that was coming soon, but for several years now they've had what they call "full self-driving" which isn't any self-driving and certainly not full. They really should be banned from using that term. If you lie to idiots that something is possible, consistently for a long time, and it never has been possible, some idiots will believe you. At what point does that become your fault, even if the idiots remain responsible for their own actions?

Science confirms what we all suspected: Four-day weeks rule

Re: Translation

It is, and I expect that many of them will find that it doesn't work and they've destroyed something that used to function quite well. In my opinion, I wouldn't mind if they could actually accomplish it; if they can make a computer do the job with equal or better quality, I don't mind that they do and someone loses the manual job that's now been automated. My primary concern is that they will pretend that's what they've accomplished when they actually have a computer that does the job much worse, then use unethical tactics to make the customers or other employees compensate for that externality.

I'm not very sure how this is related to the four-day week thing, though. I'm referring to productivity because, if we can prove that productivity is unchanged when shrinking the week, then it will be much easier to make it happen, whereas if we can't or say we can without actually proving it, it will be harder to convince those who can make it happen to do so. The people paying others to work don't want to pay more to get less, and they will try not to do that, but if we can demonstrate that they're not getting less, it might actually work out. That requires to actually know that would happen, and the people employing others are going to be harder to convince than I am.

Re: Translation

I argue about productivity because some people stated that this study found that the benefits of a shorter work week were present with equal productivity. The study did not find that. The study did not determine what happened with productivity. The people stating that it did were misinformed and their statements would mislead others.

Productivity is relevant, and it being difficult to measure will not make it irrelevant. Convincing people who can make this happen will require more information about this number, making this study incomplete. Sometimes, needed information is annoying to obtain, but we cannot ignore it because it's inconvenient. We can't decide that the answer is what we would like because it's inconvenient. There are only two things we can do:

1. Act scientifically, try to find a way to measure it, study that thing, and expose the methodology to challenge by other researchers.

2. Decide not to study it and just demand a change without research.

The approaches in the middle, only researching the parts that look good while skipping the rest, is pointless and counterproductive.

Re: stuff that involves lots of other people being at work

Nobody is saying that we should or even can deny people the choice to take worse positions. Shops that cut night shifts did not do it because we were trying to stop them. Mostly, they canceled them because those are the worst shifts, they would have to pay a lot to get people to agree to work them, and they didn't want to pay a lot. Either they didn't want to pay any shift more than another, and they would have to, or they were willing to pay a shift more, but doing so meant it wouldn't be profitable enough for their liking. That was considered easier than cutting a different shift because there are fewer people who want to take it or because there are fewer customers making it worthwhile.

I'm not sure what the alternative is here, since presumably you're not suggesting we mandate that businesses fill shifts at all hours. The businesses and workers will continue to follow the very typical pattern of deciding what shifts to offer and take based on costs and benefits.

Microsoft researchers: To fend off AI, consider a job as a pile driver

Re: Klarna Rehiring

That's Klarna's problem, but now they know what not to do next time. I think it will take more cautionary tales like that to make this sink in, but this is why I'm not a strong believer in the LLMs wiping out jobs thing. Some people will lose their jobs because their employer is an idiot and thought AI could do anything, but unfortunately, people already lose their jobs because their employer is an idiot and did something unwise from time to time. That is not the same as what AI promoters like to predict.

Re: Ask the thousands of laid off Microsoft employees

Let's consider a few of those. I already spoke about Microsoft, but you were talking about Klarna's happy announcements of cutting workers. To put it in context, they have been trying to replace customer service with AI for about three years now. They cut quite a few jobs last year, but not all of those were people getting replaced by AI, but shuttering partnerships and no longer needing the people who work on those. But maybe the most important context is the headlines we get when searching for recent information about this including:

Klarna's AI replaced 700 workers. It's trying to bring them back

Klarna Slows AI-Driven Job Cuts With Call for Real People

As it turns out, flashy tech doesn't always work. This reinforces my point in my last comment. Companies like to say they're using AI to great effect especially, but not exclusively, if they also make it. Some of them are lying. Some of them are telling the truth, but that doesn't mean that what they want to work and are saying works actually does.

And in my personal opinion, if they do find a technology that can actually replace a worker at the same quality, then I don't mind them using it. We've automated jobs before, and it will be better to help those who have to switch to something else, possibly something very similar but just not identical, rather than trying to prevent it. My biggest concern with job replacement by AI is what happens to people when a company insists on replacing someone with software that is incapable of doing the job, then tries to blame the customer for the unacceptable results so generated. As Klarna has realized, sometimes, the human does the work better.

Re: Ask the thousands of laid off Microsoft employees

But there is some reason to think that the cause isn't AI replacement, and therefore whether AI replacement is the biggest problem in the future is independent. There are a variety of causes, but the closest that AI gets to it is that the companies think they have to spend large sums on servers to train or run LLMs on and, to make up the rest so they don't show declining profits, they are cutting costs. And even that's probably not the biggest reason.

For example, Microsoft cut many jobs this year, and one of the hardest-hit areas was gaming. That might have something to do with the fact that, a couple years previously, they bought a really big gaming company, meaning they now have tons of people working on gaming. It would be quite logical to conclude that they had overinvested in that and don't need so much of that. The same kind of thing explained previous waves, because many tech companies hired like crazy during 2020-2022 then realized that they didn't need to do that and interest rates wouldn't always be basically zero.

None of this is good, but it also means that, to prove whether AI is going to replace jobs, we can't look at much of the past. The little information we have suggests that it's not making as big a dent as promoters like to imply. Promoters like that because if the LLM can do everyone's jobs, then:

A. That means their product is great and everyone should be buying it, which would earn them money.

B. Their product is a massive societal boost, so nothing should be put in their way while they build it more quickly than someone else, for example they should be allowed to break copyright law but nobody else should.

C. Their product is really scary, meaning governments should listen to them about how to regulate it, meaning that everyone except them should be prevented from doing things.

So they'll keep saying this, but that is not enough to prove it's happening. Some companies have indeed decided to fire some people and replace them with an LLM. The future will show us how often that works, and my prediction is that it won't be pleasant.

Australia bans kids from signing up for YouTube accounts, angering Google

Re: “We want kids to know who they are before platforms assume who they are.”

That would depend where they are, because in more urban areas, there are places where you can find a phone for emergency use. The fact that most phones don't charge by the minute no matter where you call (sometimes in country but that's enough) makes this easier, because now they don't even ask for the change. But that's certainly not everywhere, so what do we do in that case?

Here are some things we can do. I already mentioned the friend who got a cheap flip phone with no social media capability. Not that the flip phones that theoretically have social media are that dangerous, because they tend to have 2.8-inch screens and 2 MP cameras and those aren't going to make the process very easy. Or you can get any number of Android devices and lock access to things except the dialer and messages. That's a native feature now, was a native feature before in several manufacturers' variants, and before it was, there were many apps in the Play Store allowing you to do that (don't ask me how those got in while Nextcloud had to fight for normal files access, but it happened). If you're annoyed that Google put Google account requirements on their parental controls, I get it, that annoys me too, but you shouldn't be that surprised because it's Android and you're posting here, so you should already know how many features of Android are Google proprietary additions with data collection. Maybe you'd be happier with IOS, where the restrictions are a native OS component and require nothing more from the parent than setting up the device before handing it to the child, even if they never log into any Apple account.

From the cheapest 4G feature phone (2G in countries that haven't shut it down yet) to smartphones being handed down, there are tools available. Most of the time, the people complaining about these have put no effort into using them but demand better and better ones which would impose more onerous requirements on the rest of us. I have lost most of my sympathy for them.

Re: “We want kids to know who they are before platforms assume who they are.”

Several of those things are actually yes. For example, a parent who has used Google's parental controls can block apps, so if they don't want their children to sign into one, they can prevent that app being installed. Let's consider how social media in particular integrates with this. I don't know for sure, but I will assume you are correct that social media does not query for whether this is in child mode and take action. However, they do have child accounts, so if a parent decides that some social media is acceptable, they can only allow that app to be installed while they put it into child mode, which would involve a separate parent login. And yes, managing a device with several approved services means juggling multiple management accounts. The alternative which is perfectly acceptable is not managing that by not letting the child use that service. And Google's Family Link is not the only way to control what an Android device can do. Lots of systems exist for locking down such things. I've deployed a couple MDM systems, including a self-hosted one, and I don't even work in that area.

An alternative we can extend to whatever level we want. There's no reason why a child needs a Facebook account, a smartphone, or a phone, unless the parent agrees. I've known at least one parent who equipped their child with a flip phone because they wanted the child to be able to contact them but not to be able to use social media. That's a choice you can make, and there are plenty of flip phones on the market once again. It is not our responsibility to lock down everything we have so you don't have to work as hard to lock down the one you want locked down.

Nor is technology your primary problem. Imagine that I can give you the phone of your parenting dreams where you can do literally anything you want to prevent your child from doing something you don't like. Want a full record of every pixel they ever saw? Right over here. Want to play a recording of you yelling at them if they try entering the word TikTok into a text box? Two taps away. Your problems are far from over because you are a parent and your child can do all sorts of dangerous things from everything they encounter outside that phone. Your job is to prevent them from doing too many of those, and just because one device has bent to your whim doesn't mean the rest of the world will. Technology shouldn't be limited to make that easy because this problem is never easy and locking down the tech will not make it better but will instill a false sense of security from parents who aren't thinking it through.

Re: “We want kids to know who they are before platforms assume who they are.”

A lot of devices have parental controls. The first step is to try turning them on. It is not our fault that some parents choose not to think about this and give their children a device without any restrictions, allowing them to use a fake birth date to unlock the few remaining limitations that check the reported date on signup. No, those restrictions don't have a simple off switch, as they usually manage it by creating a parent's password which must be entered to disable them. Depending on what restrictions were applied, there may be restrictions we'd want that they don't have. Since most who complain seem unaware that these things exist, I tend not to be very sympathetic to calls for more restrictions on everyone to help with those too lazy to configure the hardware already in their hands.

Account access is a little different and more justifiable in my opinion. I don't really mind making a restriction against children creating accounts on various services, but I do object to age and therefore identity verification to enforce it, because it always seems that the identity verification was the point they really wanted to get to.

Oracle VirtualBox licensing tweak lies in wait for the unwary

Re: All because

If you want to use the book analogy, if I walk into a book shop and open a book, but then put it back on the shelf, I'm not charged for having done so. I am charged for that book when I want to take it with me out of the shop. That is how the shop determines the difference between people who intended to purchase the product and those who haven't made that choice yet. If they want, they can make sure I'm not reading that book inside the shop.

And so far, we aren't really arguing whether the thing Oracle's attempting is legal. So far, we're mostly arguing whether it is ethical, which I think it's not. There are reasons to think it's not legal either. It would be if people installed the software and used it for commercial purposes, but clicking a download link which isn't gated is unlikely to be considered entering into a contract. But most of what we dislike is definitely legal; it's totally legal for Oracle to call people who once downloaded a file, accuse them of having broken a contract, and demand lots of money, just as it is legal for me to call you out of the blue, accuse you of having broken a contract that never existed, and demand lots of money. It would only become illegal if I actually took that fallacious case to court and lied to pretend a contract existed when it didn't, which Oracle might not do because the chances are too high that they'd lose, but they can threaten it because that can scare people into paying up. However, if I did call you with such a threat, it would quite clearly make me a bad person, and some of us think that Oracle's being similarly immoral with this little change.

Re: All because

So, just to check, am I presented with that before I can download the software? Is it perhaps buried in a long document telling me that I can't reverse-engineer it, by downloading it I don't own it? Or is it just on another page?

On legitimate software, terms like that are presented before you can use it. For example, you're asked to connect it to the license you bought, which indicates that you need one. It doesn't sneak up on you by masquerading as a free trial which then springs a payment requirement on you later and uses "well you didn't read the FAQ page" when you complain. I don't care how much they charge for this or that they don't offer a trial, but I will object if they use tactics to collect on that which appear tailor-made for getting people to be unaware of it until it was harder to contest it. Legitimate businesses tell me how much it costs to do something before I do it and make it clear what limits apply on free trials they make.

US science left out in the cold amid plans to retire Antarctic icebreaker

Re: 33 year old vessel

Several of those questions are irrelevant such as:

"How much more service life do they anticipate out of this vessel?": Is this why they canceled their lease? If the vessel failed, wouldn't their lease cover this?

"What are they still researching after 33 years, as the article doesn't cite any actual accomplishments?": Oh, it was just one study. We only need to study one thing about Antarctica, sea ice, the climate in general, any of that stuff. I'm sure whatever it is was solved already. While they don't tag their discoveries by whether they occurred on this ship or not, a list of recent articles about their work include several about things determined by researchers in the Southern Ocean, including oceanographic, atmospheric, biological, and geological findings.

Your points might be relevant if they had canceled this ship because they no longer needed it. The fact that they still have a planned replacement suggests that they do need it. Okay, but maybe this ship was just too old, although thirty years isn't that unusual, but who knows, maybe it has some faults. If that was the case, they would have said that. The reason for this ship being cut is obvious: they're being asked to work with a lot less funding and this is an expensive thing they can no longer afford. That makes most of your questions wholly irrelevant, since the question should be whether the benefit they provide is worth continued funding at their previous level rather than a significant cut, and specifics about the ship could at most move the item around on the list of things they didn't want to cut but still have to.

How to find forgotten Wi-Fi passwords and SSIDs in Windows and Android

Re: Oh, the irony...

"Another alternative is the Unix style /etc/passwd encryption."

What are you talking about? /etc/passwd does not encrypt passwords. It hashes them. If you only have the hash of the WiFi password, it won't work. And encrypting passwords is what's done, with the problem that, if you want it to connect automatically, the decryption key for those passwords has to be stored somewhere anyway, and that means something with access to both data sources can access the passwords.

Re: Oh, the irony...

You do know that it must present the key in order to connect, so it has to have the ability to retrieve it in clear text, even though it's stored encrypted. And if it can, it can show it to you. What do you think happens when you connect to WiFi on Linux? Depending on how you do it, your general options are:

1. Enter the key in plain text into a configuration file, where it stays in plain text as long as you can connect.

2. Enter the key into a database, which is easily decrypted by you any time you feel like it and contains the key in plain text.

Or literally any other system that connects to WiFi. All of them can expose the key, and the only question is whether they expose it in the UI or make you try for a few minutes to get it. The alternative is not remembering it and asking you for it on each connection. That is how single-key WiFi security works.

Re: More Goddamned-Creepy User-Tracking

If you've enabled backups, then you should expect that deleting something from the main storage will not remove it from the backup because otherwise, the backup wouldn't serve its purpose as a backup. If you deleted a WiFi network you wanted to restore, getting it out of the backup is a perfectly normal thing to do.

What they were asking is whether it sneakily stays somewhere, and although that's possible, there's no evidence suggesting it does. It stays in that list because it's been told to connect again if it sees that network, so it needs to remember the network in order to do that.

Trump pushes EU into trade 'deal' that several EU leaders aren't happy about

Re: Someone from the EU please explain

The article covered this:

Trump and von der Leyen's agreement on the deal doesn't mean it's a sure thing, of course. As a matter that affects the entirety of the EU, each of the 27 member states will get to vote to ratify the trade agreement. That might be a hard sell.

No, she does not have unilateral authority, but she does have enough that she can try to negotiate something and put it before others for approval.

Problem PC had graybeards stumped until trainee rummaged through trash

Re: I'm a bit confused

Of course, some facts or the entire thing here might be made up, but the story isn't implausible. Some computer games loaded themselves at startup or kept a program running after they closed in order to make the process of starting to play them faster because a lot of the resources had been loaded. These assumed that the user would return to them quickly enough to justify doing that, but it could use up lots of RAM. Networked games might also host a server, which would use other resources depending on who might use the server. That would be really easy to change, but not everyone knew they should do that.

And to the first reply to your question, it's also not implausible that it wouldn't have been easy to track down. Does this pharmacy sound like they had a lot of admins on standby to fix this? A lot of small businesses I've seen have exactly zero people who know what netstat does, let alone how to use it properly. For that matter, I've known an IT company whose primary business was serving such organizations and not all of their tech staff knew what netstat does. Some did and would use it when necessary, but for people whose primary task was explaining how not to break things, it wasn't considered required knowledge. Even to get there, we'd have to jump from "this computer is slow" to "the problem is something running on the network", which is not the most likely cause. I wouldn't find it at all surprising if they did things like defragment the hard drive or scan for malware in the hope that that would be the cause, only to find that it was still slow afterward.

It is a lot of assumptions, but it followed the scientific method:

Hypothesis: This might mean that people have been gaming on this computer.

Experiment: Is there a game on this computer?

Result: Yes, there is.

It would have worked the same way if the hypothesis was proven false. Only if the hypothesis was considered true without an attempt to prove it would that have been a problem.

Re: Pharmacists

I don't get your logic, at least this part:

Basis: The case in this article? A game small enough to be hidden slowed the computer down enough to be noticeable?

Conclusion: That computer was ANCIENT JUNK, and should have been replaced years earlier.

How much resource a game uses is not correlated with how easy it is to hide. By your statement, my best guess is that you decided it must have been small enough as in disk usage to hide, but disk usage has no effect on RAM or CPU usage once it's executed, which is the more likely cause behind slow performance, and could have been solved by simply showing people how to terminate it when they weren't playing it. That is if they even used disk usage to try to find it, which actually wouldn't be that easy. By the late 2000s, relatively large disks were available and somewhat common. Windows updates could use plenty of that, so a game could easily be a gigabyte or two and not showed up in a cursory search, and it probably could have been twenty gigabytes and not caused a disk full problem because this computer probably didn't generate lots of data in its normal job. If the disk usage guess is wrong, what made you conclude that being able to hide a program means it won't tax a computer?

Your suggestion about the reaction isn't clearly correct either. There are reasons not to have gaming computers at workplaces. Some jobs don't apply and have nonproductive downtime when playing a game is justifiable, although in those cases it's more common for employees to be allowed to bring in their own gaming hardware rather than using the employer's hardware for that. Other jobs don't have that time and have a normal reason to not approve of employees playing games instead of doing what they are employed to do. I'm not sure which one this is, but the reaction suggests that maybe it was the latter.

Re: Late 2000s ?

That doesn't happen now. Computers are expensive. Unless I can be convinced that the computer's being troublesome because of a thing I can't fix, it's not time to junk it. If it's still happening when booted to a known clean USB disk, then I'll start to consider that there may be some bigger problem here, and even then I'll open it up and look for an overheating problem first, but this machine would have passed that test quite quickly, indicating that's likely a software problem which I can probably do something about.

Please, FOSS world, we need something like ChromeOS

Re: How difficult is it? Actually : not at all easy

I think that is what Liam should have been driving at, but I think he took a wrong turning. The problem is, as David correctly stated, getting the thing installed. To some extent, it's also getting Linux to do the things they used to do on Windows in such a way that they feel confident about achieving the same goals. Neither of those will be fixed by making Linux more like Chrome OS. The installation problem won't be fixed in the slightest, and it's a pure surrender on the functionality angle because it degrades the machine to the least common feature set of anything else, rather than normal desktop Linux which can do all of that plus lots of functional desktop applications even if we ignore virtualization and emulation.

As AI becomes more popular, concerns grow over its effect on mental health

Re: The more things change...

And, even without that, the results of the attempt were impossible to reproduce and calculated as being implausible at best. But showing the actual results of such a diet, which would be rather unhealthy but not immediately lethal, would not have made for as entertaining a documentary.

Granted, and that just makes it more popular in a different sense: the number of people willing to add it to places whether or not it is helping. Combined with the many people who voluntarily choose to use it, I'm forced to conclude that it is, in fact, getting more popular. That doesn't mean it will continue to increase in popularity, but I think it is fair to say we're on the upswing.