Posts by doublelayer

With 10 months of support remaining, Windows 10 still dominates

Re: Enshittification...

I don't hate the term, nor have I expressed criticism of it before. However, I do find that it is being used more generally than it was originally defined. It is intended to refer to specific, long-lasting policy decisions, so when people use it for "I think it got worse", it tends to lose some meaning.

If it got worse because it is a way to extract more money from you or someone else (advertisers, mostly) because that is the way the people running the business have decided to change the business model, that is what "enshittification" was supposed to mean. If it got worse because they have a new UI designer who is not good at their job, that is something else. Even if that UI designer hired more of them and they're all messing up what once made total sense, and they had some programmer friends who don't understand why you don't catch and ignore exceptions unless you have a specific reason, that's not what the term referred to.

Screwed by the cloud: Hardware vendors looking for that raison d'refresh

I don't think they're suggesting that the newer machine is more secure by design, but that if you're running eight servers, then theoretically that's eight opportunities to mess up the security of one of them, so you should definitely be trying to have as few servers as possible, and wouldn't you know we happen to have one server that can be like a bunch of older servers in one box. Of course, if you are running a lot of different servers, then you're very likely deploying the same image to most or all of them, meaning you have exactly as many opportunities to do it wrong. At that level, you also probably deploy a bunch of VMs on that server and weird manual configuration is going to be on specific VMs, meaning the number and size of physical servers is irrelevant to security at that level. Customizing each server manually is more likely to happen if you only have a few of them anyway, in which case buying an expensive newer model may be overkill because you aren't using as much performance as you just bought.

Will 2025 be the year satellite-to-smartphone services truly take off?

Re: Battery

I think this will be a nonissue for long enough that it eventually becomes a big problem. Right now, when the only thing you can do is send a message, people won't use it very often. They will use it to either send information to emergency services, where they are really not concerned about battery life, or something important enough to fiddle with things to get it out, where they are similarly not too worried. That means nobody will complain about the battery life problem for a while until the day, assuming it comes sometime, when they are able to use data connections from these satellites. When people start to browse a website, get online navigation instructions, or watch a video over these services, I expect battery life is going to start being one of their largest concerns, but it will be a while before anyone lets someone do that. There's a chance that technical limitations will prevent that working well at all, and there's a chance that they choose to charge so much for bandwidth that few ever try it, so we might not even get there.

Boffins ponder paltry brain data rate of 10 bits per second

Re: The amount of nonsense in this paper is remarkable

I think you're mostly correct, but a few points:

"Someone will have to tell these people that keyboards have a certain lay out for a reason and that subsequently it's not physically possible to type random sequences as fast as sequences in the targeted language of the layout."

I don't think this is why that difference exists. Most keyboard layouts have had a little thought put into them, but yet Dvorak, for all its adherents, doesn't actually speed typing up compared to the less thought-out QWERTY where avoiding lever jams was at least part of the consideration. I think the main differences is that I have a lot of muscle memory. When I want to type "the", I can call on a long history of typing those letters in that order without needing any conscious thought to do it correctly every time. If I have to type "ymx", I don't have that memory. Also, it is common to look at the screen while typing if only by reflex to verify that you have typed things correctly. That's easy when you're copying words, because you can remember the words for a short time. Let's see, I was typing about types of apples, the words say "green-skinned apples often used in", sounds good. If I have to copy random letters from something I'm reading, looking back and forth from one window to another is challenging because the random letters have not been memorized and I'm either comparing them manually, which is slow, or flicking back and forth for by reflex without being able to use that reflex properly, which is going to cause delays as the brain attempts to do something that it cannot do. If we created a keyboard layout by dumping all the letters into the space for letter keys at random, but we then had people only use that layout for years, I think they would be similarly fast while typing languages they know and similarly slower when typing random letters.

"The estimate for the human brain is that it uses 20W a day or on average less than 1W an hour."

Watts are already energy over time; you can't divide them like that. The brain tends to consume about 1.7 megajoules of energy in a day, meaning its average power consumption is 20 W whether it's per day or per second. Which means that you can in fact get a computer that operates with less power, and quite easily. It will be a lot more power-efficient than the brain when solving mathematical problems, but most of the tasks we want to perform daily can't be done at all or anywhere nearly as efficiently on such a machine.

Can 4G feature phones rise again on the back of QVGA, thin clients, and remote browsers?

Re: DumbOS?

That approach is not very likely because sending pixel changes, even compressed, takes a lot more bandwidth than sending text and letting the local application turn that into pixels. This is being sold to countries where people can't buy low-end smartphones, even though you can get some very cheap ones these days. I don't think the users who are buying them have unlimited data plans. In many ways, the best thing this cloud browsing thing might do for the users is find ways of radically compressing everything they send so that users can browse without running out of their allotment in a couple hours. Downscaling or eliminating images could save plenty, and elimination is the only one you can do from the phone itself.

The pixel approach is also tricky in many ways. For example, if there is an email address in a website and I activate it, the browser can open my mail client and enter the address because that's already present in text form. For one that's only operating on images, there is no access to the text, so it would have to do a round trip just to confirm that what I clicked on was clickable and you'd have to implement protocols for giving it the necessary information to act on the thing it clicked on.

I'm not sure what he thinks this needs. Many sites someone is going to try to load are going to use inefficient scripts, which I gather is what the cloud servers are for: they execute that and only send the document model to the phone. However, that introduces an extra point of latency, especially if that script is frequently changing the model. I assume this isn't sending a picture of the browser because that would increase bandwidth needs significantly. Operating those servers is going to add cost as well, and I'm not sure where the money to cover that is coming from.

In comparison, nearly all these devices already have a browser on them. Usually, it's a Gecko-based KaiOS browser or a Chromium version on something running a stripped Android build. Those browsers aren't great, and the small amounts of RAM available to the browser and the low-end processor aren't helping, but it can browse any site you want with no server in the middle. Somehow, his service has to compete with that, and I'm not sure how easy it is to sell a faster browsing experience.

Trump's tariff threats could bump PC prices by almost half

Re: But surely

That individual would be required to pay them. That might be precalculated by the vendor and added to your cost. It might be precalculated by the platform. It might be added by customs and charged to you at entry. Or they all might forget to do it and do nothing. You get to figure out which one is chosen each time you buy something.

One third of adults can't delete device data

"I'm not saying laws are/were a problem, but forcing every manufacturer into using just one connector makes life really easy for the Filth and criminals."

I'm not sure how you intend those two parts not to be contradictory or how you expect the second part not to be wrong. The box full of adapters was evidently available, so connector multitude wasn't doing it. Of course what was helping at the time is that you probably needed lots of different software to sync call history and contacts from each model. Nowadays, when there are many fewer operating systems in use, you just need exploits for those that are. The number of cables is not the problem. Nor is device covertness, because there's a really simple solution to that. Not a vape, but a small phone charger battery. It's got a USB port and plenty of spacing. Just connect a USB-C or Lightning cable to it and press the button and it can use the exploit. There's your covertness sorted. The problem has nothing to do with cables or laws about cables.

Right, the laws are the problem. It's not like, without those laws, there were a total of two connections, so all they needed was a Lightning-to-USB adapter to stick on the end of their cable. It's not like you could use a Bluetooth keyboard to do that if you had an unlocked device, which would also be compatible with everything. A standard of USB-C is the villain in this case. Physical possession of a device with the locks undone gives you a lot of power, although, like computers, there are things you can't do without authenticating again and you can add more things to that. Both Android and IOS support locking apps behind a passcode, either the normal phone one or a custom one.

Re: Surely

The biggest risk is that people might pick up some unerased devices in the hopes that some of them have a pin of 123456 or unlock without a pin at all. Then they have a low-cost way to steal things. It's not hard to press the erase button, and for most people, there is little reason for concern that there will be recoverable data left after doing that. The flash may not be entirely erased, but most phones encrypt by default and the part of the flash where the key is has been erased. For most opportunistic attackers, getting low-level access to read the flash is more work than they're going to do either.

There are exceptions for phones or drives that don't work where physical destruction is advisable. Otherwise, most people have no need to do that.

Christmas 1984: The last hurrah for 8-bit home computers

Re: Glory Days

"apart from ARM (which is technically Japanese-owned and is listed on the Nasdaq) no major technology has ever been able to establish itself in the US, e.g. Nokia, Vodafone etc."

If you're talking about the average consumer, Nokia was relatively successful in the US. In 2005, it was the third highest seller of mobile phones in that market. It didn't continue to succeed there for the same reason that it didn't continue its even more dominant position in the European market: it didn't match its earlier smartphones to the feature set of competing smartphones for a few years, then Microsoft bought it and prevented it from doing so for a few more years, and by the time it tried to fix things, it was too late. Curiously, it seems Nokia (the brand, actually HMD)was fifth by market share in the US for mobile phones in 2022. I'm not sure if they stayed there, but that's surprisingly high for a company that makes relatively unimpressive hardware compared to their earlier status.

Re: 1984: The last hurrah for 8-bit home computers

"the NES lasted until 1995 outside Japan and 2003 in Japan."

Wasn't this the point, that this didn't really count because it was only a game machine? The processor inside it doesn't matter to the article's point if it's a single-purpose device that only ran games and you couldn't write other things for it. So even if you could continue to buy them, did any significant 8-bit home computers get designed after 1984?

Bonus discussion topic: does this matter? Why is a 16-bit home computer different or worse than a 8-bit one?

OpenAI plans to ring in the New Year with a for-profit push

Re: Don't worry about it dear investors...

I think any investors they have will be fine with this*. It lets them get lots more money, and they need lots more money so that Sam Altman can get lots of it, sorry I mean they can quickly build the most amazing models possible before someone beats them to it. That's necessary if OpenAI is ever going to be profitable, and if you thought they weren't going to, you wouldn't be investing. At this stage, investors are still banking heavily on outspending and outlasting their competitors, so the need for more infusions is clear. Any investor who thinks they've stagnated is not too interested in still being an investor either, and the more for-profit they are, the easier it would be to find someone dumber to take the stake off you.

* All investors except, of course, for Elon Musk if you can call him one now, but that's not really about a for-profit move even though it's probably the lawsuit he's got the best argument for.

Just how deep is Nvidia's CUDA moat really?

Re: "but you also entirely eliminate any ability for a manufacturer to improve anything."

I don't understand how that's supposed to happen. When you say "If they added a feature then the patch would recognise it", what is "the patch"? They can change their implementation in microcode or something like it, thereby improving their implementation of the allowed instructions. There is only so much you can do to improve that. They wouldn't be allowed to have new instructions because those new instructions aren't part of the standard, so if someone compiled for those new instructions, it would only work on that manufacturer's parts. What improvements, other than microcode, do you think they'd be allowed to do with a mandated standard? While we'd see some improvements with microcode, ruthless focus on microcode alone is how we got several serious CPU security bugs, because Intel and AMD wanted more performance out of the same ISA every year and kept looking for more and more hacks to get it until one of those hacks had some nasty side-effects. Even then, they were still adding some things to the ISA which sped up some classes of program, which wouldn't be allowed if they have to stick to a standard.

Nvidia's moat is not a good thing, but attempts to eliminate it by requiring Nvidia and AMD to build the same chip isn't helping. There are a few negative effects. It seems we're disagreeing about the ability to improve on a mandatory standard, which I assume is due to a miscommunication somewhere. We also have the problem that, if any improvement one makes is immediately available to its competitor, there is less benefit to the manufacturer for improving, so why bother to do so? I think the best way to fight against Nvidia's moat is what AMD is currently doing: making it easier to port things from CUDA to run on theirs as well. They're lagging, not because Nvidia did anything wrong, but because Nvidia built something that AMD didn't bother with and people wanted that thing. AMD can catch up but they have to go to some effort to do it. I don't think we are helping the users by trying to fill in that moat for AMD. There are some times where that kind of regulation is justified, but I don't think we're there and I think attempting it nonetheless will be unwelcome to users and less effective than you expect.

Re: If OpenCL is *the* cross platform standard why is it not used?

OpenCL had various restrictions, with each of them having some downward effect on its usage. However, it didn't end up becoming the most popular neither because someone deliberately killed it nor because it had gaping technical flaws. We can debate how large each of the effects was, but I think we also need to question why we assume it was likely to be the victor.

By being a standard, OpenCL tended to run on more things, but not as efficiently on each one. Optimizing for a specific part meant you could run faster, and for something as compute-intensive as training ML models, that was a key benefit. Any other standard is likely to have the same downside. We can try to make a standard that's close to the hardware which would reduce that difference, but it's always likely to exist. That's not necessarily a problem; although compiled languages tended to run slower than hand-coded assembly*, people still chose them for the portability or writeability advantages.

Your alternative suggestion, a standard at the lowest level, is going to have other restrictions. At that point, you're no longer going to prefer one manufacturer over another, but you also entirely eliminate any ability for a manufacturer to improve anything. If they can't change the instructions their chips understand, they can't do something as simple as allowing you to work on a larger piece of data, thus accomplishing in one instruction what would have taken multiple. Other optimizations would be prohibited. Nvidia and AMD have spent lots of money building those improvements, testing them, making hardware that can do them, and building them into compilers. By requiring a standard like the one you've described, you're effectively telling every user that they will no longer get any of that work in exchange for being able to buy AMD parts. It would be similar to mandating that every CPU manufactured must be an AMD64 one, with alternative approaches like ARM and RISC-V disallowed, but anyone who wanted could try to build an AMD64 one. Do you think users are going to be pleased with that?

* Nowadays, compilers are so good at optimizing that they often produce better code than someone working in assembly. This doesn't change the point. People can still improve on the compiler's output, and in some cases they can do it quite easily because they understand the semantics of their program's contents whereas the compiler's optimizations are general ones. It is still possible to exceed a compiler's efficiency. You just have to really want to do so and be willing to put in the substantial effort.

Fining Big Tech isn't working. Make them give away illegally trained LLMs as public domain

Re: Small thinking

I'm not trying to trip it up, nor am I hung up on AGI. My problem is that, if the result of the process is something I can't check, which your example is, then I have come to expect that there may be many errors. Your clarification makes me more concerned about this. You fully expect that GPT will make these mistakes unless I prompt it with something I'm supposed to get at random online, because somehow it can't do that itself. I'm guessing this works for you because, if you're building dozens of these, you probably have all the knowledge you need to look over and confirm these designs before you pay someone to make it. For someone who doesn't, it seems like a recipe for wasting money on flawed designs, of which disconnected data lines is an example, not something that has actually happened because I'm not actually building designs.

My basis for this has been actual testing of the thing when I had software tasks to do. If someone who didn't know how to write software used GPT to write some software for them, they would not get something usable. I conclude that from several attempts to get it to write software, which I know enough about to judge its output. Its output on small tasks often contains simple yet unpredictable errors, and that's when I split it into little pieces for it which a nontechnical person will not do. I cannot give GPT to someone unfamiliar with software and expect them to get good results. I have as much familiarity with PCB design as they have with writing software, so I cannot expect good results from that. It may be that simple PCB design is simpler, with more boilerplate or fewer options for messing up, than the simple software which GPT reliably fails to write properly. In that case, maybe it is actually more capable for you than it is for me.

Re: Small thinking

This sounds great. I certainly do not have the knowledge, and I've had many ideas for something where the hardware design should be pretty simple. If I could have someone make me the board designs where usually the most complicated thing is getting buttons in the right place, that would be quite helpful.

Sadly, I have seen GPT's software output and I do know how to write software. Therefore, if I ask it to produce a PCB output which I don't know enough to check, I expect that it will come back from the contract manufacturer perfect, at least the exact shape that I expected with the USB port fully connected to the processor and RAM, except the data pins won't be connected so I can't flash software to it and a couple of the LEDs installed without the necessary components. I can get working code out of GPT. If it's a really simple problem, the code often comes out correct the first time. If it's anything else, I have to do it again and again, checking manually each time, so much that it would have been faster to do it myself, and the only reason it's correct at the end is that I have the skills to check its output manually. That's why I don't trust it for anything I can't do myself and why I don't use it for things I can.

Re: Imaginary Property does not exist

Free means several things:

1. Liberty (see freedom).

2. Lacking something (a field free of trees).

3. Not costing any money.

4. Not limited physically (let it fall free).

5. Generous or frequent.

6. Probably other things, plus all the things it means that aren't adjectives.

Stop claiming it means one thing. We all know what "free software" means. When we talk about it, we will use both words, and many of us will say "free/libre software" just in case it gets mistaken for software that is free(3). If someone says something like the thing you quoted, they and you both clearly understand that they were using the completely valid and more widely understood definition number 3. If they meant the free you're trying to claim as the only option, they would have said "released as free software". Your grammatical argument, that free(3) must always expand to "free of charge" is wrong, and it fits just fine in the way they'd express any other price ("for £20").

"There are no rights - only restrictions."

Wrong again. There are rights. If you have copyright over some code, you have a right to copyright, defended by your nation's copyright laws. "Right" has another similar meaning, the ability to do something, granted by a contract. Thus, if I sign a contract stating that I will allow you to come live in my house, you can call it perfectly grammatically, and the contract may also call it, the right to live in my house. You can negotiate for permission to do something, and while it has a slightly different meaning than the governmental right, that's a valid word for it.

Grammatical defenses of things tend to be unconvincing, even to someone who mostly probably agrees with you.

Re: Potentially another option - poison the well?

"I only contribute to projects that proudly brand themselves as free software, as I know that the project will always be working for freedom."

Ah, one of those who think that "free software" and "open source" aren't often used as synonymous terms by people who understand that they are slightly different, but not so different that ideological battles are necessary. You're not wrong that some things are described as open source when they're not, but two things apply to those: A) all the things you list are contrary to definitions of open source, including the one you linked to and B) something calling itself "free software" has done all of the same things.

You go on to decide that, somehow, GitHub is a way to tell these from another. I don't know why, you just do. Of course, everyone who has spent a little while actually working with this stuff understands that, to figure out how well it fits with the ideals of free software or open source, you actually have to check the details because what website they host their repos on tells you nothing. A good start is often to open the license file and see what it says. Is it a standard license you've already read? Is it one of those with an additional clause attached? Is it one of their own design? Does the one of their own design veer into restrictions on being allowed to use, modify, or distribute? Does it have a section that could best be summarized starting with "except for someone"? That doesn't answer all of it, but it tells you a lot of useful things, often good enough to toss out many of the fakers.

Re: Imaginary Property does not exist

I agree. It is definitely the wrong punishment. I merely wanted to indicate that it's incorrect to say that "ending copyright restrictions by releasing a work into the public domain without other restrictions is in no way a punishment, no matter the situation". It is, but it's not strong enough and it causes unnecessary and preventable harms that we shouldn't allow.

If your concern is privacy rather than copyright, your solution is somehow even worse. It would be much easier to get at any of your private data those models hold if those models were open source. I could start chopping out any protections that would prevent it from happily spitting out the data. I can run queries much faster until patterns start to show up. Asking for something that contains your private data to be made public because you're unhappy that it contains your private data is backwards.

It's the same as if I said that Google should be penalized for their data extraction by taking their records on everyone they can find and making those public. It's a real penalty, as that's the data they use to claim to advertisers that they can target ads. "See, we have every site this person ever visited and everything they ever searched for, so surely we know what they will be willing to pay for." They won't get that revenue without having exclusive access to that. However, my problem is that Google has it, and that is not solved by making sure that everyone else, from Facebook to governments to criminals has it too. It is a purely negative change.

I understand your reasons for suggesting it, but those are flawed as well. You see something that was costly to create, and you don't like getting rid of it. I can be like that as well. If something is working, even if I don't need it, I don't want to toss it into the trash. I try to find someone else who will use it, even if most of them tell me that they've got something better, because it's not dead yet. That doesn't work when the item concerned has a flaw. If, for example, I had a Samsung Galaxy Note 7 with one of the self-igniting batteries, I might be unhappy that I now have to dispose of a device that was expensive to make and so far hasn't done anything wrong. It would be dangerous to myself and others to keep using it on that basis.

Re: Imaginary Property does not exist

It would be a punishment, just a much smaller one than they deserve and one that harms someone else at the same time. Companies with LLMs would lose a lot of money if the models they spent tons of money on were released for free, so they would really hate it. For the other reasons, that's not good enough. They should be required to destroy the models that contain the data, destroy their copies of the data, and they can go back and train on the remainder of their training data, already confirmed to be something they have rights to, using their latest software. That probably makes for a bad model, but that's their problem.

Re: King Cnut

We may not be able to prevent LLMs from being created. That doesn't prove that the tide is useful. It doesn't prove that the intentions were wrong. It also suggests that your idea of why the Luddites were Ludditing might be wrong, but you're far from the only one to make that mistake.

However, in this case, that's not my opinion. Make an LLM if you want. I don't find them very useful, and I'm more than happy to ban the use of them in some cases, but if you want to make one and use it to try to do your work, have at it. Just don't steal people's work to do it. You have to buy that when it's copyrighted. If you don't want to, you can use any stuff that is in the public domain and anything people agree to give you. I may not think your electricity usage is the best use of that resource, but you're paying for it, go ahead. I may not think you have the ability to make a good product, but it's your product, don't worry about me. Trying to protect people from abuses, specifically ones that have been obviously illegal for some time, is not trying to hold back a technology. You can use copyright infringement for a number of things, and some of those are things I find useful. That still doesn't justify letting you do it.

Re: Who wins?

From this and their other comments, not only do they want that, they want to strip people of even that amount almost immediately, and they want to give the AI credit for anything it outputs. Therefore, implementing all their solutions means people who make training data get nothing, except for a few people who published very recently who get next to nothing, and the people who make the AI models get tons just on volume. There's one thing they got right though, I really disagree with almost every part of their suggestions, excluding one tiny element where their nonspecific proposal is something I agree with in principle.

"Why is that M&E innovators get 20 years benefit, but some bloke who slings together a bit of software has his entire life plus 50 years?"

Because patent protection is a lot stronger than copyright protection. If I have a patent on something, you are not allowed to use my invention without my permission. For example, if I have patented a certain chemical, you can't make that chemical. It doesn't matter if you've made a different manufacturing system and have a different use case for it, and it doesn't matter if you've invented something you mix with it to do something else, the product isn't allowed unless you get my approval. In copyright protection, you just can't use the same form I did. If I write a piece of music, you can't write the same one, but you are free to write something similar. Patent protection is a lot more than copyright protection, and thus it has a shorter time period.

Re: Delete them

That's exactly what I want. They should remove any IP they don't have the right to. If it's insignificant, then that's no problem. They are free to build an LLM out of anything to which they have legal access, whether that's public domain stuff, stuff they paid for, or stuff that someone agreed to give them for free. I just don't want them to assume that, because they found it online, that means they get to use it for free. As I've said many times, if I find a copy of their model by getting one of their computers to hand it to me, do I get to do whatever I like with it, including making lots of money, without their permission? If not, and it is not, then they should not do that to others either.

Re: Throwing the baby out with the bathwater

My post made one major point, that there are models that are not LLMs. You seem to have missed that point. Voice recognition models are not LLMs. People who make the former out of licensed data do not have enough licensed training data to make the latter. Improvements since 2005 come from more computing power, more advanced techniques, and more time and money spent on improvements. They do not come from LLMs.

You're welcome to try making an LLM forecast the weather better than tailored models. When you do, feel free to submit it to rigorous examination. You won't have much success, because the tailored models are going to improve their performance faster than your LLM will.

I don't agree with either of your points, and I'm a bit surprised to see them together. You are not the first person who wants to significantly restrict or eliminate IP protections. While I think some parts of patent and copyright need to be changed, most proposed architectural changes are, in my mind, harmful. Maybe you have specifics that I would agree with, but in the general terms you've stated so far, I don't think I would agree.

The idea that any IP-based activity should submit funds somewhere is not a problem. It's called taxes. If, however, you mean what I think you mean, that they need a special tax for the fact that there wasn't something physical involved, I disapprove of the suggestion. Physical labor is not the property of others. Intellectual labor shouldn't be either. The work of others is not free to be consumed for any reason, and making the people consuming it pay into a public fund does not cancel out the harms they've done to everyone whose work they've used without permission.

Re: Throwing the baby out with the bathwater

You're conflating two different types of AI, but it's not likely to convince anyone. Newer speech recognition systems that use AI, in the sense of modern machine learning strategies, are not covered here. They are not covered because those weren't trained on stolen data. They were trained on recordings and transcripts that the creators have the rights to. At least, that is the case for those we know about. It's possible that some of those have been trained on recordings they didn't have the rights to, but you wouldn't notice it the way you do with LLMs. Likewise for weather forecasting models, pattern recognition models, and the like.

There's something all of those have in common: none of them use LLMs. I do not get a forecast by asking a chatbot for one. I do get one by using a specialized forecasting model. The argument in this article is about LLMs, the ones trained from stolen content. Thus, your most convincing arguments, the ones that bring accessibility or other societal benefits, aren't covered in the argument at all.

Apple called on to ditch AI headline summaries after BBC debacle

Re: It's garbage

There aren't secret or copyrighted laws in these cases. There can be exceptions, for example where a law mandates a standard and ISO won't give you the standard without payment, but most cases don't involve that kind of thing so we can ignore them for now. The problem is that, even when you scrape all the laws and feed them into an LLM, they can easily mistake things the way they mistake lots of other things. A law means you are allowed to conduct a certain action, and you are sued for conducting that action, sounds like a match. Except the LLM has not noticed that the law allows you to conduct that action if you are a law enforcement officer in active duty following a disruption to communication caused by a serious natural disaster or terrorist attack, but that only appeared once in the training data so the LLM didn't recognize that you're none of those things.

Best case: a lawyer, paralegal, or other legal person reads the produced document. They weren't aware of the law, so they look it up. In the summary, they realize it doesn't apply to you. They throw out the document and start again. Maybe the LLM will produce something correct the next time. Result: the time to generate the original document and the time to review it for errors is lost.

Average case: A lawyer hands the document to a paralegal and says "check this". The paralegal reads the document and finds the reference to a law. They spend a while reading the text of that law to confirm that, even though the summary seems to limit it, the LLM which is supposed to be the next great thing may have found a cool loophole which will get this client off. They spend several hours checking this to realize that it doesn't help. They report their problems to the lawyer. The lawyer sends the report to the prompt generator. The prompt generator makes a new document and the process repeats. Result: several hours added to your legal bill.

Worst case for now: The lawyer hands the document to a paralegal and says "check this". The paralegal sees that a law is mentioned and sees the quote that the action is allowed. They check that the law exists, and it does. They check whether the quote is in there, and it is. They send the document back approved. Result: "Guilty. We are also considering contempt of court charges for council for the defendant."

Re: Not an AI Problem

Bad headlines are nothing new, but these are probably still worse. Editors may pick headlines intended to mislead, that make bad summaries, or ones designed to make the article sound more interesting than it is, but they tend not to write headlines that diametrically oppose the article unless they didn't read it or confused one article with another. This bot did that all on its own with no pressure causing it to do so. That's likely to happen a lot more frequently than an editor making such a massive mistake.

I've always wondered why they don't have the article authors write the headlines. At least in non-clickbaity examples, that should get an accurate summary in there. I suppose we've now solved that problem. The article and headline will be written by the same thing: an AI bot that made both of them up.

Re: It's garbage

Theoretically, that would be a benefit. Theoretically, it should be easier to fix an error in something than to make it from scratch. In my experience, that's not how it works. Correcting a bad document into a good one often takes longer than writing a good document in the first place, not even counting any time spent to make the bad document. The theory is only correct when I'm correcting a pretty good document by fixing small grammatical or factual errors.

When an LLM is liable to making something up and resting large parts of the document on that flimsy foundation, correcting it usually means removing everything but a mostly usable introduction paragraph and trying to do it another time. That doesn't result in less effort expended, because you're multiplying the time it takes to check every fact alleged by the document by the number of times you generate something new, plus any time you need to make manual edits. The further problem is that, if someone decides to skip one of those stages, the document still looks like it is complete, but now it's the kind of thing that results in summary judgements against you from judges annoyed that you're making them fact check your submissions.

That's fair for Apple's interest in sending random news headlines to iPhones that never asked for it. Those were often irrelevant, but at least they were true. Users can turn those off when they realize how they don't need them.

Making up headlines and sending those is much worse, and unhappiness with Apple is completely justified, especially for the reporters given credit for the false statements. Consider your reaction if you were working in an office and the person next to you frequently shouted inaccurate statements at you about things you needed to check. Yes, you could and would eventually ignore them, but it wouldn't be your fault if you wasted your time checking out one of those inaccurate statements, the problems caused by any reaction would be their fault, and someone in management might well tell them to stop doing that. We can't make Apple stop doing that, but we have lots of reasons to want them to. Asking users to turn the thing off is not good enough.

Jury spares Qualcomm's AI PC ambitions, but Arm eyes a retrial

"He believes in one set of acceptable copyright licenses of works of opinion and another set of acceptable copyright licenses for functional works."

Which is fine if you're discussing what you'd like copyright law to be, but completely meaningless when discussing what copyright law actually is. He already mixed those a bit too much for my liking in that essay, but you went the whole hog and started citing that source like it makes points about actual law which, in reality, are his opinions.

Re: Appeal

Without access to the specifics, I also thought that ARM was more likely to have written this into their licenses. I don't like to bet, but if pressed, I would have guessed that ARM was more likely than Qualcomm to win here. That they haven't could mean that the jury simply got things very wrong, but it could also suggest that ARM did in fact mess up here. I have no reason to prove otherwise other than my instincts about how smart ARM's lawyers should be, and we do have to consider that these are either the same lawyers or close colleagues of the ones that persecuted someone who wrote a good textbook used to make things ARM needs on dubious trademark terms, so they aren't infallible. My instincts on their competence aren't strong enough for me to have confidence that the jury erred here. Maybe the trial on the remaining charge will give us more information.

Re: Appeal

With ARM's contracts, at least as ARM described them, they had to demand destruction of the parts because there was no other remedy allowed. The next step, presumably, was for Qualcomm to go to ARM and ask if ARM would retract that demand for destruction if Qualcomm paid them some money, which they could negotiate. Qualcomm instead chose to challenge the accuracy of ARM's claim, and they apparently got a jury to decide that ARM was wrong after all. I'm pretty sure that ARM didn't want those destroyed, they just wanted every bit of leverage to convince Qualcomm to negotiate. They're probably quite angry now, so the question is whether they have opportunities to be vindictive and if they take them.

Are you better value for money than AI?

There are some parts of some jobs that could be done by a chatbot. For example, support requests which often fall into two categories: the one where the user has to read the manual but didn't and the part where something weird is broken and we need to figure out what it is. A chatbot could, in theory, answer the former basic queries. Pretty much everyone would be happy if it could do that. The problem is that, when it isn't a basic query, it may try to answer it anyway with an answer likely to be unhelpful at best, and there's a chance that, while answering something basic, it will go off the rails and print something ridiculous nonetheless. The technology superficially looks like you could use it for some narrow use cases, but each unreliable example makes it harder and harder to do so.

Techie fluked a fix and found himself the abusive boss's best friend

Re: Cobol...

I prefer "did". I could speculate about places where it could have come from, and I often do, but that is rarely as helpful as drilling into it and figuring where I got it from, where that thing got it from, and so on until I find the first thing in the chain that gave me the wrong thing. The other risk is that I posit something as an "it could have come from here", the recipient hears it as an "the problem is here", it turns out it didn't come from there, and we have disagreements all round about whether I'm an idiot and did what I should have done.

OneOdio Focus A5: Big battery, budget sound, and a bargain bin price

Re: Getting names wrong

I'm not sure it will be as big a problem as it seems. While you're right that odio is hate in Spanish, audio is as much a word in Spanish as it is in English, and people will understand that. I'm reminded of the urban legend that a car named the Nova sold badly in Spanish-speaking countries because "no va" means "doesn't go". Most people understood the etymology of new, easy as the Spanish word for that is nuevo/nueva, and there was not a difference in sales. For a similar reason, the word "one" is the word for sand in a number of Polynesian languages including Maori and Hawaiian, but I don't think people are going to be confused by that either.

Biden’s antitrust crackdown on tech M&As may linger into Trump’s reign

Re: Need a full ban on M&A

"Whether "consumers didn't care or even notice" is irrelevant. People often don't "notice" when a harmful strategy is imposed on them."

Point taken, and you are correct. I was trying to describe mergers that are actually inert as far as consumer experience goes, which do happen. Those are quite common for very small businesses, such as local ones, where the person who started it and operated it wants to stop doing that, either because they're retiring, moving, or they decided operating something small and retail was not fun enough that they want to do something different with the rest of their time. Successful businesses are sometimes merged with a company that just wants to keep operating them for the profit, allowing the former owner to leave without having to shut it down and liquidate it. Customers often prefer that the business didn't shut down. You are right that some harmful mergers also go unnoticed.

"Cheaters gonna cheat. It is in the best interest of society to catch and punish them."

The problem here is that those cheats wouldn't be obvious, which makes them hard to catch, and they might not be legally cheating at all, which makes them hard to punish. When mergers are possible, but sometimes we don't allow them to complete, there's an explicit denial and an arrangement to circumvent that is more easily detected. If all mergers are prohibited, then you need new laws describing how non-merger relationships are restricted and new organizations vetting what might be secret negotiations for something that looks too merger-like. Those laws don't exist right now because, when companies normally want to merge, they do merge, and that makes them easier to detect and stop if we need to.

Re: Need a full ban on M&A

"There's not a single circumstance in history that consumers were not left worse off after one company bought another."

Except, as you say, when the company was going to fail. Or when the company developed some type of technology which was expensive and used for niche purposes, but by being bought by someone that could advance it, became cheaper and available for more people and reasons. Or if a company was low on money, not so low that they were going to fail, but low enough that they either had to, or more likely were going to choose to, make their product worse in order to extract more money from the user, something they didn't end up doing when they got infusions of more cash by a company that wanted the users to be happy and knew it wouldn't happen if the product became more extractive.

This is not all or even most mergers. It is far from the none that you have claimed. There are also less helpful examples of mergers that had an indifferent result where consumers didn't care or even notice that anything happened. Those happen frequently.

This is why we actually need to review these before they happen to try to catch problems. Just forbidding it has lots of negative consequences. If you tried, you'd see lots of faux-mergers where two companies ended up making sealed agreements to operate together, and you wouldn't be able to ban all of them because, not only is that a thing that happens all the time in non-merger businesses, but those would be the only way that lots of desirable activity could happen. Many of the harms you've seen from mergers would happen from that instead. When there is a problem as multifaceted as the harms that come from a lot of private equity, the solution tends not to be as simple as we would like it to be.

Supreme Court to hear TikTok's appeal against law that would force it to shut, or sell

Restrictions on free speech do exist, although the specifics in Canada, the US, and any other country someone might be reading from will differ and if we're going to refer to them, we'll have to use the US ones because this is a US case. However, it is probably useful to understand that they aren't written as such. The law doesn't say "you have free speech except in the following cases which can be changed by the legislature". Restrictions come when some other harm, such as that experienced by people who can't enter a hospital when they need to, overrides it. In many cases, this difference may look academic, because the result is the same, so the causes can't be that important.

There are two cases where that stops looking as unimportant. The first is in a court of law. If a court is going to decide that restrictions on the content or conduct of speech are necessary, they have to provide a reason that is compliant with other laws. In this case, that reason is "national security", which is now to be debated. The generic argument that you started with, that it isn't restricted if you could go somewhere else, is much broader than that and not supported by any law, which leads us into the second case where your version and the version currently in use look very different: in a world where there are active abuses.

There are always people who think that their free speech should have no limits whatsoever who think their inability to block a hospital door if they want to without being arrested is a violation of their rights. Those people are wrong. Most others understand why that is not allowed and wouldn't want to do that anyway. When free speech is eroded, the restrictions that are placed almost never say the right has been eliminated. They just put more and more restrictions on how you can exercise your right. For instance, when there were protests in the US about civil rights, the states that sought to block them never did that by passing a law saying that those protests were not allowed. They passed laws saying those protests were definitely allowed, because those people have the same speech rights as anyone else, but coincidentally you couldn't protest in the places where those people were likely to start doing it. They would then arrest people who tried. That was an illegal restriction, struck down several times by courts, that takes the form that you defended in your original argument. That is why your argument is incorrect from a legal perspective and undesirable if you could write it into law and why, even though it is, TikTok may well lose their case because it doesn't protect them. They will be protected or unprotected on their free speech claims by the court's interpretation of the national security value of a ban.

TikTok had some better, in my mind, legal arguments. This law appears to me to be a bill of attainder, a law that simply declares a certain entity is criminal without criminalizing any action they do. Those are not allowed. If I were a lawyer, I would have tried that. Their lawyers did and a court decided it didn't apply. As I am not a lawyer, let alone a constitutional lawyer, I don't know enough to know if the court had a good reason or not to decide that. I thought that was more likely to succeed than any free speech argument, and my opinion on that has not changed.

I agree with that part. That's why I would have guessed this to be not their best argument. Yes, their local company does have its own freedom of speech rights, but I don't think that's going to help them, and I wouldn't be surprised if their attempt to argue on behalf of their users gets tossed out as they can't speak for their users.

That isn't the argument you made. Using your argument, their users could also not get heard as they are free to speak somewhere else. Not only that, your version is severe enough that it makes it seem like TikTok actually should be able to make an argument on those grounds.

Adélie Linux 1.0 – small, fast, but not quite grown up

Re: Confused

Yes, they're not being entirely honest there. Not only do they have no images that you could deploy on a smartphone, but they seem to have no software intended to provide a smartphone interface. Their compatibility guide lists only typical desktops and laptops, although a lot of PowerPC Macs are supported. On that topic, I don't think you would need anything that unusual to install it, as it's also going to involve writing the image to bootable media and selecting it in the boot manager. They also seem to support only one SBC, the Pine64 A64, and otherwise you're on your own to make the ARM versions boot right.

The Automattic vs WP Engine WordPress wars are getting really annoying

Re: nice diatribe, but not much insight

"If you doubt mullenweg's stated reason, then make the case."

They have. I have. His stated reason makes little sense and, even if that's what he actually believes, he is not implementing it in a sane way.

"But making (large) commercial concerns suppport Foss, why not ?"

When it's a large commercial concern "supporting" by paying a larger commercial concern, it somehow loses a lot of the "it's humble FOSS and deserves their support" argument. When the concern being targeted actually does contribute, that doesn't help with the argument. When the concern that would receive the "support" is one that took over an open source project, ACF, to the detriment of the users of that project, with no notice at all, it also doesn't make the case that they do things worthy of the support. I am all in favor of companies supporting open source. I am less in favor of them being forced to do it in a way that makes it non-FOSS from the perspective of the small user or contributor. I'm very much against it when they're being beaten with dubious legal threats that mostly don't work unless they changed the license to be explicit about it*, while someone like Mullenweg makes spurious arguments for why this is helping which look like lies or delusions.

* I have a feeling they don't want to change the license because they've seen what happens when others do it. If they did change the license, it would unify people who intend to fork it, while if they retain the GPL, they can hope that a lot of people just stand back because they don't use the services of either host.

It's like a lot of cloud providers. The money is in hosting and management of the software, but they wouldn't have that without the software. A lot of open source companies are in such a position, because people start with self-hosting, they eventually get so familiar with the software that they want to keep using it, and they're happy to pay for support and maintenance so they don't have to do that themselves and they also don't have to change the software they're using. Word Press hosts have two more advantages. They can attach that to paid web hosting, which is also lucrative, and the product they're hosting is understandable to nontechnical people who want a website and don't like the look of the introductory HTML tutorial that seems to be their other option.

Unfortunately, some of those companies decide to try to take the software away from all those who continued to self-host it. I wonder if this is Automattic attempting to do the same while hiding that they're trying to do so. It has many of the same hallmarks, for instance the focus on one large provider first which is often intended to signal to all the developers who are providing free work that this isn't going to impact them. It does eventually come to hit those people, but only after the biggest money sources have either paid or left.

"If WP Engine fork it then they would need to pay for development of that fork"

Yes, and they would have maintenance burdens just to keep in sync. Also, they should expect to have Automattic try to deliberately break compatibility whenever possible. All good reasons why they would want to avoid that if possible. If others in the community sided with them, they could have more success, but it's always somewhat difficult and risky.

"and the whole issue Mullenweg has was WP Engine is over WP Engine not wanting to pay for development costs."

No, that's what he said. Not even what he consistently says. It isn't consistent with reality. WP Engine does contribute code to Word Press itself in addition to the related open source components that add to its value, but that evidently isn't enough for him. Others contribute nothing, but they haven't gotten his attention yet. If this was his problem, he'd have acted differently. This suggests that, whatever his problem really is, it's something different.

Apple and Meta trade barbs over interoperability requests

Re: This is not either/or

The reason you see that isn't because we have to choose. We're not going to get to choose anyway, the EU is going to pick a winner in each of these disputes. We're commenting because we are debating who we think the winner should be. I don't really like either of them. In previous rounds of Apple vs the EU, I have sided against Apple. I think the EU was right to require Apple to allow other app stores and browser engines. I was not convinced when Apple suggested nebulous security and privacy problems and the ones I could envision were small enough that I didn't care about them and Apple didn't talk about them because others wouldn't either. Now, I'm more likely to be siding with Apple. The privacy problems they're talking about are looking a lot more real this time and the benefits of the policy are much smaller.

Re: go Apple

"Maybe everyone who doesn't have taxes automatically withheld from their checks should all together refuse to pay taxes [...] That's tens of millions of taxpayers, many with high (but not billionaire high) incomes"

Leaving aside the other problems that I think you already know about, I think you're drastically overestimating the number of people who could even try to refuse to pay taxes. In order to not have preemptive withholding, in almost every country, you have to not be an employee. Contractors and business owners may have some chances to do that, briefly anyway, but you're not going to get to tens of millions that way. Tens of millions could refuse to pay when the government didn't withhold enough, but that's a much smaller hit to the budget, one that will be quickly filled, in fact overfilled, by people getting fined for tax evasion.

Re: cry wolf

IOS app distribution. IOS browser engines.

I'm sure you have your counterarguments lined right up, for example how smartphone apps as a whole aren't an Apple monopoly. It doesn't work, and has never worked, that way because the phone you have only runs one operating system and is limited to the facilities of that. The argument didn't work for Microsoft, either in the IE case or the Windows Media Player case, even though the restrictions in Windows were much weaker than those in IOS and you could install another operating system on the same computer, either instead of or alongside Windows. Unsurprisingly, it's not working for Apple either.

In this case, I'm much more on Apple's side. I disagree with the EU that interoperability is necessary in most of these cases*, and I am much more confident in Apple's privacy practices than Meta's. Apple does abuse privacy at times**, but Meta abuses privacy every second of every day. I think a GDPR-based argument is likely to go Apple's way if they have it adjudicated by someone knowledgeable.

* For example, I assume the request to have full access to iMessage is because of the theory that all messaging apps should be interoperable and thus that WhatsApp needs to be able to send to and receive from iMessage, and to avoid confusing everyone it probably needs to sync all the messages in one place. This may be correct from a technical perspective, but I don't trust Meta to have that access. Since I see no reason why WhatsApp and iMessage need to be interoperable when I can use both if I wish, I disagree with the EU and agree with Apple here.

** I am not likely to forget Apple's on-device scanning attempt. That significantly harmed their privacy record in my mind. I don't give them credit for not actually doing it because it took a lot of screaming to get them to back down, they tried hard not to tell people important information about it, and it feels too much like giving someone credit for not actually following through on a plan to punch me.