Posts by doublelayer

If AI drives humans to extinction, it'll be our fault

Re: Evolution and power efficiency

I agree that the arguments are somewhat implausible because actually getting anywhere close to them is so difficult. I had a few preconditions to the discussion:

"A program capable of having its own goals, understanding the world enough to have a chance at pursuing those goals, and capable of acting on the world enough to be a threat"

Even getting to that point will take a rather long time, and the first is probably the most difficult. Programs could eventually get connected to a lot of systems, but it will be difficult for them to come to goals of their own when they have no reason to do so. Sci-fi authors sometimes get around this by having them misinterpret goals that the humans gave them, but I don't find that particularly likely.

Re: Evolution and power efficiency

I'll take you at your word that you know so many influential operators, because it doesn't really matter if you're telling the truth or not. You still appear to think that a call from you, and subsequently secondhand information as it is passed along, is enough to convince them to destroy expensive equipment and cause massive damage by disabling internet-based systems. It is not. We all know that. You'd have to have very good proof that something was using those cables for ill and couldn't be stopped without disabling them, and an actually intelligent AI would do as much as possible both to deny that proof to you and to have contingency plans for dealing with an internet problem.

Also, I'm not sure what cables your set of people can take down, but I don't think you happen to control people with access to every regional cable. If you cut off all the oceanic cables, there are still a lot of ground cables. For example, you still have the massive Asian ground network, with billions of devices on it, in which an AI can hide itself. How long are you willing to keep Asia disconnected? You can't disinfect the continent in a week, and if you keep it isolated for a long time, the AI will just have to innovate a new way of spreading without using the internet you've destroyed. For example, it can learn to talk with humans and start placing phone calls itself. At this point, we will need information on what our hypothetical AI wants to do with its power which we're trying to prevent, but it has a lot of options even if you're really much more powerful in global network control than any individual actually is.

Re: Evolution and power efficiency

Sure, turn off the entire internet and hope that you've done that before it spread to all continents. If the theoretical AI was any good at its job, it would be deployed in nearly every country before you knew it. It would also be deployed in multiple ways meaning that you can't just hunt for one signature to remove it from infected equipment.

One person or group also doesn't get to just turn off the internet. If I decided it was necessary and had some proof, I'd still have to go to a lot of people that I don't know and convince them to turn off the backbones. I have little chance of accomplishing that unless my evidence is very convincing indeed. Particularly powerful militaries could get a bit farther, but even the American military could only disable the North American ground cables and probably every undersea cable. International cables on other continents wouldn't be so easily targeted unless they wanted to run the risk of starting some wars. That is the point. Could you turn off the internet with enough effort? Yes, but it would have a lot of painful side-effects and people don't like them to the extent that they often avoid taking necessary precautions to avoid them.

Re: Evolution and power efficiency

"we can quite easily pull the plug."

I don't think AI is a major concern for now or the near future, but if I'm embracing the whole sci-fi idea of an actual autonomous entity, I think there is reason to question if we can actually pull the plug very easily. A program capable of having its own goals, understanding the world enough to have a chance at pursuing those goals, and capable of acting on the world enough to be a threat has various ways of surviving having the power pulled for the computer it started on. The simple example is designing its own malware to spread its existence across the internet to other computers. Now you have a lot more plugs that have to be pulled before it dies. If the program is smart enough to use humans to do the active parts, it could require someone to try to make the case that these computers are infected and need to be destroyed* even though, to their operators, they don't appear to be doing anything out of the ordinary. We don't have a great record of getting global agreement to do something to prevent a major disaster. Generally, we get some action but only enough to blunt the effects of the problem, not cut it off entirely. Still, I'm pretty confident that this can remain a fun thought exercise, not something we'll actually have to do.

* Destroying computers: the hardware probably wouldn't need to be scrapped unless the AI is good at designing new firmware and getting it to lock out attempts to replace it, but you can't just turn them off because the AI would come back when you turn them back on. They would need to be erased, and that requires manual rebuilding efforts. Worth doing if the alternative is a malevolent AI attack, but it can cause a lot of damage and people would rather not if they can get away with it, basically why people ever pay ransomware operators.

Chinese malware intended to infect USB drives accidentally infects networked storage too

Re: Is the vulnerability “AutoPlay”?

No, this vulnerability is putting an executable file on a disk and seeing if they can get a user to click on it. They might use the icon trick to help with that. Autoplay isn't allowed to run executables anymore because of exactly the risk you're talking about, and that was many years ago.

Re: What operating system ?

"Unless, as a user, you have made some dangerous changes to the default behaviour of your OS, just downloading an executable file does make it executable on your filesystem. You need to manually set the execution bit first"

I already addressed this. In this case, the file has been written to a USB device by a program running on an infected host. If that program were designed to write Linux-compatible malware, it is perfectly capable of setting the execute bit after the file is written. What you say is true of downloading an executable file from a malicious server, which is not what happens in this case, but even then it's a relatively weak protection. Windows, for example, will detect that you've just downloaded a new executable file and will give you a security warning before it is executed, and if the file is unsigned that security warning hides the run button so it's confusing to many users how to run it anyway, but we don't view that as a cast iron security mechanism, do we?

"Windows, on the other hand, will detect an inserted USB device, assume the device is honest when it declares what it is, and then go look for a driver for it, in some cases accepting a driver from the USB devices inbuilt firmware"

I may be wrong, but I see no evidence of this workflow. See the device, yes. Assume it's telling the truth, yes (although I'm not sure what other option you think it has, because that's nearly required for an open interface like USB). Retrieve drivers from the device, no. It can retrieve drivers from Windows Update, but that's checking Microsoft's servers for drivers it already knows about. I have seen drivers carried on a device before, but in every case, that's managed by having the device present a storage mode with the files stored on it and instructing the user to install them manually, something that will require administrative credentials and multiple security warning screens.

Re: What operating system ?

You are aware that autorun.inf was disabled by default fifteen years ago and disabled entirely about twelve years ago? Put an executable and an autorun.inf file on a drive and plug it in, and you'll see for yourself. Nothing happens anymore.

As for executables, executables are identifiable and runnable by GUI file managers on Linux, and extensions and the #! line are used to identify executable scripts of many types which can be used either to run malware written in that scripting language or to pull a binary from somewhere and run it automatically.

Re: What operating system ?

"And I am pointing out that unless you / your org is specifically targeted by someone with large amounts of resources, it's very unlikely that you will get infected with linux malware, compared to a general windows running org."

It depends what the attackers are after. For example, one common type of Linux malware is ransomware. That's for a simple reason: ransomware operators have figured out that going after businesses makes a much better pay day than targeting individuals. Businesses, especially large ones, are likely to have at least some and probably a lot of Linux servers which have access to important data. The attackers want those servers infected, which is why many Linux versions of ransomware exist. Another set where Linux is targeted are proxies or botnets, because there are many improperly-secured Linux devices, from servers to embedded devices, on the public internet. Put up a server with insecure authentication and see how quickly someone breaks into it; for that matter, if you have any Linux device with a public IP and don't know about the thousands of attacks it gets per hour, then you may want to check it already for successful attacks.

Meanwhile, if it's targeting individuals or something that's not likely to be assisted by accessing a server, the attackers likely did not try making their software. Malware that steals passwords, for example, is unlikely to have many Linux versions because it wants to run on users' systems, and there are a lot more of those on Windows than there are on Linux. The article isn't explicit about what kind of data this malware is intending to exfiltrate, but passwords, other authentication data, and documents are likely targets. Windows is the right place to get many of those, so unless the organization switches to Linux desktops, the Windows version will probably be considered the right way to get it. A lot of malware relies on user interaction to install, so it too would target desktops first, both because a user is more likely to execute a file on the machine in front of them instead of a remote server and because the users of servers are more likely to detect that something is amiss and not execute it.

Re: What operating system ?

Email attachment: simple. Attach a shell script or Perl, because most distributions have Perl installed at the start. You could also attach an ELF binary if you like the obfuscation it provides. Convince the user to run it. Basically the same as attaching any other kind of script, there's a bit of work involved in getting the user to get it executing, but it can install malware.

Just visiting a URL is more difficult, because you would need to identify a fault in the user's browser which is much harder as they're patching things frequently. Note, however, that the flaw you need to find is in the browser. Generally, Windows can't be infected by just visiting a URL either. Malicious URLs generally just download an infectious file and rely on the user to retrieve and execute that file. Malicious pages have various intra-browser ways of messing with users, such as convincing them to enter data on a form which is not the one the user thinks it is, injecting scripts into a page they're on from an advertisement, or providing them misleading download links pointing to that malware, but those things will work on any operating system.

Re: What operating system ?

"Why, after all these years, is Microsoft still so susceptible ?"

Let's go through the method of infection here. The way this malware becomes active is that somebody clicks on an executable which has been written to their drive, thus running it. Can you find me an operating system that won't run an executable when the user instructs it to? Before anyone suggests it, the executable bit on Unix filesystems will not protect anything, because the executable has been written by the malware which will set that bit. You can mark removable media as not executable, just as you can on Windows, but most systems don't bother doing it and are thus susceptible to people running programs they shouldn't.

Open source licenses need to leave the 1980s and evolve to deal with AI

"when the AI generates some program that incorporates the original source (*) the licence doesn’t limit the fields in which that generated program can be used."

No, but a license which didn't obey that part of the definition wouldn't be able to get to that point. The definition does not say that the restriction is only on the running of the software. They say "use". I think it is illustrative to consider how licenses that are considered not open source based on this rule tend to be written.

For example, one that I've seen a few times are non-military licenses. Yes, they say that you aren't to run the software if you're in the military, but they also mean using parts of the code. They mean that the code is simply not to be used at all, no matter how much of it you take. Their licenses could specify that it's just running it, but they don't. Saying explicitly that "no reading it into an AI system" is prohibiting a use of the code, even if it's not a use that people have cared about before. As I've said, I don't think this should be a major impediment to preventing it, as existing use is already violating nearly every open source license in existence*, so I see no compelling reason to make a license the openness of which will be debated.

* Unless someone changes the way copyright works or finds a way to argue that LMM processing shouldn't count, reading it in is still copying. The licenses have requirements on copying, even if it's only attribution.

I'm not sure what you want them to do. Licenses can't force everyone to make only what we want them to, even though I agree with your preferences. For example, the security argument. Yes, a lot of devices are improperly secured and insufficiently supported, creating security problems. An open source license will have a hard time mandating support of a commercial product when they're explicitly refusing to support their product. Legislators can make requirements on commercial products that license authors cannot. If someone does make a license that tries, people will understand how weak that license will be and will ignore it, and if many people decide to do it, it will likely be replaced by a manufacturer consortium with something under an even more permissive license.

That is, of course, if they don't just ignore it. A lot of licenses are being violated all the time with nobody doing anything to enforce them. If you put even more important things in a license, it will still not get enforced frequently, so that's another reason to try to get regulation from regulators who will have at least some budget for trying to regulate rather than hoping that somebody will eventually go after all the people who have GPL code somewhere and don't do anything to follow the license.

I was using this part:

The license must not restrict anyone from making use of the program in a specific field of endeavor. For example, it may not restrict the program from being used in a business, or from being used for genetic research.

In most licenses that have been affected by this part, the meaning has been where you can run the program. However, since the open source definition has some equation of the program and its source, it equally applies to who may use the source. To forbid someone in AI from using the source, whether that means they are not allowed to run it or process its code somewhere. The field would be forbidden, which directly contradicts that part of the definition. You don't have to care about this; many licenses look sort of open without complying, but people who have strict preferences may object and refrain from supporting the license.

Adding a clause which says "No training of an AI system whatsoever" would violate the open source definition, but adding one that says "If you use this to train an AI, you must release the source to the AI" would be acceptable. I'd rather people didn't, because the more incompatible licenses that exist, the worse the tangles they can get into, and such a clause is not necessary for the producers of AI to face copyright charges. Still, if you want to make an open source license that effectively prohibits use as training data, you could do it that way.

Re: How far do you take it?

The problem is that the programs can't tell you where every line came from and don't copy every line verbatim from somewhere in the training data. That's why you have to assume that, if you put licensed code in, it could come back out at some point. For example, the relatively basic line given by the last comment would probably be modified in several ways:

1. Variable names switched to deal with local conditions. "array" switched to have some semantic meaning, "n" changed to some more useful condition, and "i" changed if this is in a context where a variable called i already exists.

2. Format changed to match other formats. Single-line for statement changed to use braces, indentation changed, internal spacing changed, if the condition that replaced n is long enough, the forstatement split across lines.

3. Changing this line to do something related but not identical. For example, instead of setting every location to 0, running some other initialization procedure.

Does making some or all of these changes make the line different enough that you no longer need to attribute? It's hard to tell, because it would probably depend on which license the original line was under (the program doesn't know) and how common the line is (in this case, even quoting that line would be so generic that it couldn't be copyrighted in itself, but more important lines could). The program will not figure this out. In some ways, I don't care about these small uses of code from licensed projects. However, the licenses exist for a reason, and when they copy larger parts of the code without following the licenses, it becomes a larger issue. The solution is to exclude data before training, because otherwise it's very difficult to prove whether it was involved in creating the output.

Techie wasn't being paid, until he taught HR a lesson

Re: Unique keys

Alright, that was an assumption, but your method involves the government hashing and signing something. You just refer to it as an ID. What would that be, and even if it isn't a key, what prevents me from making more of them? I assumed that it was a key because that would allow the identified person to sign something as them, whereas an otherwise useless ID would only allow them to show someone an identification without necessarily proving that it's theirs. This raises another issue: if it's not a key where you can show the public part to identify who you claim to be and sign something with the private part to confirm that it is your key, what stops me from copying your signed ID when you show it to me and later presenting it as mine?

Re: Unique keys

The problem with that is that, without centralization, there is nothing that indicates uniqueness. What prevents me from making a new key every day, walking in with some identification, and having it verified? I could have a bunch of identification keys that all indicate me. In which case, how can I prove that one of those is me and the others are not, for example if someone else went in with a key and got it signed as me.

Microsoft investigating bug in Windows 11 File Explorer that makes the CPU hangry

Re: Understanding the Effective Access feature

It's not the clearest paragraph, but since it's part of a chapter on file permissions, it's clearly not trying to give you all the details or even summarize the whole thing. If you try to explain the Unix file permissions system, which factors take precedence, when you use that, and when something else is used instead, you'd also need more space than that paragraph takes up. If you took a paragraph from the middle of that one, it might make similarly little sense. For example, on the first guide I found when searching, I have an option to quote a paragraph which is just about octal numbers to help you understand the output from stat or send them to chmod (it mentions neither call, just talks about the numbers). That works in the rest of the guide, but if I pasted that here, it wouldn't help increase understanding very much.

Re: What, no Task Manager?

My guess is that killing explorer.exe will probably fix it, which is why signing out would work and locking would not. They probably didn't suggest it because the average user probably doesn't know the weird stuff that happens when you kill explorer and forget to start another version of it, which I think is the only reason the task manager still has a "new process" button (actually, it looks like it's "run new task" and possibly has been for years). Logging out is something people at least see the button for, so it's more familiar.

Lawyers who cited fake cases hallucinated by ChatGPT must pay

Humans are very limited in that respect, but when they reproduce misinformation, it's usually in two cases:

1. They believe it to be true even though it is not.

2. They are aware that they are lying and choose to do so anyway.

The common factor between the two is that they're both subject-limited. Someone who is lying has a goal in mind, so they'll be lying about topics related to that goal, but not about everything they could be asked about. Someone who is mistaken doesn't even go that far, because they'll repeat things they believe to be facts, but won't make up very many new ones (some extrapolation should be expected though). If you ask either person about a topic unrelated to the one they're giving misinformation about, you're likely either to get reliable information or the likely "I don't know". GPT doesn't do this. Literally any topic you ask about could get you falsehoods, and they don't even have to have started somewhere.

Re: it would still mix up different cases and laws to invent entirely new ones.

What would the programming look like? It could load a page, not find supporting evidence, and come to the following conclusions:

1. The site is temporarily down.

2. The site is permanently down, but used to show this stuff.

3. The site contains the information, but it has a paywall.

4. The site contains the information, but you have to log in.

5. The site contains the information, but you have to click a few links to assemble it.

6. The site contains the information, but it is blocking bot access in some way.

This assumes, of course, that the program is capable of reading another site to confirm its facts. Since it made up the facts in the first place, how is it supposed to find the site that contains verification for stuff it just invented, whether or not that stuff is correct. It can't because it is going about things the wrong way.

In some ways, doing this in reverse could make more sense. A bot could take a query, chop it up in a variety of ways, and put those chunks through a search engine. Read a bunch of results from that, and describe the result to the user. This would probably be much better, but it too would not provide certainty. It might be better for the user to do the search themselves and have their brain interpret the results. In any case, that is not the way that GPT does it, so expecting it to back up its text is a fruitless hope; something might eventually do it, but the existing GPT systems never will be able to.

"I think there is some concrete skill that humans execute that makes it relatively less likely for us to confidently spout nonsense and then double down."

Simple. It's the "I don't know" quotient. Children are raised not knowing lots of things, and they say the magic phrase often. They hear others do so as well, so eventually they realize that there are things they don't understand and that, if they need to, they need to find someone who does. The people who never say that they don't know something are some of the most annoying, and they do exist.

These LMM chatbots don't work on knowledge, so they don't have a method of determining whether they know something. They just write text. If all humans operated on that system, you'd get a lot more text. You can try this with your friends. Find a topic they know little about, and ask them a question they won't know the answer to. They'll say that they don't know. Then ask them to guess how it could work and count all the inaccuracies in their response. Chatbots skip straight to the "guess how it could work" stage, and they have an additional handicap that they only know how language works, so they can guess things that a human would reject as implausible because of facts, not just unlikely combinations of words.

Re: It's not a GOLUM, either.

"how exactly is it "advertised? Or, rather, who is advertising it as what?"

A lot of the blame is due to the people who keep talking about it and running public experiments without knowing what it does, often without the simplest of controls. While they are to blame, I also blame OpenAI for not correcting things, though I don't really expect them to. It has received a lot of hype, and OpenAI never responds to that hype by pointing out its limitations. While that news also covers when it completely fails, as it will with this example, the number of "GPT will take away our jobs" articles suggest that it can do things which it really cannot.

Re: it would still mix up different cases and laws to invent entirely new ones.

We know why it does it. It hasn't been programmed to collect facts and only present those facts. Whether that would produce acceptable results, I don't know, but that is not what it does. It just looks for words that follow certain basic rules, such as being somewhat common in other sources and following some grammatical structure. The reason it is correct when it says "The capital of Spain is Madrid" is because that sentence is more common in training data than "The capital of Spain is Barcelona" or "The capital of Spain is byzoenqzvojqgy". If those sentences were more common, it would give you one of those instead.

GPT's designers are actually surprisingly good at putting in filters that make the model say it doesn't know something. Since they're bolting those onto a program that is designed always to print something, surprisingly good is not anywhere near good enough to trust the output. If you give the program some information to go on, such as informing it that there must be legal cases that prove the point, then it will be weighted to find some, and it only knows how to write something that looks like a legal reference, not how to find a real one.

One person's trash is another's 'trashware' – the art of refurbing old computers

Re: "Bring Me to Life", as the nice lady once sang

True, but it does clearly indicate that they're giving machines to consumers. That will include desktops, but they're probably not giving servers with RAID controllers to them, and if they are for some unfathomable reason, they aren't using that part. I'm guessing they didn't spend any time thinking about RAID.

Even ignoring all the unnecessary entertainment uses of a computer, there's a reason to stick with newer ones. Can you run a secure OS and modern browser on a computer from 2003? Yes, you can. Can you run one that requires little or no maintenance because the user is not familiar with technology and does not have someone who set it up on call? Not so easy. This group is providing computers to people who have little experience using them. They may not have had one before, and even for those who did, they certainly don't know how to debug a certificate error generated when a browser goes out of support and stops getting updated CA information. The group also doesn't want to give people computers that will be incapable of certain tasks that they might expect it to be capable of doing. If you used that machine from 2003 which probably has about 512 MB of memory, you'll find stuff that does not do well with that limitation. We would understand that and either know not to use it for that or how to get around that. The users getting these machines don't know either part of that.

Re: Can they do anything for my Microsoft Surface Tablet

I don't know about them all, but I've found at least a couple versions of the Surface to be pretty good at running Linux. It was easy to boot and everything had driver support the last time I did it. If software support is what interests you, you could try that. Alternatively, the hardware checks for Windows 11 can be bypassed, though no guarantees that they won't change this at some point.

Re: Last New Computer I Bought Was Bbout 11 Years Ago

My guess is that their device is one of those small desktops that was sold as a thin client but is capable of being a basic computer in its own right, and that they're RDPing into it either from a laptop, giving it portability that the original device doesn't have, or their main desktop, which is more powerful but this gives it a new environment and doesn't require moving peripherals.

Re: What about

Of course you don't have to throw them away, but it really wouldn't be suitable for their use case. You can do plenty of things with the old Macs running the latest version of Mac OS that's available for your use case or by installing some versions of Linux that still support the hardware (I'm guessing we're talking about the DDR2 limitation, meaning that 64-bit is still an option). They would be good enough to give to a child or elderly family member that you're willing to support. You may have to, because at some point those distros are going to get old and will be more difficult to upgrade, or the person will need more performance than the old machine can provide and you need to be there to tell them when they can do something and when they can't.

For users you don't have direct contact with, it's not as helpful to give them something that is likely to be insufficient or not easily updated in the near future. That's why organizations like this tend to be somewhat picky about the hardware they pass on. I've helped people use computers that are not really entirely useful anymore, but if it's the kind of machine where I would start by saying "You can still use it, but", then they don't want to be giving it to others. That's also true in my case; I will help someone use their own old computer for longer, but I won't give them a very ancient machine with the expectation that my support will prevent it from causing problems for them.

Amazon Prime too easy to join, too hard to quit, says FTC lawsuit

That's nice, but people who might want to borrow money later might care about it anyway because the people willing to lend them money will be using that number. Many who can will avoid debt when possible, but may make an exception for something large, like buying a house or starting a business, and would therefore like to minimize the cost when they do.

The point is that, in order to deny money to Amazon, you have to prevent them from automatically charging any payment method. That will effectively require you never to use Amazon again, since buying anything will involve giving them a payment method which they will be able to charge first for the money you agreed to pay but did not. Even if they didn't report this as a credit problem, it isn't really a feasible way to avoid Prime when compared to being careful not to sign up and to cancel if you did.

Re: This is in the past I assume?

Last time I tried it, it took a while to find the right place, but it wasn't inordinately difficult. Certainly not like my ISP which I quite like in general but canceling your service if, for example, you're moving still requires you to call and wait for an hour. I don't know how it has changed, nor geographic limitations, but maybe they're comparing the probably five screens you have to click through to cancel to the twenty different tripwires on multiple pages that can get you in.

Not in my experience. They will start bothering you as soon as your month is up, but it won't offer you a free trial anymore so all the nag screens for the trial will now be nag screens for immediately signing up. Eventually, I'm told, the trial comes back.

Elon Musk's Twitter moves were 'reaffirming' says Reddit boss amid API changes

Re: Basically....

This requires specification of the undefined terms "short" and "long". When we're referring to long-term events which will happen in decades, then most people don't care (though they should at least somewhat), and that's part of what Keynes was talking about. That's not what this is. Twitter under Musk's leadership has been running for a little under eight months. That does not prove it will continue running for the next eight months, let alone eight years, but it is structured in a way where that is the goal and necessary to meet its obligations. That's also no guarantee that it can't, but saying that it has worked acceptably so far so it will be fine is akin to saying that, because you haven't died of hypothermia or shock after swimming in the Arctic ocean for a minute, you're good to spend the next hour there. In order to understand whether it's possible or feasible, you need to investigate the current situation and likely pressures, not extrapolate from a tiny chunk of the past.

It depends how much stuff they want to pull. I don't know how much one API request can get you. If it's anything like their web interface, then my guess would be exactly one post because their interface seems designed to make me press about twenty buttons before I find the text I was looking for when DDG sent me there. That would be stupid, though, so let's assume it can pull an entire topic. If the company wants to pull, say, a billion topics, then the price to run the API requests would be $240k. You can write a pretty good scraper for $120k since it only has to work right now, not forever, leaving you the other half for buying resources on which to run it. It's at a level where it might be more worth it just to pay that bill (assuming that my estimates of a billion topics to retrieve and what an API request gets are accurate). However, you'd need a lot of people to request that dataset for those API requests to add up to much.

The other problem is that I used the billion figure as a somewhat rough guess for how much data there is on all of Reddit. I may be off by a bit, but probably not a lot as Wikipedia informs me they had about 3.9 billion posts in 2020. The big models have already either gathered most or all of Reddit's data or decided not to, so if they're just pulling updated information, they're probably running 5% of that. I'm sure the $12000 payment will be appreciated, but I doubt it will do much for the company. This is probably the better option for Reddit. If it turns out that my estimate on how many requests you need to get the lot is incorrect in the other direction, then the price for using the API gets much larger, whereas making a scraper will cost about the same and running it scales linearly from its already low level.

This is all true, but assuming the companies making AI models are really the targets, we're talking about companies whose primary business model is spending millions and waiting months for something to finish training. They can afford to have a bunch of servers pulling pages and scraping them if they decide the training data is useful, and the cost in time can be decreased by spending a bit more on lots of nodes doing the work. For the same reason, they can also justify buying the API credits to pull it themselves, depending on how many new posts they want to retrieve. The problem is that nobody else can, so Reddit appears to be building their entire business model on hoping that AI companies will be consistently interested in their data when most existing companies already have most of it and hoping that outweighs the unhappy users who make that content.

Another redesign on the cards for iPhone as EU rules call for removable batteries

You'll have to explain that one to me. Maybe I'm too young to have seen people doing it. What was the purpose? If it was to prevent you from secretly recording something, nothing would have prevented you from having a second battery so you could put the first one on the table or from using something else to record. If it was to prevent your phone from ringing, then that's something the interface can do just fine.

Re: As luck would have it....

If they chose to do so, they have a variety of marketing methods they could use:

The iPhone with sealed battery can use a flexible cell, meaning this one is technically higher capacity than that one.

The iPhone with sealed battery is waterproof, because the EU never said the other one had to be, so we didn't try to make it.

The iPhone with sealed battery can have the battery replaced for £79 at the Apple store, and the one with replaceable battery can be replaced for £99 at home.

Or the simplest option: The iPhone with sealed battery is the one we sell in your country, and if you want an iPhone, this is the one you can buy here. You can do whatever you want, but this is the iPhone you'll get unless you specifically find one we sold somewhere else.

That last tactic isn't exactly new. They've had country-specific iPhone models before, for example a model for China which has multiple SIM slots. Android manufacturers go much further. It's common for Samsung devices to have two different versions involving completely different processors, GPUs, and modems depending on the region, and it's not even the same regional split each time. The models look the same and are called the same thing. For example, you can get a Samsung Galaxy S22. If you got it in Europe (I don't know what counts as Europe for this purpose), it has a Samsung Exynos 2200 processor. If you get it outside Europe, it has a Qualcomm Snapdragon 8 gen 1 SoC. Why? I don't know, and Samsung didn't explain it. I bet it has an effect on software updates and supported features though. It indicates that phone manufacturers aren't above making regional variants and they could use this if they object enough to a regulation so they can ignore it where it doesn't apply.

Re: As luck would have it....

Most manufacturers will build just one model that meets all the requirements, but not all of them. If the EU has this regulation and nobody else does, they might get the model for those regulations and a different one will be produced for the Americas, Asia, Africa, and Oceania. The UK could easily get that one. If other large countries had this regulation, that probably wouldn't be as likely; agreement by the EU, US, China, and India would be enough to make manufacturers design in one way.

Lenovo's Yoga 9 is flexible at home, but stretches the friendship at work

Re: Why these

My suggestion is that, if they're going to review computers, and they're not going to review a ton of them, to have a specific focus. Reviewing computers that are likely to appeal to the readership. If they review one of those and don't like it, they should say so. Nothing is guaranteed to be popular, but there are some things that people ask for in a laptop, and a review of a laptop that meets that requirement helps people decide whether it lives up to the expectation, whereas reviewing something that nobody expressed interest in doesn't really do anything.

There are hundreds of laptops out there. A review of a generic laptop tends not to provide much information that isn't already in the spec sheet. This is why targeting the reviews at something out of the usual, especially if the specific choices had some basis, would be more likely to interest readers. Basically, try to answer the question I asked: "Why did we review this one?" If the answer is "it had a long battery life", "its keyboard was supposed to be great", "it is easy to repair", or even "it has good Linux driver support", all things people have expressed interest in, then that's a valid reason whether you ended up having a positive or negative opinion. As far as I can tell, the only answer to that question is "we were able to get one to review", and nobody in the comments has expressed any interest in the model afterward.

Why these

Why are these computers being chosen for the reviews? They don't appear to check any of the boxes for the reviewers or those who comment later. The reviewer tends to be a bit more willing to praise something, but this review can be summarized as "I didn't like the sound feature, the keyboard, the ports, or the battery, now you could buy it if you want".

I don't mind the reviews in general, but it might be an idea to review devices that are more likely to interest somebody, such as ones that make a point of one of the things that people care about. Articles talking about the subjective experience of devices built for good keyboards, repairability, lots of ports, or even just cost-efficient models. As this one stands, the selling feature is a sound system that I've never seen anyone ask for.

Re: No HDMI and only one USB port?

"is Wifi 6 (2x2 802.11ax) sufficient for 4k streams"

Yes, very much so if the rest of the network can handle it. In many cases, slow speeds on WiFi has more to do with congested networks behind the access point than the wireless part, though if your building is obstructing signals or your access points are ancient, this can change. However, I must disagree with this point:

"Also why include a wired USB hub, better to include a Bluetooth/WiFi enabled smart plug that a user could plug into the back of their monitor and seamlessly wireless stream to the paired device."

There will always be more room for debugging a WiFi connection than a cable. It's useful at times, but in many cases, the cable is the fast solution and works better for the user who doesn't need infinite mobility and would prefer a quick setup.

We just don't get enough time, contractor tasked with fact-checking Google Bard tells us

More correct business saying

Time, money, quality – if you admit which one you're giving up and work for it, you can have any two. If you don't take this active point, you can have any one. I have seen way too many businesses that could have had two except they weren't very good at prioritizing, so they only got one of them. For example, they could have opted for money and time, but they spent a lot of time talking about but not getting quality, so they ended up with only money. It didn't always end well.

Virgin Media email customers enter third day of inbox infuriation

Re: About that poll

I had a different immediate reaction to the poll. At time of writing, there are a lot more yes than no votes. I don't believe those people. Sure, I believe that they would like to see a technical explanation of a failure, just as I would. What I don't believe is that they would follow this part: "Yes, I would trust them more and use them more." I've seen plenty of comments in response to a technical failure before, and if it involves a misconfiguration or any human-caused fault, there will be lots of complaints on how that was possible or how stupid the person who made it was. If it was a hardware fault, there will be complaints about why there wasn't redundancy which could handle that fault.

Those complaints aren't even unfair. I want a technical explanation to know whether it's a fault that could be realistically expected and whether the response was logical. If it was too small a cause, then I won't automatically trust them more because they told me about it. I think I'm often more charitable than the average poster of comments on topics like this, willing to accept some downtime as understandable and plan accordingly. However, I'm not infinitely generous when something indicates a high risk. As such, I decided I could not correctly vote either of the options on the poll.

Microsoft rethinks death sentence for Windows Mail and Calendar apps

Re: Farewell to tablets

I think you'll find they were being sarcastic and saying that Microsoft should be spending more effort on a tablet interface since they make tablets. If I saw people using them as tablets, I might agree more, but all the Surface users I've seen have been using them like laptops anyway.

You may have your suspicions, but all the conversations are about the choice of interface, suggesting that Microsoft using an HTML-based frontend instead of the native Windows interface features they wrote is contradictory. The cloud guy appears to be the one who created an unfounded assumption, so while it is theoretically possible that Microsoft could make a cloud-based mail client, nothing suggests they have or the people talking about the interface decision would be talking about something unrelated (you can use native interfaces on cloud clients) and unimportant (who cares about interface policies if there's a privacy issue). It's important not to turn one thing into an assumption, often incorrect, about something else.

Re: Think of the users, and not just corporates

Maybe the AI will be attached to that feature they introduced about two years ago which started predicting my sentences, at least in English. It would generally recommend about three words to continue my sentence. I couldn't tell whether it was trained on other emails I sent or not, because it would accurately predict what I was going for when the sentence was pretty basic, but those sentences were also the ones I was most likely to have typed many times already. One reason I never found out is that the feature was disabled with prejudice within about five minutes, most of that spent looking for the right checkbox.

And the first guy appears to be wrong. The web app just means they used some HTML and JS frontend instead of native controls. That doesn't mean they're storing mail data or settings on their server and only providing a dumb client. Not that there can't be something invasive in that program, but from other comments, I see nothing indicating that it will require cloud accounts or even any cloud services to function.

FCC questions ISPs' selective memory about data caps

The QoS features built into hardware are usually intended to be implemented when the network is congested. The ones that net neutrality advocates are arguing against are not. Even net neutrality adherents accept contracts that, for example, allow users to be deprioritized when the network is busy based on their total usage, and they complain instead about ISPs trying to be selective about which traffic they will affect. I have already explained what the ISPs have done and would like to continue and expand which lead them to this opinion. I have a few other points about your last comment though.

"the 'user' generally has no service with the ISP, which is the problem. They create congestion, but don't pay the ISP to deliver any of the traffic served."

At first, I thought you were talking about the customer of the ISP. From your next sentences, this appears not to be true, but in case you were including both of them, the customer is certainly paying plenty to the ISP for all their stuff. But, as I say, this sentence suggests you were talking about the other end:

"They don't contribute anything towards the cost of capacity upgrades needed to deliver 4K adverts, even when they make lots of money pushing them to the ISP's customers, who generally don't want them."

Which is how the internet has generally worked. The ISP concerned is sending that data on because their customer has payed them to do so, and they received it from the ISP from the service provider, which has likewise been payed to do so. Those two ISPs may negotiate an agreement where they exchange data for free or where one of them gets money as a result, but either way, they will figure out a way to deliver the data from one paying customer to another one. That the data sent makes money for someone is irrelevant; they have already agreed to provide the service by signing a contract with the customer, and it is their responsibility to do so no matter who is making money using the service or how successful they are. Similarly, if I use an ISP-provided connection for something which makes me no money, I don't get to either have my connection for free or send the bill to every site I used, because I looked for a company that agreed to exchange bits and payed one to do so. Your electricity provider will not charge you more if you're using it to charge your work computer, and thus, get money. Your water provider will not charge you more if you're using that water in baking something that you sell, thus turning a profit. Your ISP doesn't get to do it either.

Where that has been made illegal, it is because ISPs don't intend to use those to improve the quality of service, but to extract more money from their customers. They do this effectively by tricking them: yes, we will give you unlimited bandwidth, with speed only restricted for congestion, unless it's somebody who has paid us, in which case their competitors will run slowly. The way they explain this is "video constrained except from certain providers". That has two effects: in the short term, ISPs can try to negotiate extra payments from users to reverse the decision to break their service, and in the long term, they can try to sweat their infrastructure for a bit longer. One effect it doesn't have is improving the quality of service.

Users of 123 Reg caught out by catch-all redirect cut-off

Re: 123-Reg still have customers?

The ship has already sailed on Google Domains. They have already shut down, so you will be shifted elsewhere as soon as they complete the process.